阅读说明：本技术 用于确定开关故障类型的方法、装置、处理器及配电设备 (Method, device, processor and power distribution equipment for determining switch fault type ) 是由 尹志斌 霍超 程显明 甄岩 陈文彬 郑利斌 陈静 孙海鹏 于 2021-07-14 设计创作，主要内容包括：本申请实施例提供一种用于确定开关故障类型的方法、装置、处理器及配电设备。方法包括：获取开关的实时运行数据；对所述实时运行数据进行预处理；将预处理后的实时运行数据输入至故障识别模型；通过所述故障识别模型确定所述开关的故障类型。该方法基于深度学习技术构建深度神经网络模型,通过神经网络模型输出智能开关的故障类型并对智能开关执行进行相应的保护动作,实现对保护线路及用电设备以精准、可靠的保护,还能够及时发现电气线路的超温、过载、短路和接地等安全隐患,并快速动作,实现保护智能化。(The embodiment of the application provides a method, a device, a processor and power distribution equipment for determining a switch fault type. The method comprises the following steps: acquiring real-time operation data of a switch; preprocessing the real-time operation data; inputting the preprocessed real-time operation data into a fault recognition model; and determining the fault type of the switch through the fault identification model. The method is characterized in that a deep neural network model is built based on a deep learning technology, the fault type of an intelligent switch is output through the neural network model, corresponding protection actions are carried out on the intelligent switch, the protection circuit and the electric equipment are accurately and reliably protected, potential safety hazards such as over-temperature, overload, short circuit and grounding of the electric circuit can be timely found, the actions are fast, and protection intellectualization is realized.)

1. A method for determining a type of switch failure, the method comprising:

acquiring real-time operation data of a switch;

preprocessing the real-time operation data;

inputting the preprocessed real-time operation data into a fault recognition model;

and determining the fault type of the switch through the fault identification model.

2. The method for determining a type of switch fault of claim 1, wherein said determining a fault type of said switch by said fault identification model comprises:

acquiring fault types output by the fault identification model and prediction probabilities corresponding to the fault types;

and determining the fault type corresponding to the maximum prediction probability as the fault type of the switch.

3. The method for determining the type of switch fault of claim 2, further comprising:

comparing the prediction probability of the fault type with a preset probability threshold;

and executing a protection action corresponding to the fault type under the condition that the prediction probability of the fault type is greater than the preset probability threshold.

4. The method for determining a type of switch fault as claimed in claim 3, wherein said performing a protection action corresponding to said type of fault comprises:

searching a database for a protection action corresponding to the fault type;

and under the condition that the protection action corresponding to the fault type is found, executing the corresponding protection action on the switch.

5. The method for determining a type of switch fault as recited in claim 4, wherein said performing a protection action corresponding to the type of fault further comprises:

and uploading the real-time operation data to a remote control center under the condition that the protection action corresponding to the fault type is not found in a database, so as to monitor and/or control the switch through the remote center.

6. The method for determining a type of switch fault of claim 1, wherein the pre-processing the real-time operational data comprises:

and normalizing the operation parameters contained in the real-time operation data to scale the operation parameters to a preset range.

7. The method for determining a type of switch fault of claim 6, wherein the pre-processing the real-time operational data comprises:

scaling the real-time operation data to a preset range according to a formula (1):

wherein x is real-time operation data, x' is preprocessed data, A_minAnd A_maxRespectively, the minimum and maximum values in x.

8. The method for determining the type of switch fault of claim 1, further comprising:

acquiring historical operating data of a plurality of switches;

preprocessing the historical operating data;

inputting the preprocessed historical operating data into a fault recognition model so as to train the fault recognition model.

9. The method for determining the type of switch fault of claim 8, wherein inputting the preprocessed historical operating data into a fault recognition model to train the fault recognition model comprises:

performing weighted calculation on the historical operating data through a weight vector matrix of the fault identification model to obtain updated historical operating data;

adjusting the weight of the fault identification model through the updated historical operation data;

determining an error value of the fault identification model;

and determining that the fault recognition model is trained completely under the condition that the error value is lower than a preset error threshold value.

10. The method for determining a type of switch fault of claim 9, wherein the fault identification model includes an input layer, a hidden layer, and an output layer;

the adjusting the weight of the fault identification model through the updated historical operation data comprises: adjusting the weight of each network layer of the fault identification model according to the formula (2) and the formula (3):

Δw_jk＝ηy_j(d_j-y_j)(1-y_j)y_kformula (2);

wherein, w_jkIs the connection weight between hidden layer neuron k and output layer neuron j, v_kiIs the connection weight between the input layer neuron i and the hidden layer neuron k, eta is the preset learning rate, y_jRepresenting the output result of the network layer, d_jTo desired output, y_kAs output result of the hidden layer, x_iIs input into the input layer.

