阅读说明：本技术 基于云边协同的空压机集群运行调节方法及系统 (Air compressor cluster operation adjusting method and system based on cloud edge cooperation ) 是由 周开乐 张增辉 胡定定 费志能 郭金环 于 2021-07-28 设计创作，主要内容包括：本发明提供一种基于云边协同的空压机集群运行调节方法及系统,涉及空压机调节技术领域。本发明将获取的空压机集群中所有空压机运行数据发送至云端；并对空压机运行数据进行异常检测和数据预测处理,然后基于异常检测数据和预测值进行故障预警和执行运行决策,最后利用强化学习的思想对运行决策执行结果进行记录和反馈以优化下一轮决策过程。本发明不仅节约了大量本地算力和存储空间,提高了调节效率,同时还提高了决策的可靠性和质量,使得空压机集群运行调节的自动化程度高,同时调节决策质量高。(The invention provides an air compressor cluster operation adjusting method and system based on cloud-edge coordination, and relates to the technical field of air compressor adjustment. The method comprises the steps of sending all acquired air compressor operation data in an air compressor cluster to a cloud end; and performing anomaly detection and data prediction processing on the operation data of the air compressor, performing fault early warning and operation decision execution based on the anomaly detection data and the predicted value, and finally recording and feeding back the operation decision execution result by using the idea of reinforcement learning so as to optimize the next decision process. The invention not only saves a large amount of local computing power and storage space, improves the regulation efficiency, but also improves the reliability and quality of decision making, so that the automation degree of the operation regulation of the air compressor cluster is high, and the regulation decision making quality is high.)

1. The air compressor cluster operation adjusting method based on cloud-edge coordination is characterized by comprising the following steps:

sending the acquired operation data of all air compressors in the air compressor cluster to a cloud terminal;

performing anomaly detection and data prediction processing on the air compressor operation data at a cloud end to obtain anomaly data and a predicted value;

performing fault early warning and executing operation decision on the side based on the abnormal data and the predicted value;

and recording and feeding back an operation decision execution result based on the idea of reinforcement learning so as to optimize the next operation decision of the air compressor.

2. The method of claim 1, wherein the method further comprises: collecting the operation data of all air compressors in the air compressor cluster, and preprocessing the operation data of the air compressors.

3. The method of claim 1, wherein the performing of the anomaly detection and data prediction processing on the air compressor operation data at the cloud end comprises: judging whether the operation data of the air compressor is abnormal or not by using a multivariate statistical regression method; and carrying out data prediction processing on the air compressor operation data by utilizing a multivariate linear model.

4. The method of claim 1, wherein the performing fault early warning and operational decisions on the side based on the anomaly data and predicted values comprises:

s31, performing fault early warning on the side based on the abnormal data; the corresponding relation between the abnormal data and the fault early warning is as follows:

if no abnormal data exists, no alarm signal exists;

if the calculated parameter of the abnormal detection satisfies a < x < b, a three-level alarm signal is sent out;

if the calculated parameter of the abnormal detection meets x > b, a secondary alarm signal is sent out;

if the parameters calculated by the abnormality detection of the plurality of air compressors all meet x > b, a primary alarm signal is sent out;

wherein x represents a parameter calculated by abnormality detection; a and b are judgment threshold values when whether abnormal data exist or not is detected;

s32, executing operation decision at the side based on the predicted value; the corresponding relation between the predicted value and the operation decision is as follows:

if 0 < | X₁|-|X₂If the | is less than or equal to m, the variable frequency air compressor carries out pre-load shedding operation;

if | X₁|-|X₂If the | is greater than m, pre-shutting down a certain air compressor;

if-m < | X₁|-|X₂If the | is less than or equal to 0, performing preloading operation on the variable frequency air compressor;

if | X₁|-|X₂If the | is less than or equal to-m, a certain air compressor is started in advance;

wherein, X₁Representing the current working capacity of the air compressor; x₂Expressing the predicted value of the workload of the air compressor; i X₁|-|X₂L represents the difference between the current value and the predicted value; m represents a threshold value; | X | represents the modulus of the vector X,

5. the method of claim 1, wherein the recording and feeding back the execution result of the operation decision based on the idea of reinforcement learning to optimize the next operation decision of the air compressor comprises:

s41, recording the operation decision and the operation decision execution result given by the system each time;

s42, if the execution result of the operation decision conforms to the actual operation condition of the system at the next moment, encapsulating the system operation decision and the decision instruction at the moment into a reward group; if the execution result of the operation decision given by the system is contrary to the actual operation condition of the system at the next moment, encapsulating the operation decision, the decision instruction and the deviation value of the instruction and the actual operation of the system at the moment into a penalty group;

and S43, optimizing the next operation decision of the air compressor based on the reward group and the penalty group.

