阅读说明：本技术 一种基于缺陷模式的变电主设备诊断方法和系统 (Fault mode-based substation main equipment diagnosis method and system ) 是由 杨兆静 王海龙 陈操 崔玮 谢茜 于 2021-08-30 设计创作，主要内容包括：本发明涉及一种基于缺陷模式的变电主设备诊断方法和系统。该方法采用基于缺陷模式的设备缺陷诊断方法,通过明确变电主设备的缺陷模式,定义基于检测项目的缺陷判据,定义分析矩阵和缺陷概率综合算法,对缺陷进行定性分析,有效实现变电主设备的缺陷诊断；并通过计算状态量的严重程度,对缺陷进行定量分析,同时也为开展状态评价和风险评估提供数据基础。(The invention relates to a method and a system for diagnosing power transformation main equipment based on a defect mode. The method adopts a defect mode-based equipment defect diagnosis method, defines a defect criterion based on a detection project by defining the defect mode of the main power transformation equipment, defines an analysis matrix and a defect probability comprehensive algorithm, carries out qualitative analysis on the defects, and effectively realizes the defect diagnosis of the main power transformation equipment; and the defects are quantitatively analyzed by calculating the severity of the state quantity, and meanwhile, a data basis is provided for developing state evaluation and risk assessment.)

1. A method for diagnosing a substation main device based on a defect mode is characterized by comprising the following steps:

s1, constructing a defect diagnosis model:

s11, defining a defect mode, and defining typical defect types of the substation main equipment;

s12, defining detection items, and carrying out category grouping on all state information under the existing conditions of the power transformation main equipment according to the source and the incidence relation with the defect mode to form the detection items for judging whether the detection is qualified;

s13, defining an analysis matrix, determining the corresponding relation between the detection items and the defect mode, and assigning each corresponding relation to a basic defect probability;

s14, defining a defect probability comprehensive algorithm for comprehensively calculating the defect probability of a plurality of unqualified detection items related to one defect;

s2 model application:

s21, judging detection items, namely inputting full-dimensional data into the detection items, judging whether the input data are abnormal or not through the detection items, and screening out unqualified detection items;

s22, matching the defect modes to obtain the defect mode corresponding to each unqualified detection item and the corresponding defect probability;

s23. defect probability synthesis and severity analysis, including the following parallel steps:

s231, integrating the defect probability, and comprehensively calculating the defect mode and the basic defect probability of the unqualified detection item corresponding to the defect mode;

s232, analyzing and calculating the severity, namely a quantitative index used for representing the defect/defect development degree in the model;

and S24, outputting a diagnosis result based on the calculation result of the step S23.

2. The method for diagnosing a substation main equipment based on the defect mode according to claim 1, wherein the step S13 specifically includes: the analysis matrix is a D multiplied by M matrix, wherein D is a defect set, M is a detection item set, a defect probability exists in a defect mode reflected by each detection item, the defect probability value is a decimal between 0 and 1, and the numerical value of the defect probability value is a statistical result of empirical data and/or an empirical value provided by an expert.

3. The method for diagnosing a substation main equipment based on the defect mode according to claim 1, wherein the step S21 specifically includes: inputting full-dimensional data into detection items to form three types of detection item sets:

1) the detection item set without parameters, and the data has no access model or detection means;

2) a set of qualified test items with parameters;

3) a set of disqualified test items with parameters.

4. The method for diagnosing a substation main equipment according to claim 3, wherein the step S22 specifically includes:

the detection items judged to be parameter-free detection items and qualified detection items with parameters in the rejection detection item judgment stage cannot participate in defect diagnosis of equipment, and invalid data dimensionality is actively reduced;

and performing association retrieval on the unqualified detection items with the parameter judgment result in the detection item judgment stage by combining the analysis matrix to obtain a defect mode corresponding to each unqualified detection item and a corresponding defect probability.