11. The method for determining switch fault type of claim 10, wherein inputting the preprocessed historical operating data into a fault recognition model to train the fault recognition model further comprises:

the output of each network layer is determined by equation (4) and equation (5):

wherein z is_kRepresenting the intermediate calculation result, the value range of k is [1, q ]]Q is the number of neurons in the hidden layer, y_jThe output result of the network layer is shown, and the value range of j is [1, m]M is the number of neurons in the output layer, x_iRepresenting input data of the network layer, iThe value range is [1, n]N is the number of neurons in the input layer, f₁(x) As an excitation function of said hidden layer, f₂(x) Is the excitation function of the output layer.

12. The method for determining a type of switch fault of claim 11, characterized in that the excitation function is defined as formula (6):

wherein, theta₁To vary the parameters of the function amplitude values, theta₂To change the parameters of the variable elasticity, x is the input data of the network layer.

13. The method for determining the type of switching fault according to claim 9, characterized in that the expression of the weight vector matrix is defined as formula (7):

wherein, T_nExpressed as the weight vector matrix, r is expressed as the weight of the eigenvalue.

14. The method for determining a type of switch fault of claim 9, wherein said determining an error value of said fault identification model comprises: determining an error value of the fault identification model by equation (8):

where E is the error value, n is the number of training samples, and E (i) is the training error of the ith training sample.

15. The method for determining the type of switch fault of claim 14, wherein e (i) is obtained by equation (9):

wherein d (i) represents the expected output of input x (i), x (i) being the input data to the fault identification model; k is hidden layer neuron, and the value range of k is [1, q ]]N is the number of neurons in the input layer, y_kIs the output result of the hidden layer.

16. The method for determining the type of switch fault of claim 1, further comprising:

adding the real-time operation data of the fault type into a historical data set so as to update the historical data set;

training the fault type through the updated historical data set.

17. The method for determining the type of switch fault of claim 1, wherein the fault identification model comprises a back-propagation neural network.

18. The method for determining the type of switch fault of any of claims 1 to 17, wherein said real-time operational data includes at least one of current, voltage, current harmonics, voltage harmonics and junction temperature captured by said switch.

19. The method for determining the type of switch fault according to any one of claims 1 to 17, characterized in that the fault type comprises at least one of overload, short circuit, undervoltage and open circuit.

20. A processor configured to perform the method for determining a type of switch failure according to any one of claims 1 to 19.

21. An apparatus for determining a type of switch failure, comprising the processor of claim 20.

22. An electrical distribution apparatus comprising an apparatus for determining a type of switch fault according to claim 21.

Technical Field

The application relates to the technical field of computers, in particular to a method, a device, a processor and power distribution equipment for determining a switch fault type.

Background

The intelligent switch, one of the most important devices in the power system, has double tasks of control and protection, and the reliability of the performance of the intelligent switch is related to the safe operation of the power system. The intelligent switch can be used for distributing electric energy, protecting a power supply circuit, a motor and the like, and automatically cutting off a circuit when a power system has serious faults such as overload, short circuit, undervoltage and the like, and has the function equivalent to the combination of a fuse type switch, an over-and-under-heat relay and the like. Intelligent switches are widely used in power distribution systems, power transmission systems, and electrical equipment to protect electrical equipment and cables, etc.

The traditional switch equipment utilizes certain physical effects to realize the closing and the opening of a circuit through mechanical action, so that the traditional switch equipment has the advantages of large volume, large power consumption, high temperature rise, low accuracy, low speed, strong protection pertinence, incapability of changing along with the change of protected equipment and inconvenience for specific application. The biggest defect is that the intelligent detection is not realized, and the detection of the switch performance is mainly realized by performing preventive detection under the condition of power failure to detect the mechanical and electrical performance of the switch. However, the method cannot find the abnormal condition of the accident in time, and the normal operation of the switch can be influenced on the contrary by excessively disassembling and overhauling the switch. The traditional power distribution protection relies on terminal equipment of each connection point to realize on-site protection of a circuit, and each terminal equipment is required to upload collected power information to a power distribution master station, and the master station makes a decision and issues a control command. With the rise of the comprehensive automation and the transfer automation of modern power systems, the power grid protection is not an isolated, single-task and passive standby device, but a computer automatic control system which participates actively and maintains the overall safe and stable operation of the power system together. The power grid protection not only needs to realize the removal or automatic superposition of protected equipment, but also can be used as a terminal of an automatic control system, receives a scheduling command to realize the operations of tripping, closing and the like, and fault identification, state prediction, safety monitoring, load control and the like.

Disclosure of Invention

An object of the embodiment of the application is to provide a method, a device, a processor and power distribution equipment for determining the type of switch fault.