6. The utility model provides an air compressor machine cluster operation governing system based on cloud limit is cooperative, its characterized in that, the system includes:

the data sending module is used for sending the acquired running data of all the air compressors in the air compressor cluster to the cloud;

the anomaly detection and prediction module is used for performing anomaly detection and data prediction processing on the air compressor operation data at the cloud end to obtain anomaly data and a predicted value;

the early warning and operation decision module is used for carrying out fault early warning and executing operation decision on the side based on the abnormal data and the predicted value;

and the operation result recording and feedback module is used for recording and feeding back the operation decision execution result based on the idea of reinforcement learning so as to optimize the next operation decision of the air compressor.

7. The system of claim 6, wherein the system further comprises: and the data acquisition and preprocessing module is used for acquiring the operation data of all air compressors in the air compressor cluster and preprocessing the operation data of the air compressors.

8. The system of claim 6, wherein the anomaly detection and prediction module performs anomaly detection and data prediction processing on the air compressor operation data at a cloud side comprises: judging whether the operation data of the air compressor is abnormal or not by using a multivariate statistical regression method; and carrying out data prediction processing on the air compressor operation data by utilizing a multivariate linear model.

9. The system of claim 6, wherein the early warning and operational decision module to perform fault early warning and operational decision wayside based on the anomaly data and predicted values comprises:

s31, performing fault early warning on the side based on the abnormal data; the corresponding relation between the abnormal data and the fault early warning is as follows:

if no abnormal data exists, no alarm signal exists;

if the calculated parameter of the abnormal detection satisfies a < x < b, a three-level alarm signal is sent out;

if the calculated parameter of the abnormal detection meets x > b, a secondary alarm signal is sent out;

if the parameters calculated by the abnormality detection of the plurality of air compressors all meet x > b, a primary alarm signal is sent out;

wherein x represents a parameter calculated by abnormality detection; a and b are judgment threshold values when whether abnormal data exist or not is detected;

s32, executing operation decision at the side based on the predicted value; the corresponding relation between the predicted value and the operation decision is as follows:

if 0 < | X₁|-|X₂If the | is less than or equal to m, the variable frequency air compressor carries out pre-load shedding operation;

if | X₁|-|X₂If the | is greater than m, pre-shutting down a certain air compressor;

if-m < | X₁|-|X₂If the | is less than or equal to 0, performing preloading operation on the variable frequency air compressor;

if | X₁|-|X₂If the | is less than or equal to-m, a certain air compressor is started in advance;

wherein, X₁Representing the current working capacity of the air compressor; x₂Expressing the predicted value of the workload of the air compressor; i X₁|-|X₂L represents the difference between the current value and the predicted value; m meterIndicating a threshold value; | X | represents the modulus of the vector X,

10. the system of claim 6, wherein the operation result recording and feedback module records and feeds back the operation decision execution result based on the idea of reinforcement learning to optimize the next operation decision of the air compressor comprises:

s41, recording the operation decision and the operation decision execution result given by the system each time;

s42, if the execution result of the operation decision conforms to the actual operation condition of the system at the next moment, encapsulating the system operation decision and the decision instruction at the moment into a reward group; if the execution result of the operation decision given by the system is contrary to the actual operation condition of the system at the next moment, encapsulating the operation decision, the decision instruction and the deviation value of the instruction and the actual operation of the system at the moment into a penalty group;

and S43, optimizing the next operation decision of the air compressor based on the reward group and the penalty group.

Technical Field

The invention relates to the technical field of air compressor adjustment, in particular to an air compressor cluster operation adjusting method and system based on cloud-edge cooperation.

Background

As terminal energy utilization equipment widely used by key energy utilization enterprises, the proportion of the air compressors (which generally appear in an air compressor cluster mode) in the power consumption of the enterprises is very high, so that how to improve the energy efficiency level of the air compressors has important significance in promoting energy conservation, consumption reduction and green development of the key energy utilization enterprises. At present, compared with a common technology for improving the energy efficiency level of an air compressor, the operation of the air compressor is reasonably regulated and controlled, so that the no-load operation of the air compressor is reduced to the maximum extent, the electric energy consumption of an enterprise can be greatly reduced, and the air compressor cluster regulation energy-saving point is realized.

