Fault diagnosis system for relay protection device

Abstract

The embodiment of the invention discloses a fault diagnosis system for a relay protection device, and belongs to the technical field of fault diagnosis, and the system comprises a data collection module which obtains the equipment data, the fault failure case type, the fault result and the failure mechanism of the relay protection device, and the state data of a target relay protection device; the model establishing module extracts fault features from the data acquired by the data acquisition module, and establishes a reliability index model according to the fault features; and the fault diagnosis module performs operation monitoring, fault identification diagnosis and reliability evaluation on the state data of the target relay protection device by using the reliability index model. According to the embodiment of the invention, abnormal identification and failure positioning of the relay protection device can be realized, and possible defects of the relay protection device can be pre-judged; a relay protection device state evaluation system is planned to be constructed, and assistance is provided for comprehensively improving the safety, reliability and intelligent operation and maintenance level of a relay protection system.

Description

Technical Field

[0001] The present application belongs to the technical field of fault diagnosis, and in particular, relates to a fault diagnosis system for a relay protection device. Background Art

[0002] Relay protection devices are the first line of defense to ensure the safe operation of the power grid and prevent catastrophic accidents. Their own health and reliability have a significant impact on the entire power system. It is vital to carry out condition maintenance on the relay protection device system. Relay protection devices are an important line of defense for power grid security and are key equipment for building a three-dimensional defense system for the safe and stable operation of large power grids. New situations and new scenarios in the power grid have put forward new requirements for the digitalization and intelligence of key technologies of relay protection devices.

[0003] The current relay protection operation and maintenance system faces severe challenges. First, the operating status information of the relay protection device inside the plant cannot be pre-processed in the first time, and it is impossible to identify effective fault data from the massive alarm signal data and filter out invalid interference data. It is impossible to reduce the data processing pressure of front-line personnel and improve the processing efficiency of alarm signals; secondly, with the large-scale commissioning of smart substations, the widespread application of unmanned substations and remote operation, how to ensure the safe and reliable operation of secondary equipment and solve the contradiction between the configuration of operation and maintenance personnel and the ever-expanding scale of the power grid has become a major and urgent problem facing the current power grid equipment operation and maintenance. Problems are not solved until they occur, resulting in the inability of the power grid to operate smoothly. Summary of the invention

[0004] In order to solve the above problems and technical defects, the embodiment of the present application adopts the following technical solution, a fault diagnosis system for a relay protection device, comprising:

[0005] Data acquisition module, used to obtain equipment data of historical relay protection devices, fault failure case categories, fault results, failure mechanisms and status data of relay protection devices to be monitored;

[0006] A model building module is used to extract fault features from the data acquired by the data acquisition module and to build a reliability index model based on the fault features;

[0007] The fault diagnosis module is used to use the reliability index model to perform operation monitoring, fault identification and diagnosis, and reliability evaluation on the status data of the target relay protection device.

[0008] Preferably, the model building module pre-processes the data collected by the data collection module, and performs data cleaning and removal of abnormal data through scale normalization processing.

[0009] Furthermore, the specific steps of extracting fault features from the data acquired by the data acquisition module are as follows:

[0010] Obtain the equipment models of various historical relay protection devices, obtain the internal component parameters of each model of relay protection device, and the historical status data of each operating situation in different scenarios;

[0011] Combined with the internal component parameters, the status features of various historical status data in operation under different scenarios are extracted, and the historical fault status data is obtained from the various historical status data according to the extracted status features;

[0012] The historical fault status data is combined with the fault failure case categories corresponding to the historical fault status data to perform feature extraction to obtain fault features.

[0013] Furthermore, the specific steps of establishing the reliability index model according to the fault characteristics are as follows:

[0014] Establish a fault classification library, which has multiple fault subsets. Each fault subset corresponds to a type of fault failure case. Store historical fault status data into the corresponding fault subset according to fault characteristics.

[0015] Training calculations are performed on each fault subset to obtain a fault training subset, which is then used to train a base classification fault sub-model. The base classification fault sub-model is trained for multiple rounds to generate multiple base classification fault sub-models. The multiple base classification fault sub-models are weighted and fused to generate a reliability index model.

[0016] Furthermore, the specific steps of the fault diagnosis module for performing fault identification and diagnosis are as follows:

[0017] First, a feature vector of each historical fault state data is generated according to the fault feature of each historical fault state data in the fault subset;

[0018] Extracting features of target state data of the target relay protection device, and generating a target feature vector according to the extracted features;

[0019] Calculate the similarity between the target feature vector and the feature vector of each historical fault state data to obtain the similarity vector between the target state data and each fault subset;

[0020] The maximum similarity vector is extracted from each similarity vector, and the fault failure case category corresponding to the fault subset to which the maximum similarity vector belongs is used as the fault failure case category of the target state data.