To achieve the above object, a first aspect of the present application provides a method for determining a type of a switch fault, comprising:

acquiring real-time operation data of a switch;

preprocessing the real-time operation data;

inputting the preprocessed real-time operation data into a fault recognition model;

and determining the fault type of the switch through the fault identification model.

A second aspect of the present application provides a processor configured to perform the above-described method for determining a type of switch failure.

A third aspect of the present application provides an apparatus for determining a type of switch failure, comprising the processor described above.

A fourth aspect of the present application provides a power distribution apparatus comprising the above-described apparatus for determining a type of switch fault.

A fifth aspect of the present application provides a machine-readable storage medium having stored thereon instructions which, when executed by a processor, cause the processor to be configured to perform the above-described method for determining a type of switch fault.

According to the technical scheme, the method for determining the switch fault type comprises the steps of acquiring real-time operation data of the switch, preprocessing the real-time operation data, inputting the preprocessed real-time operation data into a fault recognition model, and determining the fault type of the switch through the fault recognition model. The method is characterized in that a deep neural network model is built based on a deep learning technology, the fault type of an intelligent switch is output through the neural network model, corresponding protection actions are carried out on the intelligent switch, the protection circuit and the electric equipment are accurately and reliably protected, potential safety hazards such as over-temperature, overload, short circuit and grounding of the electric circuit can be timely found, the actions are fast, and protection intellectualization is realized.

Additional features and advantages of embodiments of the present application will be described in detail in the detailed description which follows.

Drawings

The accompanying drawings, which are included to provide a further understanding of the embodiments of the disclosure and are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and together with the description serve to explain the embodiments of the disclosure, but are not intended to limit the embodiments of the disclosure. In the drawings:

fig. 1 schematically shows an application environment diagram of a method for determining a type of switch failure according to an embodiment of the present application;

FIG. 2 schematically illustrates a flow diagram of a method for determining a type of switch fault according to an embodiment of the present application;

FIG. 3 schematically illustrates a flow diagram of a method for determining a type of switch fault according to another embodiment of the present application;

FIG. 4 schematically shows a structural diagram of a BP neural network according to an embodiment of the present application;

FIG. 5 schematically shows a step process diagram of a BP neural network according to an embodiment of the present application;

fig. 6 schematically shows an internal structure diagram of a computer device according to an embodiment of the present application.

Detailed Description

The following detailed description of embodiments of the present application will be made with reference to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the embodiments of the application, are given by way of illustration and explanation only, not limitation.

The method for determining the type of the switch fault provided by the embodiment of the application can be applied to an application environment as shown in fig. 1. As shown in fig. 1, the intelligent switch 102 communicates with the processor 104 over a network. The intelligent switch 102 may upload real-time operation data corresponding to the intelligent switch in an actual operation process to the processor 104 through a network, and the processor 104 may preprocess the real-time operation data and input the preprocessed real-time operation data to the fault recognition model to determine a fault type of the intelligent switch 102. The smart switch 102 is one of important devices in an electric power system, has dual functions of controlling and protecting the electric power system, and can be used for distributing electric energy and protecting a power line, a motor and the like.

Fig. 2 schematically shows a flow diagram of a method for determining a type of switch failure according to an embodiment of the application. As shown in fig. 2, in an embodiment of the present application, there is provided a method for determining a type of switch failure, comprising the steps of:

step 201, acquiring real-time operation data of the switch.

Step 202, preprocessing the real-time operation data.

And step 203, inputting the preprocessed real-time operation data into a fault identification model.

And step 204, determining the fault type of the switch through a fault identification model.

The switch can be an intelligent switch, has a communication function, has dual functions of controlling and protecting a power system, and can be used for distributing electric energy and protecting a power supply circuit, a motor and the like. When the power system has serious overload or short circuit, undervoltage and open circuit, it can automatically cut off the circuit, and its function is equivalent to the combination of fuse switch and over-and-under-heat relay. Intelligent switches are widely used in, for example, power distribution systems, power transmission systems, and electrical equipment to protect electrical equipment and cables. The intelligent switch in this application possesses communication function, can be with the real-time operation data transmission to the treater of self in the use. In particular, the real-time operating data may include at least one of current, voltage, current harmonics, voltage harmonics, and joint temperature acquired by the smart switch. The joint temperature refers to the temperature of the switch and can be acquired through a temperature sensor, and the temperature sensor can be arranged on the switch so that the temperature of the switch can be acquired in real time.

The processor can be used for preprocessing the real-time operation data after acquiring the real-time operation data of the intelligent switch, and can be used for normalizing a plurality of operation parameters included in the real-time operation data through the preprocessing process.

In one embodiment, pre-processing the real-time operational data comprises: and normalizing the operation parameters contained in the real-time operation data to scale the operation parameters to be within a preset range.