The traditional air compressor operation adjustment is air supply adjustment through continuous loading and unloading, the air compressor set can continuously impact an electric network in the frequent loading and unloading process in the adjustment mode, the service life of equipment is shortened, the adjustment speed is low, and the automation degree is low; the air compressor is adjusted by adopting a variable-frequency speed regulation adjusting mode, and the air compressor can be adjusted only in real time according to the current workload, so that the equipment can make an error decision which is only beneficial to the operation at the current moment, the reliability is poor, and the long-term operation of the equipment is not facilitated; meanwhile, the response speed of the equipment is low, a feedback link is not provided, feedback adjustment cannot be performed on an improper adjustment instruction, and the adjustment decision quality is not high.

Therefore, the problems of low adjusting efficiency, low adjusting decision quality and the like exist in the conventional air compressor operation adjusting technology.

Disclosure of Invention

Technical problem to be solved

Aiming at the defects of the prior art, the invention provides an air compressor cluster operation adjusting method and system based on cloud-edge coordination, and solves the problems of low adjusting efficiency and low adjusting decision quality in air compressor operation adjustment in the prior art.

(II) technical scheme

In order to achieve the purpose, the invention is realized by the following technical scheme:

in a first aspect, the present invention first provides a cloud-edge coordination-based air compressor cluster operation adjustment method, where the method includes:

sending the acquired operation data of all air compressors in the air compressor cluster to a cloud terminal;

performing anomaly detection and data prediction processing on the air compressor operation data at a cloud end to obtain anomaly data and a predicted value;

performing fault early warning and executing operation decision on the side based on the abnormal data and the predicted value;

and recording and feeding back an operation decision execution result based on the idea of reinforcement learning so as to optimize the next operation decision of the air compressor.

Preferably, the method further comprises: collecting the operation data of all air compressors in the air compressor cluster, and preprocessing the operation data of the air compressors.

Preferably, the performing, at the cloud, the abnormality detection and data prediction processing on the air compressor operation data includes: judging whether the operation data of the air compressor is abnormal or not by using a multivariate statistical regression method; and carrying out data prediction processing on the air compressor operation data by utilizing a multivariate linear model.

Preferably, the performing fault early warning and executing operation decision at the side based on the abnormal data and the predicted value includes:

s31, performing fault early warning on the side based on the abnormal data; the corresponding relation between the abnormal data and the fault early warning is as follows:

if no abnormal data exists, no alarm signal exists;

if the calculated parameter of the abnormal detection satisfies a < x < b, a three-level alarm signal is sent out;

if the calculated parameter of the abnormal detection meets x > b, a secondary alarm signal is sent out;

if the parameters calculated by the abnormality detection of the plurality of air compressors all meet x > b, a primary alarm signal is sent out;

wherein x represents a parameter calculated by abnormality detection; a and b are judgment threshold values when whether abnormal data exist or not is detected;

s32, executing operation decision at the side based on the predicted value; the corresponding relation between the predicted value and the operation decision is as follows:

if 0 < | X₁|-|X₂If the | is less than or equal to m, the variable frequency air compressor carries out pre-load shedding operation;

if | X₁|-|X₂If the | is greater than m, pre-shutting down a certain air compressor;

if-m < | X₁|-|X₂If the | is less than or equal to 0, performing preloading operation on the variable frequency air compressor;

if | X₁|-|X₂If the | is less than or equal to-m, a certain air compressor is started in advance;

wherein, X₁Representing the current working capacity of the air compressor; x₂Expressing the predicted value of the workload of the air compressor; i X₁|-|X₂L represents the difference between the current value and the predicted value; m represents a threshold value; | X | represents the modulus of the vector X,

preferably, the recording and feeding back the operation decision execution result based on the idea of reinforcement learning to optimize the next operation decision of the air compressor includes:

s41, recording the operation decision and the operation decision execution result given by the system each time;

s42, if the execution result of the operation decision conforms to the actual operation condition of the system at the next moment, encapsulating the system operation decision and the decision instruction at the moment into a reward group; if the execution result of the operation decision given by the system is contrary to the actual operation condition of the system at the next moment, encapsulating the operation decision, the decision instruction and the deviation value of the instruction and the actual operation of the system at the moment into a penalty group;

and S43, optimizing the next operation decision of the air compressor based on the reward group and the penalty group.