[0021] Furthermore, after obtaining the fault failure case category, the target state data is updated to the corresponding fault subset according to the fault failure case category, and the reliability indicator model is updated every time a preset period is reached.

[0022] Furthermore, the fault diagnosis module is provided with an evaluation diagnosis module, and the evaluation diagnosis module is provided with an evaluation diagnosis model. After obtaining the category of the fault failure case, the evaluation diagnosis model performs corresponding fault evaluation to obtain a reliability evaluation result.

[0023] Furthermore, the evaluation and diagnosis model establishment process is as follows:

[0024] Obtain historical fault evaluation texts, extract feature words from each fault evaluation text, and calculate feature word weights;

[0025] Roughly classify each fault evaluation text, and calculate the similarity between fault evaluation texts based on the rough classification results and feature word weights;

[0026] According to the similarity calculation results, the historical fault evaluation texts are classified and processed, and the evaluation diagnosis model is obtained according to the classification processing results.

[0027] Furthermore, the calculation of feature word weights is to first screen and calculate the weights of key feature words in the fault evaluation text, then calculate the weights of the remaining feature words, calculate the final weight of the feature words based on the weights of the key feature words and the weights of the remaining feature words, and use the final weight of the feature words as the feature word weight.

[0028] Furthermore, a manual evaluation module is also provided in the evaluation and diagnosis module. When the reliability index model and the evaluation and diagnosis model are unable to complete fault identification diagnosis and fault assessment for the target state data of the target relay protection device, an expert performs manual evaluation processing to manually perform fault identification diagnosis and fault assessment. After the manual evaluation processing is completed, the reliability index model and the evaluation and diagnosis model are iteratively updated according to the fault identification diagnosis results and the fault assessment results.

[0029] Compared with the prior art, the beneficial effects of the embodiments of the present application are:

[0030] This application establishes a reliability index model to conduct operation monitoring, fault identification and diagnosis, and reliability evaluation on the status data of the target relay protection device. By extracting status features of various historical status data in operation under different scenarios, historical fault status data is obtained from various historical status data according to the extracted status features. Fault features of the historical fault status data are extracted in combination with the fault failure case categories corresponding to the historical fault status data to obtain fault features, and a fault classification library is established. The fault classification library has multiple fault subsets, and multiple rounds of training are performed to generate multiple base classification fault sub-models. These multiple base classification fault sub-models are weighted and fused to generate a reliability index model, so as to realize abnormal identification and failure location of relay protection devices, predict possible defects of relay protection devices, and intend to build a relay protection device status evaluation system to provide support for comprehensively improving the safety, reliability and intelligent operation and maintenance level of the relay protection system. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In the attached picture:

[0032] Figure 1 Schematic diagram of the system structure of an embodiment of the present application. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Generally, the components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0034] Example

[0035] like Figure 1 As shown, a fault diagnosis system for a relay protection device comprises:

[0036] Data acquisition module, used to obtain equipment data of historical relay protection devices, fault failure case categories, fault results, failure mechanisms and status data of relay protection devices to be monitored;

[0037] A model building module is used to extract fault features from the data acquired by the data acquisition module and to build a reliability index model based on the fault features;

[0038] The fault diagnosis module is used to use the reliability index model to perform operation monitoring, fault identification and diagnosis, and reliability evaluation on the status data of the target relay protection device.

[0039] The model building module preprocesses the data collected by the data acquisition module and cleans the data to remove abnormal data through scale normalization.

[0040] The specific steps for extracting fault features from the data acquired by the data acquisition module are as follows:

[0041] Obtain the equipment models of various historical relay protection devices, obtain the internal component parameters of each model of relay protection device, and the historical status data of each operating situation in different scenarios;

[0042] Combined with the internal component parameters, the status features of various historical status data in operation under different scenarios are extracted, and the historical fault status data is obtained from the various historical status data according to the extracted status features;

[0043] The fault characteristics of the historical fault status data are extracted in combination with the fault failure case categories corresponding to the historical fault status data to obtain the fault characteristics.

[0044] The specific steps of establishing a reliability index model based on fault characteristics are as follows:

[0045] Establish a fault classification library, which has multiple fault subsets. Each fault subset corresponds to a type of fault failure case. Store historical fault status data into the corresponding fault subset according to fault characteristics.