5. The method for diagnosing a substation main equipment based on the defect mode according to claim 4, wherein the step S231 specifically includes: traversing the analysis result of defect mode matching, carrying out comprehensive calculation on the defect mode and the basic defect probability of a plurality of unqualified detection items corresponding to a certain defect mode, and outputting the unique defect probability of each defect mode and the integral defect probability of the equipment; the step S232 specifically includes: and mapping the ratio to the [0, 1] interval according to the comparison between the test measurement value and the standard value in the same item or related procedures in the handover test table.

6. A system for diagnosing a substation master based on a fault mode, the system comprising:

the model building module is used for building a defect model;

the model application module is used for carrying out defect diagnosis;

the model building module specifically comprises:

the defect mode defining unit is used for defining a defect mode and defining a typical defect type of the power transformation main equipment;

the detection item definition unit is used for defining detection items, and carrying out category grouping on all state information under the existing conditions of the power transformation main equipment according to the source of the state information and the incidence relation between the state information and the defect mode to form a detection item for judging whether the detection is qualified or not;

the analysis matrix definition unit is used for defining an analysis matrix, determining the corresponding relation between the detection items and the defect mode and assigning each corresponding relation to a basic defect probability;

the defect probability comprehensive algorithm definition unit is used for defining a defect probability comprehensive algorithm and comprehensively calculating the credibility of a plurality of unqualified detection items related to one defect;

the model application module specifically comprises:

the detection item judgment unit is used for inputting the full-dimensional data into a detection item, judging whether the input data is abnormal or not through the detection item, and screening out unqualified detection items;

the defect mode matching unit is used for obtaining a defect mode corresponding to each unqualified detection item and corresponding defect probability;

the defect probability synthesis and severity analysis unit comprises: the defect probability comprehensive subunit is used for comprehensively calculating the defect mode and the basic defect probability of the unqualified detection item corresponding to the defect mode; the severity analysis subunit is used for analyzing and calculating the severity, namely a quantitative index used for representing the defect/defect development degree in the model;

and the diagnosis result output unit is used for outputting the calculation result based on the defect probability integration and severity analysis unit.

7. The system according to claim 6, wherein the analysis matrix definition unit is specifically configured to: and defining an analysis matrix as a D multiplied by M matrix, wherein D is a defect set, M is a detection item set, a defect probability exists in a defect mode reflected by each detection item, the defect probability value is a decimal between 0 and 1, and the numerical value is a statistical result of empirical data and/or an empirical value provided by an expert.

8. The system according to claim 6, wherein the detection item determination unit is specifically configured to: inputting full-dimensional data into detection items to form three types of detection item sets:

1) the detection item set without parameters, and the data has no access model or detection means;

2) a set of qualified test items with parameters;

3) a set of disqualified test items with parameters.

9. The defect diagnostic system of claim 8, wherein the defect pattern matching unit is specifically configured to:

the detection items judged to be parameter-free detection items and qualified detection items with parameters in the rejection detection item judgment stage cannot participate in the analysis of equipment defects, and invalid data dimensions are actively reduced;

and performing association retrieval on the unqualified detection items with the parameter judgment result in the detection item judgment stage by combining the analysis matrix to obtain a defect mode corresponding to each unqualified detection item and a corresponding defect probability.

10. The system according to claim 9, wherein the defect probability integration subunit is specifically configured to: traversing the analysis result of defect mode matching, carrying out comprehensive calculation on the defect mode and the basic defect probability of a plurality of unqualified detection items corresponding to a certain defect mode, and outputting the unique defect probability of each defect mode and the integral defect probability of the equipment; the severity analysis subunit is specifically configured to: and mapping the ratio to the [0, 1] interval according to the comparison between the test measurement value and the standard value in the same item or related procedures in the handover test table.

Technical Field

The invention relates to the technical field of equipment diagnosis, in particular to the technical field of power transformation main equipment diagnosis based on a defect mode.