In a second aspect, the present invention further provides a cloud-edge coordination based air compressor cluster operation adjusting system, where the system includes:

the data sending module is used for sending the acquired running data of all the air compressors in the air compressor cluster to the cloud;

the anomaly detection and prediction module is used for performing anomaly detection and data prediction processing on the air compressor operation data at the cloud end to obtain anomaly data and a predicted value;

the early warning and operation decision module is used for carrying out fault early warning and executing operation decision on the side based on the abnormal data and the predicted value;

and the operation result recording and feedback module is used for recording and feeding back the operation decision execution result based on the idea of reinforcement learning so as to optimize the next operation decision process of the air compressor.

Preferably, the system further comprises: and the data acquisition and preprocessing module is used for acquiring the operation data of all air compressors in the air compressor cluster and preprocessing the operation data of the air compressors.

Preferably, the anomaly detection and prediction module performs anomaly detection and data prediction processing on the air compressor operation data at a cloud end, and includes: judging whether the operation data of the air compressor is abnormal or not by using a multivariate statistical regression method; and carrying out data prediction processing on the air compressor operation data by utilizing a multivariate linear model.

Preferably, the performing, by the early warning and operation decision module, a fault early warning and an operation decision at a side based on the abnormal data and the predicted value includes:

s31, performing fault early warning on the side based on the abnormal data; the corresponding relation between the abnormal data and the fault early warning is as follows:

if no abnormal data exists, no alarm signal exists;

if the calculated parameter of the abnormal detection satisfies a < x < b, a three-level alarm signal is sent out;

if the calculated parameter of the abnormal detection meets x > b, a secondary alarm signal is sent out;

if the parameters calculated by the abnormality detection of the plurality of air compressors all meet x > b, a primary alarm signal is sent out;

wherein x represents a parameter calculated by abnormality detection; a and b are judgment threshold values when whether abnormal data exist or not is detected;

s32, executing operation decision at the side based on the predicted value; the corresponding relation between the predicted value and the operation decision is as follows:

if 0 < | X₁|-|X₂If the | is less than or equal to m, the variable frequency air compressor carries out pre-load shedding operation;

if | X₁|-|X₂If the | is greater than m, pre-shutting down a certain air compressor;

if-m < | X₁|-|X₂If the | is less than or equal to 0, performing preloading operation on the variable frequency air compressor;

if | X₁|-|X₂If the | is less than or equal to-m, a certain air compressor is started in advance;

wherein, X₁Representing the current working capacity of the air compressor; x₂Expressing the predicted value of the workload of the air compressor; i X₁|-|X₂L represents the difference between the current value and the predicted value; m represents a threshold value; | X | represents the modulus of the vector X,

preferably, the recording and feedback module of the operation result records and feeds back the execution result of the operation decision based on the idea of reinforcement learning to optimize the next operation decision of the air compressor includes:

s41, recording the operation decision and the operation decision execution result given by the system each time;

s42, if the execution result of the operation decision conforms to the actual operation condition of the system at the next moment, encapsulating the system operation decision and the decision instruction at the moment into a reward group; if the execution result of the operation decision given by the system is contrary to the actual operation condition of the system at the next moment, encapsulating the operation decision, the decision instruction and the deviation value of the instruction and the actual operation of the system at the moment into a penalty group;

and S43, optimizing the next operation decision of the air compressor based on the reward group and the penalty group.

(III) advantageous effects

The invention provides an air compressor cluster operation adjusting method and system based on cloud-edge coordination. Compared with the prior art, the method has the following beneficial effects:

1. the method comprises the steps of sending all acquired air compressor operation data in an air compressor cluster to a cloud end; and carrying out anomaly detection and data prediction processing on the operation data of the air compressor at the cloud end, then carrying out fault early warning and operation decision execution based on the anomaly detection data and the predicted value, and finally recording and feeding back the operation decision execution result by utilizing the idea of reinforcement learning so as to optimize the next decision process. The invention not only saves a large amount of local computing power and storage space, improves the regulation efficiency, but also improves the reliability and quality of decision making, so that the automation degree of the operation regulation of the air compressor cluster is high, and the regulation decision making quality is high.

2. According to the method, a cloud edge cooperation technology is utilized, so that a large amount of local computing power can be saved, data abnormity detection and prediction and data storage operation are all placed at the cloud end, the local storage space is saved, and the operation and adjustment efficiency of the air compressor is improved;

3. the method is based on the decision obtained by the predicted value and the historical result, and has higher reliability; meanwhile, the operation result is continuously recorded and fed back based on the reinforcement learning method, correct decision is reinforced, wrong decision is weakened, and decision quality of the system can be continuously improved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.