In actual operation, there are various types of faults that may occur in the intelligent switch, such as overload, short circuit, undervoltage, open circuit, and the like. When each fault type occurs, the real-time operation data corresponding to the intelligent switch is correspondingly changed. Therefore, it can be understood that whether the intelligent switch has a fault or not and the type of the fault can be analyzed according to the real-time operation data of the intelligent switch. For example, assuming that there are N different fault types (overload, short circuit, undervoltage, open circuit, etc.), the number of features for each fault type is M. { m₁,m₂,m₃...m_MIs e.g. M, where M₁Voltage, m, obtained for the intelligent switch₂Current, m, obtained for the intelligent switch₃The data such as voltage harmonic acquired by the intelligent switch. These real-time operational data can be preprocessed by min-max normalization to scale the feature data to [0, 1%]Or [ -1,1 [ ]]Within the scope, to achieve a standardized processing of the data.

In one embodiment, pre-processing the real-time operational data comprises: scaling the real-time operation data to a preset range according to a formula (1):

wherein x is real-time operation data, x' is preprocessed data, A_minAnd A_maxRespectively, the minimum and maximum values in x.

After the real-time operation data are preprocessed, the processor can input the preprocessed real-time operation data into the fault recognition model, the fault recognition model can extract the characteristics of the input real-time operation data, and the fault type of the intelligent switch can be predicted according to the extracted characteristics. The fault identification model may be a neural network model, which may include, for example, a back-propagation neural network.

In one embodiment, determining the type of fault of the switch via the fault identification model comprises: acquiring fault types output by a fault identification model and prediction probabilities corresponding to the fault types; and determining the fault type corresponding to the maximum prediction probability as the fault type of the switch.

The input data of the fault identification model is real-time operation data after preprocessing. The output data of the fault identification model is a plurality of predicted fault types and the corresponding prediction probability of each fault type. For example, the output of the fault identification model is: the output of overload (20%), short circuit (85%), undervoltage (60%) and open circuit (75%) shows that the fault recognition model judges that the probability of overload fault occurrence of the intelligent switch is 20%, the probability of short circuit fault occurrence is 85%, the probability of undervoltage fault occurrence is 60% and the probability of open circuit fault occurrence is 75% according to the input data. In this case, the fault type corresponding to the prediction probability with the largest value may be determined as the fault type of the intelligent switch, for example, the short circuit corresponding to 85% with the largest value in the above example may be determined as the fault type of the intelligent switch.

Further, in one embodiment, the method further comprises: comparing the prediction probability of the fault type with a preset probability threshold; and executing a protection action corresponding to the fault type under the condition that the prediction probability of the fault type is greater than a preset probability threshold.

After the fault type of the intelligent switch is determined through the fault identification model, whether the fault is processed needs to be determined. For example, assume that the output of the fault identification model is: overload (20%), short (40%), undervoltage (35%), open (55%). At this time, it can be determined that the open circuit corresponding to the 55% with the maximum prediction probability value is the fault type which the fault identification model determines that the switch may have. In practice, the treatment process after the conclusion is also considered. When the probability of the occurrence of the open circuit fault in the smart switch is 55%, it indicates that the smart switch is not necessarily actually open-circuited, and therefore, it is necessary to further confirm the prediction result of the fault recognition model. Specifically, a preset probability threshold may be preset, and then the finally determined prediction probability of the fault type may be compared with the preset probability threshold. If the finally determined prediction probability of the fault type is greater than the preset probability threshold, it can be determined that the intelligent switch is actually in fault and a protection action corresponding to the fault type needs to be executed on the intelligent switch; otherwise, it is determined that the intelligent switch has not failed, and corresponding protection actions do not need to be executed. For example, according to the above example, the open circuit fault corresponding to 55% of the maximum prediction probability value may be determined as the final prediction result of the fault identification model. Namely, the type of the intelligent switch with the fault is determined to be open circuit according to the fault identification model. At this time, the predicted probability 55% corresponding to the open circuit is compared with the preset probability threshold 85%. Obviously, the predicted probability 55% of the fault type of the switch is less than the preset probability threshold 85%, and at this time, the corresponding protection action does not need to be executed on the intelligent switch. If the prediction probability corresponding to the open circuit is 90%, and the prediction probability 90% of the fault type of the switch is smaller than the preset probability threshold 85%, then the corresponding protection action needs to be executed on the intelligent switch to process the open circuit fault of the intelligent switch.

In one embodiment, performing the protection action corresponding to the fault type includes: searching a database for a protection action corresponding to the fault type; and under the condition that the protection action corresponding to the fault type is found, executing the corresponding protection action on the switch.