Background

The defect diagnosis is the precondition of equipment maintenance and is the guarantee of stable operation of electrical equipment. The basis for defect diagnosis is derived from the state detection of the device. The state detection can be obtained through inspection information, operation maintenance information, conventional electrical tests, infrared temperature measurement, protection device action information and application of various online monitoring devices. The current defect diagnosis methods are mainly classified into the following 3 categories.

1) Conventional defect diagnosis method

The conventional defect diagnosis method mainly comprises a post diagnosis method and a pre diagnosis method. The post-diagnosis is performed when a defect occurs. Defects are directly detected according to various physical and chemical means. For example: and when the GIS equipment has a defect, the appearance of the equipment is checked on site, the action report of the smart protection device is analyzed, the gas is analyzed, and the like. The method is quick, simple and effective.

The prior diagnosis is performed before the occurrence of the defect. And analyzing and comparing the data of the electrical equipment in normal operation to find out the defect development trend. The method is based on historical data, compares the equipment parameters, maintenance and operation data provided by manufacturers, analyzes the development tendency of the equipment state and obtains a diagnosis result, and the method is based on the defects and repeatedly summarizes rules.

2) Intelligent defect diagnosis method

With the maturity of artificial intelligence technology, various intelligent defect diagnosis systems have been widely applied in power systems, such as: the inspection of a transformer substation robot, an online monitoring system of transformer oil chromatography, an online monitoring device of leakage current of a lightning arrester, a deformation monitoring system of a transformer winding and the like become main diagnosis means of defects of transformer substation equipment.

3) Mathematical method defect diagnosis method

Mathematical models of equipment defects are established by mathematical means such as mathematical statistics, neural networks, various algorithms and the like, various defect information is screened, defect rules are summarized by methods such as deduction and the like, defect symbols are reversely deduced, and then verification is carried out on the defect symbols and on-site actual data to form an effective defect diagnosis method.

The power transformation main equipment comprises electric equipment such as a transformer, a reactor, a combined electrical apparatus, a current transformer, a voltage transformer, a lightning arrester, an isolating switch, a circuit breaker and the like. The operation state of the main power transformation equipment plays a decisive role in the safety and stability of the operation of the transformer substation, and the defect diagnosis of the main power transformation equipment is a subject of constant exploration in the technical field.

At present, the state perception capability, data dimensionality and historical data integrity of the main power transformation equipment have obvious difference, which is mainly shown in that 1, partial equipment lacks effective technical means such as online monitoring and live detection, and cannot be substantially improved in a short period. Part of the available dimensional data is missing during state analysis. 2. The method is influenced by various factors, the integrity of historical state data of part of equipment is poor, and state data for longitudinal comparison is lacked during state analysis. 3. The online monitoring data has good timeliness but poor data quality, and the offline detection data has poor timeliness but good data quality. There are differences in data availability across dimensions.

Disclosure of Invention

In order to overcome the defects of the prior art, the invention provides a method and a system for diagnosing the power transformation main equipment based on the defect mode.

The invention adopts the following technical scheme:

a method for diagnosing a substation main device based on a defect mode comprises the following steps:

s1, constructing a defect diagnosis model:

s11, defining a defect mode, and defining typical defect types of the power transformation main equipment;

s12, defining detection items, and carrying out category grouping on all state information under the existing condition of the power transformation main equipment according to the source of the state information and the incidence relation between the state information and the defect mode to form a detection item for judging whether the detection is qualified or not;

s13, defining an analysis matrix, determining the corresponding relation between the detection items and the defect mode, and assigning each corresponding relation to a basic defect probability;

s14, defining a defect probability comprehensive algorithm for comprehensively calculating the defect probability of a plurality of unqualified detection items related to one defect;

s2, model application:

s21, judging detection items, namely inputting full-dimensional data into the detection items, judging whether the input data is abnormal or not through the detection items, and screening out unqualified detection items;

s22, matching the defect modes to obtain a defect mode corresponding to each unqualified detection item and corresponding defect probability;

s23, defect probability synthesis and severity analysis, which comprises the following parallel steps:

s231, defect probability synthesis, namely performing comprehensive calculation on the defect mode and the basic defect probability of the unqualified detection item corresponding to the defect mode;

s232, analyzing and calculating the severity, namely a quantitative index used for representing the defect/defect development degree in the model;

and S24, outputting a diagnosis result, and outputting recommended measures based on the calculation result of the step S23.