Fig. 1 is a flowchart of an air compressor cluster operation adjustment method based on cloud-edge coordination in the embodiment of the present invention;

fig. 2 is a block diagram of an air compressor cluster operation adjustment system based on cloud-edge coordination in an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described, and it is obvious that the described embodiments are a part of the embodiments of the present invention, but not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The embodiment of the application provides an air compressor cluster operation adjusting method and system based on cloud-edge cooperation, and solves the problems that in the prior art, the air compressor operation adjustment is low in adjusting efficiency and adjusting decision quality.

In order to solve the technical problems, the general idea of the embodiment of the application is as follows:

in order to save a large amount of local computing power and storage space, improve the adjustment efficiency of the operation of the air compressor cluster and improve the reliability and quality of operation decision, the invention sends the acquired operation data of all the air compressors in the air compressor cluster to the cloud; and carrying out anomaly detection and data prediction processing on the operation data of the air compressor at the cloud end, then carrying out fault early warning and operation decision execution based on the anomaly detection data and the predicted value, and finally recording and feeding back the operation decision execution result by utilizing the idea of reinforcement learning so as to optimize the next decision process. The invention finally ensures that the automation degree of the operation regulation of the air compressor cluster is high, the efficiency is high, and the regulation decision quality is high.

In order to better understand the technical solution, the technical solution will be described in detail with reference to the drawings and the specific embodiments.

Cloud-edge cooperation refers to cooperative cooperation and combined action of cloud computing and edge computing. The cloud computing has the characteristics of global property, long-term property and good performance in big data computing; compared with cloud computing, the short-period characteristic of edge computing can better support local services, the cloud edge cooperation can better integrate the advantages of cloud computing and edge computing, and the cloud edge cooperation is a distributed open platform integrating communication, computing power, data storage and application services. Under the large framework of cloud edge collaboration, the edge side and the cloud side are not simple alternative relationships, but are complementary collaborative relationships. By constructing a unified and efficient collaborative framework, cloud-edge complementation and resource fusion can be realized through cloud-edge collaboration.

Example 1:

in a first aspect, the present invention first provides a method for adjusting operation of an air compressor cluster based on cloud-edge coordination, and referring to fig. 1, the method includes:

s1, sending the acquired operation data of all air compressors in the air compressor cluster to a cloud terminal;

s2, carrying out anomaly detection and data prediction processing on the air compressor operation data at the cloud end to obtain anomaly data and a predicted value;

s3, performing fault early warning and executing operation decision on the side based on the abnormal data and the predicted value;

and S4, recording and feeding back an operation decision execution result based on the idea of reinforcement learning so as to optimize the next operation decision of the air compressor.

Therefore, the acquired operation data of all the air compressors in the air compressor cluster are sent to the cloud end; and performing anomaly detection and data prediction processing on the operation data of the air compressor, performing fault early warning and operation decision execution based on the anomaly detection data and the predicted value, and finally recording and feeding back the operation decision execution result by using the idea of reinforcement learning so as to optimize the next decision process. The invention not only saves a large amount of local computing power and storage space, improves the regulation efficiency, but also improves the reliability and quality of decision making, so that the automation degree of the operation regulation of the air compressor cluster is high, and the regulation decision making quality is high.

The following describes the implementation of one embodiment of the present invention in detail with reference to the explanation of specific steps S1-S4.

And S1, sending the acquired operation data of all the air compressors in the air compressor cluster to a cloud terminal.

Suppose that the air compressor cluster comprises N air compressor branches, wherein the ith branch (i is less than or equal to N) comprises(n_iNot less than 1) air compressors,for rounding up, the smallest integer greater than or equal to x is represented, perIn the air compressor with the branch line,the platform is a standby unit and can be used only when a common unit fails, and other n units_iThe platform is a common unit, and the common unit comprisesThe platform is frequency conversion air compressor machine, can satisfy the variable frequency speed governing demand, and others are ordinary air compressor machine.

The operation data of each air compressor of each branch is collected in real time, and the operation data comprises various operation parameters of the power consumption, the air inflow, the air inlet pressure, the air outlet pressure, the motor rotating speed, the compressor exhaust temperature, the host oil temperature, the circulating water main pipe pressure and temperature and the like of each air compressor. When the operating data of the air compressor is collected, the data are directly collected and obtained by an intelligent sensing unit on the air compressor side, after the data are collected and obtained, preprocessing work needs to be carried out on the data, and then the preprocessed data are sent to a cloud end. When data are collected, the intelligent sensing unit comprises various sensors, such as a pressure sensor for collecting pressure data of an inlet and an outlet, a temperature sensor for collecting temperature data and the like. When the data is preprocessed, the preprocessing mainly comprises data cleaning, data missing value filling, abnormal data processing and the like.