[0046] Training calculations are performed on each fault subset to obtain a fault training subset, which is then used to train a base classification fault sub-model. The base classification fault sub-model is trained for multiple rounds to generate multiple base classification fault sub-models. These multiple base classification fault sub-models are weighted and fused to generate a reliability index model.

[0047] There are m training samples (x1, y1), (x2, y2), ... (x m ,y m ), where x i ∈X,y i ∈Y={-1,+1}, initialize sample weight D1(i)=1 / m, t=1,...,T, where T is the number of integrated models, in the sample distribution D t Upper base classification fault submodel h t , the formula is as follows:

[0048] h t (X)→{-1,+1}

[0049] Base classification fault sub-model h t The weight vector calculation formula is as follows:

[0050]

[0051] where ε t h t In the distribution D tClassification error rate on ;

[0052] Update sample weights

[0053] Among them, Z t =Σ i D t (i)exp ( -α t y i h t (x i ))

[0054] Reliability Index Model

[0055] Classification error rate ε t The calculation formula is as follows:

[0056]

[0057] The specific steps for fault identification and diagnosis using the reliability index model are as follows:

[0058] First, a feature vector of each historical fault state data is generated according to the fault feature of each historical fault state data in the fault subset;

[0059] Extracting features of target state data of the target relay protection device, and generating a target feature vector according to the extracted features;

[0060] Calculate the similarity between the target feature vector and the feature vector of each historical fault state data to obtain the similarity vector between the target state data and each fault subset;

[0061] The maximum similarity vector is extracted from each similarity vector, and the fault failure case category corresponding to the fault subset to which the maximum similarity vector belongs is used as the fault failure case category of the target state data.

[0062] The fault subsets are A i×n , B p×n and C q×n , i, p and q are the number of historical fault status data in the fault subset, and the feature vector of the historical fault status data is n-dimensional: XA = (xa1, xa2, ...xa n )、XB=(xb1,xb2,...xb n ) and XC=(xc1,xc2,...xc n ),

[0063] The target state data X of the target relay protection device generates a feature vector X=(x1, x2, ...x n), the target state data X and the fault subset A i×n , Fault subset B p×n and the fault subset C q×n The similarity calculation is performed on each historical fault state data in the , and the similarity vector S=9d1,d2,...d i )、S=(d1,d2,...d p ) and S=(d1,d2,...d q ), take the target state data X and each fault subset A i×n , B p×n and C q×n The mean values ​​of similarities SA, SB and SC are compared;

[0064]

[0065] Take the largest value among them, then the target state data X belongs to the fault failure case category to which the fault subset belongs.

[0066] The target state data X and each fault subset A i×n , B p×n and C q×n The mean calculation formula of the similarity is as follows:

[0067]

[0068]

[0069] The similarity calculation is performed using the correlation coefficient similarity, and the calculation formula is as follows:

[0070]

[0071] Where X is the target state data, and Y is the historical fault state data.

[0072] After the target state data obtains the fault failure case category, it is stored in the corresponding fault subset according to its own fault failure case category. Every time a preset cycle is reached, the reliability index model is updated according to the update of the fault subset in the fault classification library.

[0073] An evaluation diagnosis module is provided in the fault diagnosis module, and an evaluation diagnosis model is provided in the evaluation diagnosis module. After obtaining the category of the fault failure case, the evaluation diagnosis model performs the corresponding fault evaluation to obtain the reliability evaluation result.

[0074] The process of establishing the evaluation and diagnosis model is as follows:

[0075] Obtain historical fault evaluation texts, extract feature words from each fault evaluation text, and calculate feature word weights;

[0076] Roughly classify each fault evaluation text, and calculate the similarity between fault evaluation texts based on the rough classification results and feature word weights;

[0077] According to the similarity calculation results, the historical fault evaluation texts are classified and processed, and the evaluation diagnosis model is obtained according to the classification processing results.

[0078] The weight of feature words is calculated by first screening and calculating the weight of key feature words in the fault evaluation text, and then calculating the weight of other feature words. The final weight of feature words is calculated based on the weight of key feature words and the weight of other feature words, and the final weight of feature words is used as the weight of feature words.

[0079] There is also a manual evaluation module in the evaluation and diagnosis module. When the reliability index model and the evaluation and diagnosis model are unable to complete fault identification diagnosis and fault assessment for the target state data of the target relay protection device, experts will perform manual evaluation processing to manually perform fault identification diagnosis and fault assessment. After the manual evaluation processing is completed, the reliability index model and the evaluation and diagnosis model will be iteratively updated according to the fault identification diagnosis results and fault assessment results.