Further, the step S13 specifically includes: the analysis matrix is a D multiplied by M matrix, wherein D is a defect set, M is a detection item set, a defect probability exists in a defect mode reflected by each detection item, the defect probability value is a decimal between 0 and 1, and the numerical value of the defect probability value is a statistical result of empirical data and/or an empirical value provided by an expert.

Further, the step S21 specifically includes: inputting full-dimensional data into detection items to form three types of detection item sets:

1) the detection item set without parameters, and the data has no access model or detection means;

2) a set of qualified test items with parameters;

3) a set of disqualified test items with parameters.

Further, the step S22 specifically includes:

the detection items judged to be parameter-free detection items and qualified detection items with parameters in the rejection detection item judgment stage cannot participate in defect diagnosis of equipment, and invalid data dimensionality is actively reduced;

and performing association retrieval on the unqualified detection items with the parameter judgment result in the detection item judgment stage by combining the analysis matrix to obtain a defect mode corresponding to each unqualified detection item and a corresponding defect probability.

Further, the step S231 specifically includes: traversing the analysis result of defect mode matching, carrying out comprehensive calculation on the defect mode and the basic defect probability of a plurality of unqualified detection items corresponding to a certain defect mode, and outputting the unique defect probability of each defect mode and the integral defect probability of the equipment; the step S232 specifically includes: and mapping the ratio to the [0, 1] interval according to the comparison between the test measurement value and the standard value in the same item or related procedures in the handover test table.

The invention also includes a system for diagnosing a substation master based on a fault mode, the system comprising:

the model building module is used for building a defect model;

the model application module is used for carrying out defect diagnosis;

the model building module specifically comprises:

the defect mode defining unit is used for defining a defect mode and defining a typical defect type of the power transformation main equipment;

the detection item definition unit is used for defining detection items, and carrying out category grouping on all state information under the existing conditions of the power transformation main equipment according to the source of the state information and the incidence relation between the state information and the defect mode to form a detection item for judging whether the detection is qualified or not;

the analysis matrix definition unit is used for defining an analysis matrix, determining the corresponding relation between the detection items and the defect mode and assigning each corresponding relation to a basic defect probability;

the defect probability comprehensive algorithm definition unit is used for defining a defect probability comprehensive algorithm and comprehensively calculating the credibility of a plurality of unqualified detection items related to one defect;

the model application module specifically comprises:

the detection item judgment unit is used for inputting the full-dimensional data into a detection item, judging whether the input data is abnormal or not through the detection item, and screening out unqualified detection items;

the defect mode matching unit is used for obtaining a defect mode corresponding to each unqualified detection item and corresponding defect probability;

the defect probability synthesis and severity analysis unit comprises: the defect probability comprehensive subunit is used for comprehensively calculating the defect mode and the basic defect probability of the unqualified detection item corresponding to the defect mode; the severity analysis subunit is used for analyzing and calculating the severity, namely a quantitative index used for representing the defect/defect development degree in the model;

and the diagnosis result output unit outputs recommended measures based on the calculation results of the defect probability integration and severity analysis unit.

Further, the analysis matrix definition unit is specifically configured to: and defining an analysis matrix as a D multiplied by M matrix, wherein D is a defect set, M is a detection item set, a defect probability exists in a defect mode reflected by each detection item, the defect probability value is a decimal between 0 and 1, and the numerical value is a statistical result of empirical data and/or an empirical value provided by an expert.