In order to effectively process the fault after the fault type of the intelligent switch is determined, the intelligent switch is effectively protected, and a corresponding database can be established according to historical experience and historical operation processing. The database contains the protection actions that the intelligent switch can take in the event of each fault type. For example, the protection action that the intelligent switch can take when the overload fault occurs is determined as A; when the intelligent switch has short-circuit fault, the protection action can be taken as B; when the intelligent switch has an undervoltage fault, the protection action which can be adopted is taken as C; when the intelligent switch has an open circuit fault, the protection action which can be taken is D, and the like. In the case where it is determined that the predicted probability of the fault type is greater than the preset probability threshold, it may be looked up in the database whether a protection action corresponding to the fault type is included. If the protection action corresponding to the fault type is found in the database, the protection action corresponding to the fault type stored in the database can be executed on the intelligent switch. Otherwise, if the protection action corresponding to the fault type is not found in the database, the real-time operation data of the intelligent switch can be uploaded to the server, namely to the remote control center, so that the electric equipment related to the intelligent switch can be monitored and/or controlled through the remote center. Specifically, the real-time operation data of the intelligent switch can be pushed to a remote control center through a remote 4G or 5G communication interface, so that the on-line monitoring and the remote control of the electric equipment are realized.

According to the method for determining the switch fault type, the real-time operation data of the switch are acquired, the real-time operation data are preprocessed, the preprocessed real-time operation data are input into the fault recognition model, and the fault type of the switch is determined through the fault recognition model. The method is characterized in that a deep neural network model is built based on a deep learning technology, the fault type of an intelligent switch is output through the neural network model, corresponding protection actions are carried out on the intelligent switch, the protection circuit and the electric equipment are accurately and reliably protected, potential safety hazards such as over-temperature, overload, short circuit and grounding of the electric circuit can be timely found, the actions are fast, and protection intellectualization is realized. Compared with the traditional switch which only has a short-circuit fault protection function, the intelligent power grid fault protection switch has the advantages of single function, low measurement precision, no pre-detection function before power-on, no detection function in the power-on process, failure reason and position cannot be determined, no communication capability exists, stateless alarm exists, the power-off rated current is large, high temperature and electric spark are easily generated when a fault is generated, manual switching-on and power-on are needed after the fault is eliminated, and the traditional power distribution protection scheme needs a master station to issue a control command, the information quantity needing to be processed by the master station is very large, a decision cannot be made in time, the intelligent power grid fault protection switch is influenced by communication delay and the like, timely processing and fine management of the fault cannot be realized, and the requirement for intelligent power grid construction cannot be met. And intelligent switch in this application can initiatively acquire each electric power parameter of the circuit of being connected with it and each consumer to can accurately determine the fault type that circuit and consumer took place through the fault identification model, in order can carry out corresponding protection action fast, can be accurate, protect circuit and consumer reliably, in time discover potential safety hazards such as the overtemperature of electric circuit, overload, short circuit and ground connection, the fast motion realizes the intellectuality of protection, user's power consumption security has also been improved.

In one embodiment, as shown in fig. 3, there is also provided a method for determining a type of switch failure by a user, comprising:

step 301, acquiring real-time operation data of the switch.

Step 302, normalizing the operation parameters included in the real-time operation data to scale the operation parameters to a preset range.

And step 303, inputting the preprocessed real-time operation data into a fault identification model.

And step 304, acquiring the fault types output by the fault identification model and the prediction probability corresponding to each fault type.

And step 305, determining the fault type corresponding to the maximum prediction probability as the fault type of the switch.

Step 306, the predicted probability of the fault type is compared with a preset probability threshold.

And 307, searching a protection action corresponding to the fault type in the database under the condition that the prediction probability of the fault type is greater than a preset probability threshold.

Step 308, judging whether the protection action corresponding to the fault type is found, if so, executing step 309; if not, go to step 310.

Step 309, perform the corresponding protection action on the switch.

And step 310, uploading the real-time operation data to a remote control center so as to monitor and/or control the switch through the remote center.

The intelligent switch has a communication function and can transmit real-time running data of the intelligent switch in the using process to the processor. In particular, the real-time operating data may include at least one of current, voltage, current harmonics, voltage harmonics, and joint temperature acquired by the smart switch. In actual operation, there are various types of faults that may occur in the intelligent switch, such as overload, short circuit, undervoltage, open circuit, and the like. When each fault type occurs, the real-time operation data corresponding to the intelligent switch is correspondingly changed. Therefore, it can be understood that whether the intelligent switch has a fault or not and the type of the fault can be analyzed according to the real-time operation data of the intelligent switch. The processor can be used for preprocessing the real-time operation data after acquiring the real-time operation data of the intelligent switch, and can be used for normalizing a plurality of operation parameters included in the real-time operation data through the preprocessing process. The standard processing is carried out on the sample data of the fault identification model, so that the inconsistency of the data can be eliminated, the calculated amount is reduced, and the convergence speed of the fault identification model is accelerated. Meanwhile, the training data are weighted through the weight vector matrix, so that indexes which are beneficial to the model are strengthened, indexes which are unfavorable to the model are weakened, and accurate identification of the fault type is achieved.