And S2, carrying out abnormity detection and data prediction processing on the air compressor operation data at the cloud end to obtain abnormal data and predicted values.

Since historical operating data of the air compressors have complex characteristics of large scale, diversity, dynamics, high dimensionality and the like, and the processing and storage of the data consume a large amount of computing power and storage space, the data of the operation of the air compressors are processed and stored at the cloud end.

1) And (6) cloud data processing. The cloud data processing refers to analyzing and processing the stored data and the newly received data, and the processing process mainly comprises the steps of carrying out anomaly detection on the data and predicting the workload of the air compressor.

And (4) detecting the abnormality. Comparing with a large amount of historical data of the air compressors, judging whether the air compressors are abnormal or not according to the current operation data of the air compressors, if the analysis result is abnormal, sending an overhaul signal, determining the abnormal or fault operation data and respectively storing the abnormal or fault operation data into the abnormal data or the fault data, and storing the normal data into the normal data. In particular, when it is determined whether or not there is an abnormality in the data, abnormality detection of the device may be performed by using a method such as multivariate statistical regression or an automatic encoder network. Taking multivariate statistical regression as an example, principal component analysis is carried out on data to reduce dimensionality, a covariance matrix is calculated by using training data only containing normal operating conditions, an ellipsoid which can represent the set probability distribution most is estimated through the covariance matrix, then the mahalanobis distance between a sample point and the data centroid is calculated, and if the distance is higher than a certain threshold value, the test point is classified as abnormal. The automatic encoder network is similar to the automatic encoder network, the network reconstruction loss after the data are reconstructed through network compression is calculated, the network reconstruction loss is compared with a set threshold value, and the operation abnormity is considered to exist when the network reconstruction loss exceeds the threshold value.

And predicting the working capacity of the air compressor. The method mainly predicts the operation parameters of each air compressor at the next moment and the medium-long term operation trend of the air compressor cluster, and takes the prediction as one of the bases for the variable-frequency speed-regulating operation regulation of the air compressors. When the workload of the air compressor at the next moment is predicted, the workload can be predicted by using a multivariate linear model, a multivariate nonlinear model, a neural network model and the like, specifically, the workload of the air compressor at the current moment, the historical workload of the same period, the workload of other air compressors at the current moment, the current operation of the air compressor and the deviation between the historical predicted value and the actual value are used as input in the prediction process, and the predicted workload value of the air compressor at the next moment is used as final output.

2) And storing the cloud data. The received data sent by the air compressor is stored by combining with cloud data processing classification results, the data are stored according to three types of normal data, abnormal data and fault data during storage, each type of data comprises various operation parameters related to the acquired data, the various types of data are grouped according to air compressor branches, and the data are arranged according to time periods in the groups.

And S3, performing fault early warning and executing operation decision on the side based on the abnormal data and the predicted value.

The real-time requirement on the operation decision in the operation process of the air compressor is high, the decision process is placed at the side end, the decision result can be better supported by a system to be executed, and the recording and the feedback of the decision result are facilitated, wherein the decision process comprises two parts, and on one hand, the fault level is judged according to the early warning signal; on the other hand, based on the fault level, decision suggestions are given to equipment regulation and control by combining the relation between the current working capacity of the air compressor and the predicted working capacity value of the air compressor. The specific process is as follows:

1) the relationship between fault classification and alarm is as follows:

no alarm signal: abnormal data indicates that no abnormal equipment exists and all equipment normally operates;

three-level alarm signals: at this time, the parameter calculated by the abnormality detection satisfies a < x ≦ b. The method comprises the following steps that (1) operation abnormity of one or more air compressors is shown, and the air compressors can still operate due to limited working capacity at the moment;

secondary alarm signal: at this time, the parameter calculated by abnormality detection satisfies x > b. The method comprises the following steps that operation faults of a certain air compressor are shown, and at the moment, the air compressor can hardly operate or can not operate;

primary alarm signal: at the moment, the calculated parameters for the abnormity detection of the plurality of air compressors all meet x > b. And indicating that a plurality of air compressors in the circuit have operation faults, wherein at the moment, the air compressors can hardly operate or can not operate.