Further, the detection item determination unit is specifically configured to: inputting full-dimensional data into detection items to form three types of detection item sets:

1) the detection item set without parameters, and the data has no access model or detection means;

2) a set of qualified test items with parameters;

3) a set of disqualified test items with parameters.

Further, the defect pattern matching unit is specifically configured to:

the detection items judged to be parameter-free detection items and qualified detection items with parameters in the rejection detection item judgment stage cannot participate in the analysis of equipment defects, and invalid data dimensions are actively reduced;

and performing association retrieval on the unqualified detection items with the parameter judgment result in the detection item judgment stage by combining the analysis matrix to obtain a defect mode corresponding to each unqualified detection item and a corresponding defect probability.

Further, the defect probability integration subunit is specifically configured to: traversing the analysis result of defect mode matching, carrying out comprehensive calculation on the defect mode and the basic defect probability of a plurality of unqualified detection items corresponding to a certain defect mode, and outputting the unique defect probability of each defect mode and the integral defect probability of the equipment; the severity analysis subunit is specifically configured to: and mapping the ratio to the [0, 1] interval according to the comparison between the test measurement value and the standard value in the same item or related procedures in the handover test table.

The invention achieves the following beneficial effects: influence factors of data quality and data type difference are considered in the corresponding relation between the detection items and the defect modes, so that the influence caused by the data quality and the data type difference is solved.

The defect diagnosis model can diagnose the defect type and the defect component, and can provide a mathematical analysis basis for the defect development and evolution process, other potential hidden dangers possibly existing in the power transformation main equipment and the prediction of the defects possibly occurring in the future; the defect diagnosis model can also calculate the severity of the state quantity corresponding to the defect mode, and provides a data basis for equipment state evaluation.

Drawings

Fig. 1 is a schematic structural diagram of a defect diagnosis model according to the present invention.

FIG. 2 is a schematic diagram of an analysis matrix according to the present invention.

FIG. 3 is a schematic diagram of the diagnostic analysis process of the present invention.

Detailed Description

The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby. It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.

The invention simulates the intelligent reasoning process idea of applying the symptom phenomenon/data to the defect diagnosis of the main power transformation equipment by human experts. Example 1 of the present invention comprises the following method steps:

s1, constructing a defect model, comprising the following steps:

s11, defining a defect mode: and defining typical defect/defect type of the transformer master equipment, namely a defect mode.

To accurately define the defect mode, the transformer main equipment needs to be further divided into various sub-equipment/subsystems. The division of the subsystems can adopt two methods: functional methods and component methods. The functional method is to divide the main equipment of the transformer according to the function of the component, such as SF₆The circuit breaker is divided into an insulation system, a breaking system, a current carrying system, an SF₆Gas, mechanical systems, etc. The component rule divides the substation main equipment according to the physical position of the component.

The defect mode of the sub-device/subsystem includes a functional failure mode of the sub-device/subsystem and a defect mode, wherein the failure refers to the loss of function of the component, the defect is a component abnormality but not necessarily loss of function, wherein the functional failure mode may cause the loss of function of the component, and the defect mode is an abnormality of the component but not necessarily causes the loss of function, and the component can only be operated for a certain time.

Still at SF₆The circuit breaker is taken as an example, and the defect mode is shown in table 1 below.

TABLE 1 Defect mode of SF6 circuit breaker

Each defect pattern may be further divided into a plurality of sub-defect patterns. The sub-defect modes are shown in table 2 below, taking a transformer as an example.

TABLE 2 partial Defect mode of transformers

S12, defining detection items: all state information under the existing condition of the power transformation main equipment is classified and grouped according to the source and the incidence relation with the defect mode, and the grouped data is logically and mathematically combined by combining threshold values, trends, correlation regulations and expert experiences to form the exclusive criterion of the minimum unit, namely a detection item for judging whether the detection is qualified.