After the real-time operation data are preprocessed, the processor can input the preprocessed real-time operation data into the fault recognition model, the fault recognition model can extract the characteristics of the input real-time operation data, and the fault type of the intelligent switch can be predicted according to the extracted characteristics. The fault identification model may be a neural network model, which may include, for example, a back-propagation neural network. The back propagation neural network may be a BP neural network. The input data of the fault identification model is real-time operation data after preprocessing. The output data of the fault identification model is a plurality of predicted fault types and the prediction probability corresponding to each fault type, and the fault type corresponding to the prediction probability with the maximum value can be determined as the fault type of the intelligent switch. After the fault type of the intelligent switch is determined through the fault identification model, whether the fault is processed needs to be determined. The processor may further perform another confirmation on the prediction result of the fault recognition model, preset a preset probability threshold, and compare the finally determined prediction probability of the fault type with the preset probability threshold. In the case where it is determined that the predicted probability of the fault type is greater than the preset probability threshold, it may be looked up in the database whether a protection action corresponding to the fault type is included. If the database is searched for the protection action corresponding to the fault type, the protection action corresponding to the fault type stored in the database may be executed on the intelligent switch. Otherwise, if the protection action corresponding to the fault type is not found in the database, the real-time operation data of the intelligent switch can be uploaded to the server, namely to the remote control center, so that the electric equipment related to the intelligent switch can be monitored and/or controlled through the remote center.

Further, after the fault type of the intelligent switch is determined through the fault identification model, the finally determined prediction probability of the fault type is compared with the preset probability threshold, and the difference value between the two types is determined. And determining the level of the fault according to the difference. Specifically, under the condition that the prediction probability of the finally determined fault type is greater than a preset probability threshold and the difference between the prediction probability and the preset probability threshold reaches a first difference threshold, the fault level can be determined to be a high level; determining the fault level as a medium level under the condition that the prediction probability is greater than a preset probability threshold value, and the difference value between the prediction probability and the preset probability threshold value is greater than a second difference value threshold value and smaller than a first difference value threshold value; and under the condition that the prediction probability is greater than a preset probability threshold value and the difference between the prediction probability and the preset probability threshold value is greater than a third difference threshold value and less than a second difference threshold value, determining the fault level as a low level, wherein the first difference threshold value > the second difference threshold value > the third difference threshold value. Then, a protection action specific to the fault may be determined according to the level of the fault. Correspondingly, when the database stores the protection actions corresponding to each fault type, the protection actions may be subdivided according to the level corresponding to each fault type. For example, there are three types of fault levels of short-circuit fault, i.e., high, medium, and low, and a protection action corresponding to each fault level is stored.

In one embodiment, before the fault recognition model is put into practical use, the fault recognition model may be trained, and then the fault type of the switch may be determined by using the trained fault recognition model. Specifically, the training step of the fault recognition model comprises the following steps: acquiring historical operating data of a plurality of switches; preprocessing historical operating data; and inputting the preprocessed historical operating data into a fault recognition model so as to train the fault recognition model.

The historical operation data of the switch refers to operation data of the switch in a historical period of time, and comprises operation data when the intelligent switch is not in fault and operation data corresponding to the intelligent switch when various types of faults occur. Therefore, historical operation data of the intelligent switches can be acquired, and the acquired historical operation data can be preprocessed. The pre-processing operation on the historical operating data and the pre-processing operation on the real-time operating data may be consistent, that is, the historical operating data is standardized, and a plurality of operating parameters contained in the historical operating data are scaled to be within a preset range, for example [0,1] or [ -1,1 ]. And then, inputting the preprocessed historical operating data into a fault recognition model, and training the fault recognition model. The operating parameters may refer to power parameters acquired by the smart switch, such as current, voltage, current harmonics, voltage harmonics, and junction temperature acquired by the smart switch.

In one embodiment, inputting the preprocessed historical operating data into the fault recognition model to train the fault recognition model comprises: performing weighted calculation on the historical operating data through a weight vector matrix of the fault identification model to obtain updated historical operating data; adjusting the weight of the fault identification model through the updated historical operation data; determining an error value of the fault identification model; and determining that the fault recognition model is trained completely under the condition that the error value is lower than a preset error threshold value.

After the preprocessed historical operating data is input into the fault identification model, the power parameters can be further processed through the weight vector matrix of the fault identification model. The established mathematical model reveals the integrally associated behavior microscopic features in the macroscopic data, which is beneficial to grasping more essential data information, strengthens indexes favorable for the data information, weakens indexes unfavorable for the data information, and thus realizes accurate identification of fault types. Specifically, in one embodiment, the expression of the weight vector matrix is defined as shown in the following equation (7):

wherein, T_nExpressed as a matrix of weight vectors and r as the weight of the eigenvalues.