Wherein x represents a parameter calculated by abnormality detection; and a and b are judgment threshold values when abnormal data are detected, and the values of a and b can be preset according to the actual production running condition of the air compressor.

2) And (5) making operation decisions of air compressor equipment.

2.1) firstly making a response according to the alarm level, and if no operation alarm exists, directly giving a decision suggestion; if the three-level alarm is received, when the decision-making suggestion is to shut down a certain air compressor, the abnormal air compressor is shut down preferentially, and equipment maintenance is carried out after the shutdown; if the secondary alarm is received, immediately shutting down the fault air compressor, starting a standby unit and carrying out equipment maintenance; and if the primary alarm is received, immediately closing all the fault air compressors, starting the standby units with the corresponding number, and carrying out equipment maintenance. When abnormal equipment is overhauled, the method mainly comprises the steps of receiving a unit abnormal signal sent in an abnormal detection stage of cloud data processing, quickly locating a relevant unit, carrying out equipment inspection, locating a fault type according to operation data, grading the fault, sending a fault unit number, a fault level and a fault type by a system, and waiting for manual overhaul; and if the system can not locate the fault type, classifying the fault only according to the operation data, and sending the fault unit number and the fault level by the system to wait for manual maintenance.

2.2) then, based on the corresponding relation between the current value of the workload of the air compressor and the predicted value of the workload of the air compressor, giving a decision suggestion for the regulation and control of the equipment, which is as follows:

with X₁Indicates the current working capacity, X, of the air compressor₂Represents the predicted value of the working capacity of the air compressor, | X₁|-|X₂I represents the difference value between the current value and the predicted value, a threshold value m is set according to the type and the operation condition of the air compressor, wherein, X represents the modulus of a vector X,the specific regulation and control process is as follows:

if 0 < | X₁|-|X₂If the | is less than or equal to m, suggesting a variable frequency air compressor to carry out pre-load shedding operation;

if | X₁|-|X₂If the | is greater than m, the air compressor is recommended to be shut down in advance. Specifically, the priority of pre-shutdown of the air compressor is respectively an abnormal operation unit, a standby unit and a common unit from high to low;

if-m < | X₁|-|X₂If the | is less than or equal to 0, suggesting a variable frequency air compressor to perform preloading operation;

if | X₁|-|X₂If the | is less than or equal to-m, the air compressor is recommended to be started in advance. Specifically, when the air compressor is pre-started, the priority of the common unit is greater than that of the standby unit.

Finally, the early warning is integratedAnd the reaction and equipment regulation and control suggestions based on the predicted values give equipment operation decisions. For example, all the current line-all the commonly used units are running fully, and the equipment n in the line-one₁When the operation fault occurs, the current working capacity X of the air compressor is₁And the predicted value X of the workload of the air compressor₂Satisfy 0 < | X₁|-|X₂If | is less than or equal to m, the final decision result is' turning off the line equipment n1, and turning on the line equipment n₁+1, pre-offloading the line-device 1 ".

And S4, recording and feeding back an operation decision execution result based on the idea of reinforcement learning so as to optimize the next operation decision of the air compressor.

And optimizing the next operation decision suggestion continuously according to the system operation result based on the idea of reinforcement learning. Specifically, the execution result of the operation decision given by the system each time is recorded, and the execution result of the operation decision corresponding to the operation decision at each moment is recorded. And if the execution result of the operation decision given by the system accords with the actual operation condition of the system at the next moment, packaging the operation decision and the decision instruction of the system at the moment into a reward group as a reference for decision later. If the execution result of the operation decision given by the system is contrary to the actual operation condition of the system at the next moment, the operation decision, the system decision instruction and the deviation value of the system instruction and the actual operation at the moment are packaged into a penalty group, when the system is similar to the current moment again, the operation decision of the system considers the historical instruction packaged by the penalty group, and the operation decision of a new round is adjusted towards the deviation value on the basis of the original result.

Therefore, the whole process of the air compressor cluster operation adjusting method based on cloud-edge coordination is completed.

Example 2:

in a second aspect, the present invention further provides a cloud-edge coordination-based air compressor cluster operation regulation system, referring to fig. 2, where the system includes:

the data sending module is used for sending the acquired running data of all the air compressors in the air compressor cluster to the cloud;

the anomaly detection and prediction module is used for performing anomaly detection and data prediction processing on the air compressor operation data at the cloud end to obtain anomaly data and a predicted value;

the early warning and operation decision module is used for carrying out fault early warning and executing operation decision on the side based on the abnormal data and the predicted value;

and the operation result recording and feedback module is used for recording and feeding back the operation decision execution result based on the idea of reinforcement learning so as to optimize the next operation decision of the air compressor.