Taking a transformer as an example, the test items are shown in table 3 below.

TABLE 3 partial inspection items of transformers

S13, defining an analysis matrix: determining the corresponding relation between the detection items and the defect mode according to typical defect cases and experience, endowing each corresponding relation with a basic defect probability according to different capabilities of the detection items for representing the defect mode, and constructing a top-level analysis matrix with many-to-many detection items and defect modes by taking the basic defect probability as a main line. The analysis matrix is a D × M matrix, where D is the set of defects and M is the set of defect symptoms or the set of inspection items. According to typical defect cases and experiences, the corresponding relations between the detection items and the defect modes are determined, and the capabilities of characterizing the defect modes according to the detection items are different, so that the situation that positive correlation and negative correlation exist between defects and symptoms in the analysis matrix is analyzed, each corresponding relation is given a basic defect probability, and the basic defect probability is used as a main line to construct a top-level analysis matrix with many pairs of detection items and defect modes, as shown in fig. 2. In the diagnosis model, a confidence coefficient or defect probability exists in a defect mode reflected by each detection item, the defect probability value is a decimal between 0 and 1, and the numerical value is a statistical result of empirical data and an empirical value provided by an expert. Taking a transformer as an example, the analysis matrix is shown in the following table 4

TABLE 4 partial analysis matrix of transformers

S14, defining a defect probability comprehensive algorithm: and defining a defect probability comprehensive algorithm, and reasonably changing the output defect probability when the directivity criterion of a certain defect mode is increased. In the defect diagnosis process, if a plurality of unqualified detection items correspond to a certain defect, the credibility of the evidences needs to be integrated. In order to solve the comprehensive problem of confidence, the following method is adopted:

in the confidence method, the diagnosis knowledge is expressed by a production rule for an arbitrary inference rule k_ijThe standard form is as follows: IF m_j THEN d_i WITH CF(d_i|m_j)

Wherein m is_jRepresents evidence, d_iIndicating a defect conclusion, confidence CF (d)_i|m_j) Is shown in evidence m_jAssuming defect d in the presence_iTo the extent of confidence.

From conditional probability and prior probability, CF (d)_i|m_j) Is defined as:

when a diagnosis model is actually established, the CF assignment method is flexible, and the statistical result of empirical data can be adopted, and the empirical value provided by experts can be used for replacing the CF assignment method.

The belief function CF should satisfy the following basic properties:

(1).-1≤CF(d_i|m_j)≤+1，CF<0 represents the degree of distrust for the proposition.

(2) If d₁,d₂,L,d_mAre mutually independent defect hypotheses, m_jFor evidence of their benefits, then

(3) If m₂For unknown evidence, then CF (d)_i|m₁∧m₂)＝CF(d_i|m₁)。

In the defect diagnosis process, if a plurality of evidences exist to support or deny D ═ D₁,d₂,L,d_mA certain defect d in_iIt is desirable to integrate the trustworthiness of these proofs. Suppose there are two rules

IF m₁ THEN d_i WITH CF(d_i|m₁)

IF m₂ THEN d_i WITH CF(d_i|m₂)

The associated composite confidence is defined as

Wherein, CF (d)_i|m₁∧m₂) Representing a comprehensive credibility value, namely the defect probability after the comprehensive; and X and Y are defect probabilities corresponding to the criterion pointing to the defect mode.

The following aspects are mainly considered for the operation experience of the power transformation equipment: whether the equipment is subjected to outlet short circuits, overvoltage surges, long-term overloads, and lengthy downtime, etc. The possible effects of these operational experiences on the equipment can be introduced into the diagnostic reasoning process in the form of weighting the confidence in the defects.

The reliability correction formula is

CF^*＝(1+λ)CF

Wherein λ is artificially set according to the operation experience of the power transformation equipment.