Weighting each power parameter by a weight vector matrix, i.e. feature calculationThereafter, updated historical operating data may be obtained. For example, if there are N different fault types (overload, short circuit, undervoltage, open circuit, etc.), the number of features corresponding to each fault type is M, that is, the power parameter of each fault type is different, and if the type of the power parameter is M, the number of features corresponding to each fault type is M. E.g. m₁Voltage, m, obtained for the intelligent switch₂Current, m, obtained for the intelligent switch₃Data such as voltage harmonics, m, acquired for a smart switch_M…, etc., and m₁,m₂,m₃,...,m_MIs ∈ M. That is, for each fault type, there is a feature set m corresponding to it_nFeature set m corresponding to each fault type_nIn, M power parameters are included, each being M₁,m₂,m₃,...,m_M. Aiming at feature set m corresponding to any fault type_nA set of feature m formed by the weight vector matrix to the power parameters_nTo obtain a corresponding updated feature set m'_n. Specifically, the updated feature set m 'may be represented according to the following calculation formula'_n:

m′_n＝m₁*r_1n+m₂*r_2n+…m_M*r_MnWherein r is_1n，r_2n，…，r_MnRepresented as a weight corresponding to each power parameter. After the power parameters are weighted and calculated in this way, corresponding updated historical operating data can be obtained.

In one embodiment, the fault identification model comprises a back propagation neural network, which may be in particular a BP neural network. The calculation process of the BP neural network consists of a forward calculation process and a backward calculation process. And in the forward propagation process, the input mode is processed layer by layer from the input layer through the hidden unit layer and is transferred to the output layer, and the state of each layer of neurons only affects the state of the next layer of neurons. If the expected output can not be obtained at the output layer, the reverse propagation is carried out, the error signal is returned along the original connecting path, and the weight of each neuron is modified to minimize the error signal. A BP neural network algorithm model is adopted, and the error of the whole network is minimized by utilizing back propagation mainly through a gradient descent method. And learning and training the BP neural network through a plurality of power parameters contained in the collected historical operating data so as to complete the optimization and training process of the model.

In one embodiment, the fault identification model includes an input layer, a hidden layer, and an output layer. Adjusting the weight of the fault identification model through the updated historical operating data comprises: and (3) adjusting the weight of each network layer of the fault identification model according to the formula (2) and the formula (3):

Δw_jk＝ηy_j(d_j-y_j)(1-y_j)y_kformula (2);

wherein, w_jkIs the connection weight between hidden layer neuron k and output layer neuron j, v_kiIs the connection weight between the input layer neuron i and the hidden layer neuron k, eta is the preset learning rate, y_jRepresenting the output result of the network layer, d_jTo desired output, y_kAs output result of the hidden layer, x_iIs input into the input layer.

As shown in fig. 4, the fault recognition model includes an input layer (status information input), a hidden layer (status information training, learning), and an output layer (information output), and the data information in the above-described power parameter feature set is input through the information input layer. In order to achieve a certain adjusting effect, different weights or thresholds are set for each network layer of the fault recognition model, so that the training precision of the model is improved, and the calculation error is reduced. The fault identification model can be a BP neural network, and the BP neural network mainly comprises two processes of forward propagation and backward propagation. Assuming that the input layer has n neurons, use x_i(i 1,2,3.., n) represents an input signal, the output layer has m neurons, and y represents an input signal_j(j ═ 1,2,3.., m) represents the output result, with q neurons in the hidden layer,by z_k(k ═ 1,2,3.., q) denotes the intermediate calculation results. Then, the training process of the BP neural network is as shown in fig. 5, specifically, the training process of the BP neural network is as follows:

1. initialization:

randomly assigning numbers to the weight matrixes v and w of each network layer, setting the number p of samples input into the BP neural network and the training times q of the BP as 1, resetting the error E of the BP neural network to zero, setting the learning rate eta as a decimal within 0 to 1, and setting the precision E achieved after the BP neural network is trained_minA fraction greater than zero is set.

2. Inputting training data, calculating each layer output:

and inputting the historical operating data into the BP neural network, and calculating the output of each network layer contained in the BP neural network. Specifically, the output of each network layer may be determined by the following equations (4) and (5):

wherein z is_kRepresenting the intermediate calculation result, the value range of k is [1, q ]]Q is the number of neurons in the hidden layer, y_jThe output result of the network layer is shown, and the value range of j is [1, m]M is the number of neurons in the output layer, x_iRepresenting input data of the network layer, i having a value range of [1, n]N is the number of neurons in the input layer, f₁(x) For the excitation function of the hidden layer, f₂(x) Is the excitation function of the output layer. Wherein the excitation function is defined as shown in equation (6):

wherein, theta₁For varying parameters of function amplitude values，θ₂To change the parameters of the variable elasticity, x is the input data of the network layer.