Optionally, the system further includes: and the data acquisition and preprocessing module is used for acquiring the operation data of all air compressors in the air compressor cluster and preprocessing the operation data of the air compressors.

Optionally, the anomaly detection and prediction module performs anomaly detection and data prediction processing on the air compressor running data at a cloud end, and includes: judging whether the operation data of the air compressor is abnormal or not by using a multivariate statistical regression method; and carrying out data prediction processing on the air compressor operation data by utilizing a multivariate linear model.

Optionally, the performing, by the early warning and operation decision module, a fault early warning and an operation decision at a side based on the abnormal data and the predicted value includes:

s31, performing fault early warning on the side based on the abnormal data; the corresponding relation between the abnormal data and the fault early warning is as follows:

if no abnormal data exists, no alarm signal exists;

if the calculated parameter of the abnormal detection satisfies a < x < b, a three-level alarm signal is sent out;

if the calculated parameter of the abnormal detection meets x > b, a secondary alarm signal is sent out;

if the parameters calculated by the abnormality detection of the plurality of air compressors all meet x > b, a primary alarm signal is sent out;

wherein x represents a parameter calculated by abnormality detection; a and b are judgment threshold values when whether abnormal data exist or not is detected;

s32, executing operation decision at the side based on the predicted value; the corresponding relation between the predicted value and the operation decision is as follows:

if 0 < | X₁|-|X₂If the | is less than or equal to m, the variable frequency air compressor carries out pre-load shedding operation;

if | X₁|-|X₂If the | is greater than m, pre-shutting down a certain air compressor;

if-m < | X₁|-|X₂If the | is less than or equal to 0, performing preloading operation on the variable frequency air compressor;

if | X₁|-|X₂If the | is less than or equal to-m, a certain air compressor is started in advance;

wherein, X₁Representing the current working capacity of the air compressor; x₂Expressing the predicted value of the workload of the air compressor; i X₁|-|X₂L represents the difference between the current value and the predicted value; m represents a threshold value; | X | represents the modulus of the vector X,

optionally, the recording and feeding back the operation decision execution result by the operation result recording and feeding back module based on the idea of reinforcement learning to optimize the next operation decision of the air compressor includes:

s41, recording the operation decision and the operation decision execution result given by the system each time;

s42, if the execution result of the operation decision conforms to the actual operation condition of the system at the next moment, encapsulating the system operation decision and the decision instruction at the moment into a reward group; if the execution result of the operation decision given by the system is contrary to the actual operation condition of the system at the next moment, encapsulating the operation decision, the decision instruction and the deviation value of the instruction and the actual operation of the system at the moment into a penalty group;

and S43, optimizing the next operation decision of the air compressor based on the reward group and the penalty group.

It can be understood that, the air compressor cluster operation adjusting system based on cloud-edge coordination provided by the embodiment of the present invention corresponds to the air compressor cluster operation adjusting method based on cloud-edge coordination, and the explanation, examples, beneficial effects, and other parts of the relevant contents may refer to the corresponding contents in the air compressor cluster operation adjusting method based on cloud-edge coordination, and are not described herein again.

In summary, compared with the prior art, the method has the following beneficial effects:

1. the method comprises the steps of sending all acquired air compressor operation data in an air compressor cluster to a cloud end; and carrying out anomaly detection and data prediction processing on the operation data of the air compressor at the cloud end, then carrying out fault early warning and operation decision execution based on the anomaly detection data and the predicted value, and finally recording and feeding back the operation decision execution result by utilizing the idea of reinforcement learning so as to optimize the next decision process. The invention not only saves a large amount of local computing power and storage space, improves the regulation efficiency, but also improves the reliability and quality of decision making, so that the automation degree of the operation regulation of the air compressor cluster is high, and the regulation decision making quality is high.

2. According to the method, a cloud edge cooperation technology is utilized, so that a large amount of local computing power can be saved, data abnormity detection and prediction and data storage operation are all placed at the cloud end, the local storage space is saved, and the operation and adjustment efficiency of the air compressor is improved;

3. the method is based on the decision obtained by the predicted value and the historical result, and has higher reliability; meanwhile, the operation result is continuously recorded and fed back based on the reinforcement learning method, correct decision is reinforced, wrong decision is weakened, and decision quality of the system can be continuously improved.

It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, 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 a process, method, article, or apparatus that comprises the element.

The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.