And establishing a defect diagnosis model by defining a defect mode, defining defect criteria, defining an analysis matrix and defining a defect comprehensive algorithm.

The fault diagnosis model of the substation main equipment is established on the basis of the early-stage basic research. And (4) determining the characteristic state quantity set, the typical defect mode set and the incidence relation of each item in the two sets to construct a main framework of the model. The main logic in the model includes three parts, defect identification, defect confidence analysis (qualitative) and defect severity analysis (quantitative). The input design for the model is: all characteristic state quantities which can be collected; the output design is as follows: possible defect modes of the device, confidence and severity of individual defects. The main logical structure of the defect diagnosis model of the present invention is shown in fig. 1.

And S2, applying the model. According to the established defect reasoning idea, the system divides the diagnosis and analysis process of the model into three stages: the first stage, detecting item judgment; in the second stage, defect mode matching is carried out; and in the third stage, integrating the defect probability, wherein the specific process is shown in the figure 3.

And S21, judging detection items, namely inputting the full-dimensional data into the detection items, rapidly analyzing the detection items and judging whether the input data is abnormal or not, and screening out unqualified detection items, namely equipment abnormal states, so that the data dimension can be reduced, and the defect diagnosis rate and precision can be improved. Each detection item is a judgment formula formed by applying a mathematical relationship or a logical relationship to one or more state quantities. And extracting and analyzing the input experimental data, online monitoring data, charged detection data, inspection data, operation condition data, load, environment and other data, and transmitting the data into corresponding detection project models to obtain the conclusion whether the detection projects are qualified or not. Three types of detection item sets are formed at this stage:

1) the data has no access model or detection means.

2) A collection of qualified test items with parameters.

3) A set of disqualified test items with parameters.

S22, defect pattern matching: when defect analysis is carried out, the system firstly rejects the detection items judged to be parameter-free and qualified detection items judged to be parameter-free in the detection item judgment stage, so that the detection items cannot participate in the analysis of the equipment defects, namely, invalid data dimensionality is actively reduced. And then, performing correlation retrieval on the unqualified detection items with the parameter judgment result in the detection item judgment stage by combining a top layer analysis model (matrix) to obtain a defect mode corresponding to each unqualified detection item and a corresponding defect probability.

S23, defect probability synthesis and severity analysis, which comprises the following parallel steps:

and S231, integrating the defect probability, and comprehensively calculating the defect mode and the basic defect probability of the unqualified detection item corresponding to the defect mode. Traversing the analysis result of the defect pattern matching, and if a plurality of unqualified detection items correspond to a certain defect, comprehensively calculating the probability of the basic defect of the defect pattern and the detection items. Finally realizing defect probability calculation unique to each defect mode and equipment overall defect probability output

And S232, analyzing and calculating the severity, namely a quantitative index used for representing the defect/defect development degree in the model. The severity value of the defect is a quantitative indicator used in the model to characterize the defect/degree of defect development. The parameters or parameter sets that may characterize the severity are different for each defect. Aiming at the characteristics, a differentiated severity analysis formula is formed for a typical defect mode set in the quantitative analysis process of the defect mode. And mapping the ratio to the [0, 1] interval according to the comparison between the test measurement value and the standard value in the same item or related procedures in the handover test table. The mapping rule is not general, and each defect is different from the mapping rule of the corresponding data.

In the model processing process, after the defect mode is determined and the severity value is output, the calculation of the severity is mainly applied to a defect prediction and state evaluation model.

And S24, outputting a diagnosis result, and outputting recommended measures based on the calculation result of the step S23.

The present invention in embodiment 2 will be described by taking the diagnosis of the discharge defect of the transformer as an example.

In this embodiment, the online monitoring and early warning system sends out early warning according to the abnormal growth trend of two characteristic gases in the online monitoring data to trigger defect diagnosis, and provides recommended measures to perform offline oil chromatographic analysis as soon as possible.