3. Calculating the error of each network layer:

since each network layer has a certain error in transferring data. I.e. from y_jTo z_k、z_kTo x_iThere will be error propagation in between. Assuming that the number of training samples input to the BP neural network for training is n: { (x (1), y (1)), (x (2), y (2)),. x (i) is expressed as an input signal representing the network layer, y (i) is expressed as an output result representing the network layer, d (i) is expressed as an expected output of the input x (i), and then the error value E of the BP network can be calculated by the formula (8):

where E is an error value, n is the number of training samples, and E (i) is a training error of the ith training sample. E (i) is calculated by the following formula (9):

wherein d (i) represents the expected output of input x (i), x (i) being the input data for the fault identification model; k is a hidden layer neuron.

4. Adjusting the weight of each network layer:

specifically, the weight of each network layer of the fault identification model may be adjusted according to the following formula (2) and formula (3):

Δw_jk＝ηy_j(d_j-y_j)(1-y_j)y_kformula (2);

wherein, w_jkIs the connection weight between hidden layer neuron k and output layer neuron j, v_kiIs the connection weight between the input layer neuron i and the hidden layer neuron k, eta is the preset learning rate, y_jRepresenting the output result of the network layer, d_jTo desired output, y_kAs output result of the hidden layer, x_iIs input into the input layer.

5. Completing training of all training samples:

in the case where the number p of input samples is smaller than the total number S of training samples, both the counters p and q are incremented by 1, and in this way, training of all samples is completed. In case the number p of input samples is larger than the total number S of training samples, the next step may be entered.

6. Determining whether the total error of the BP neural network reaches a preset precision standard:

and under the condition that the error value of the BP neural network is lower than a preset error threshold value, determining a fault recognition model, namely, finishing training of the BP neural network.

In one embodiment, the method further comprises: adding real-time operation data of the fault type into a historical data set so as to update the historical data set; and training the fault type through the updated historical data set.

After the fault type of the intelligent switch is identified through the fault identification model, the currently acquired real-time operation data of the intelligent switch can be added into a training data set of the fault identification model, namely the real-time operation data of the intelligent switch is added into a historical data set, so that the training data set of the fault identification model is updated, the training data of the fault identification model is enriched, the fault type can be trained through the updated training set, the fault identification model is optimized, and the prediction accuracy of the fault identification model is improved.

The embodiment of the application provides a processor, which is used for running a program, wherein the program runs to execute the method for determining the type of the switch fault.

In one embodiment, an apparatus for determining a switch fault type is provided, which includes the processor described above, and may determine the fault type of the intelligent switch by a method for determining the switch fault type executed by the processor, and perform a corresponding protection action on the associated circuit and the electrical device.

The processor comprises a kernel, and the kernel calls the corresponding program unit from the memory. The kernel can be provided with one or more, and the method for determining the switch fault type is realized by adjusting the kernel parameters.

The memory may include volatile memory in a computer readable medium, Random Access Memory (RAM) and/or nonvolatile memory such as Read Only Memory (ROM) or flash memory (flash RAM), and the memory includes at least one memory chip.

Embodiments of the present application provide a storage medium having stored thereon a program that, when executed by a processor, implements the above-described method for determining a type of switch failure.

In one embodiment, there is also provided a power distribution apparatus comprising the above-described apparatus for determining a type of switch fault.

In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 6. The computer device includes a processor a01, a network interface a02, a memory (not shown), and a database (not shown) connected by a system bus. Wherein processor a01 of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises an internal memory a03 and a non-volatile storage medium a 04. The non-volatile storage medium a04 stores an operating system B01, a computer program B02, and a database (not shown in the figure). The internal memory a03 provides an environment for the operation of the operating system B01 and the computer program B02 in the nonvolatile storage medium a 04. The database of the computer device is used for storing the operation data of the intelligent switch and the like. The network interface a02 of the computer device is used for communication with an external terminal through a network connection. The computer program B02 is executed by the processor a01 to implement a method for determining a type of switch failure.

Those skilled in the art will appreciate that the architecture shown in fig. 6 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

An embodiment of the present application provides an apparatus comprising a processor, a memory, and a program stored on the memory and executable on the processor, the processor implementing the steps of the method for determining a type of switch failure when executing the program.

The present application further provides a computer program product adapted to perform a program for initializing the method steps for determining the type of switch failure as follows, when executed on a data processing device.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.

Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in the process, method, article, or apparatus that comprises the element.

The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.