A defect pattern and detection items are first defined. For the transformer discharge defect of this example 2, the defect mode relates to an insulating oil defect, and the sub-defect mode is a partial discharge in oil.

And in the defect diagnosis process, firstly, judging detection items, then, carrying out defect mode matching, and eliminating non-parameter detection items and qualified detection items with parameters, thereby reducing data dimensionality and improving the operation speed of the diagnosis model.

Then, performing defect diagnosis according to the analysis data of the gas in the off-line oil to obtain that the defect result is a discharge defect CF (confidence coefficient) of 0.99; wherein high energy discharge CF is 0.4, low energy discharge CF is 0.6 and solid insulation CF is not involved is 0.8.

The detection items also comprise off-line oil gas analysis data, iron core and clamp grounding current data, high-frequency partial discharge detection data in live detection, ultrasonic signal detection data and ultrahigh frequency detection data input. And the off-line gas in oil analysis, ultrasonic signal detection and ultrahigh frequency detection are unqualified.

And inputting the defect types of the diagnosis results into a state evaluation and result output unit, giving recommended measures by the state evaluation and result output unit, carrying out tests according to the measures by operation and maintenance personnel, supplementing data to the defect diagnosis model, and diagnosing the defect diagnosis model again according to the newly supplemented data. The second defect diagnosis result is that the discharge defect CF (confidence coefficient) is 0.99; wherein, high energy discharge CF is 0.4, low energy discharge CF is 0.6, solid insulation CF is 0.8 and suspension discharge CF is 0.7. The third defect diagnosis and detection items comprise off-line oil gas analysis, partial discharge detection, direct current resistance detection, insulation resistance detection, high-frequency partial discharge detection, ultrasonic signal detection and ultrahigh frequency detection data input in winding voltage ratio detection and charged detection. The detection data of off-line gas analysis in oil, partial discharge detection, high-frequency partial discharge detection, ultrasonic signal detection and ultrahigh frequency detection are unqualified, and the defect diagnosis result is that the discharge defect CF (confidence coefficient) is 0.99; the high-energy discharge CF is 0.4, the low-energy discharge CF is 0.6, the solid insulation CF is not involved, the suspension discharge CF is 0.8, the defect position is the suspension discharge at the middle-phase pressure sleeve flange of the B phase, the corresponding defect mode is an insulating oil defect, and the sub-defect mode is partial discharge in oil.

For the severity analysis formula, currently, a set of corresponding severity differentiation analysis formula is formed for each type of typical defect set of power transformation main equipment (such as a transformer, a breaker, a disconnecting switch, a grounding disconnecting link, a lightning arrester, a mutual inductor and the like). Taking the device type as a main transformer, the part to which the device belongs as a current carrying system, and the defect mode as a winding short circuit as an example, a calculation formula 1 is as follows:

equation 1:

key parameters: a direct current leakage current;

the data source is as follows: dc leakage test-leakage current (uA) of the winding;

the calculation method comprises the following steps: x is the same item in the measurement value/handover test table;

calculating the formula: calculation by four-region linear algorithm

Zone 1: x <1.0 severity: 0;

zone 2: x is more than or equal to 1.0 and less than 1.5 severity: ((X-1.0)/0.5) × 0.5;

zone 3: x is more than or equal to 1.5 and less than 3.0 severity: ((X-1.5)/1.5) × 0.5+ 0.5;

zone 4: x severity of 3.0 ≦ X: 1

And inputting the severity analysis result into a state evaluation and result output unit, giving recommended measures by the state evaluation and result output unit, carrying out tests according to the measures by operation and maintenance personnel, supplementing data to the defect diagnosis model, and diagnosing the defect diagnosis model again according to the newly supplemented data. The cases were the same as for the defect analysis, with a first diagnosis of 0.75 severity; the severity of the second diagnosis result is 1; the severity of the third diagnosis was 1.

The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.