[0080] The above-mentioned embodiments only express the preferred implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for ordinary technicians in this field, several modifications, improvements and substitutions can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application.

Claims

1. A fault diagnosis system for a relay protection device, characterized in that: include: Data acquisition module, used to obtain equipment data of historical relay protection devices, fault failure case categories, fault results, failure mechanisms and status data of relay protection devices to be monitored; A model building module is used to extract fault features from the data acquired by the data acquisition module and to build a reliability index model based on the fault features; The fault diagnosis module is used to use the reliability index model to perform operation monitoring, fault identification and diagnosis, and reliability evaluation on the status data of the target relay protection device.

2. A fault diagnosis system for a relay protection device according to claim 1, characterized in that: The model building module preprocesses the data collected by the data acquisition module and cleans the data to remove abnormal data through scale normalization.

3. A fault diagnosis system for a relay protection device according to claim 2, characterized in that: The specific steps of extracting fault features from the data acquired by the data acquisition module are as follows: Obtain the equipment models of various historical relay protection devices, obtain the internal component parameters of each model of relay protection device, and the historical status data of each operating situation in different scenarios; Combined with the internal component parameters, the status features of various historical status data in operation under different scenarios are extracted, and the historical fault status data is obtained from the various historical status data according to the extracted status features; The historical fault status data is combined with the fault failure case categories corresponding to the historical fault status data to perform feature extraction to obtain fault features.

4. A fault diagnosis system for a relay protection device according to claim 3, characterized in that: The specific steps of establishing the reliability index model according to the fault characteristics are as follows: Establish a fault classification library, which has multiple fault subsets. Each fault subset corresponds to a type of fault failure case. Store historical fault status data into the corresponding fault subset according to fault characteristics. Training calculations are performed on each fault subset to obtain a fault training subset, which is then used to train a base classification fault sub-model. The base classification fault sub-model is trained for multiple rounds to generate multiple base classification fault sub-models. The multiple base classification fault sub-models are weighted and fused to generate a reliability index model.

5. A fault diagnosis system for a relay protection device according to claim 4, characterized in that: The specific steps of the fault diagnosis module for fault identification and diagnosis are as follows: First, a feature vector of each historical fault state data is generated according to the fault feature of each historical fault state data in the fault subset; Extracting features of target state data of the target relay protection device, and generating a target feature vector according to the extracted features; Calculate the similarity between the target feature vector and the feature vector of each historical fault state data to obtain the similarity vector between the target state data and each fault subset; The maximum similarity vector is extracted from each similarity vector, and the fault failure case category corresponding to the fault subset to which the maximum similarity vector belongs is used as the fault failure case category of the target state data.

6. A fault diagnosis system for a relay protection device according to claim 5, characterized in that: After obtaining the failure case category, the target state data is updated to the corresponding failure subset according to the failure case category, and the reliability indicator model is updated every preset period.

7. A fault diagnosis system for a relay protection device according to claim 5, characterized in that: The fault diagnosis module is provided with an evaluation diagnosis module, and the evaluation diagnosis module is provided with an evaluation diagnosis model. After the fault failure case category is obtained, the evaluation diagnosis model performs corresponding fault evaluation to obtain a reliability evaluation result.

8. A fault diagnosis system for a relay protection device according to claim 7, characterized in that: The evaluation and diagnosis model establishment process is as follows: Obtain historical fault evaluation texts, extract feature words from each fault evaluation text, and calculate feature word weights; Roughly classify each fault evaluation text, and calculate the similarity between fault evaluation texts based on the rough classification results and feature word weights; According to the similarity calculation results, the historical fault evaluation texts are classified and processed, and the evaluation diagnosis model is obtained according to the classification processing results.

9. A fault diagnosis system for a relay protection device according to claim 8, characterized in that: The calculation of feature word weights is to first screen and calculate the weights of key feature words in the fault evaluation text, then calculate the weights of the remaining feature words, calculate the final weight of the feature words based on the weights of the key feature words and the weights of the remaining feature words, and use the final weight of the feature words as the feature word weight.

10. A fault diagnosis system for a relay protection device according to claim 9, characterized in that: The evaluation and diagnosis module is also provided with a manual evaluation module. When the reliability index model and the evaluation and diagnosis model are unable to complete fault identification diagnosis and fault assessment for the target state data of the target relay protection device, an expert performs manual evaluation processing to manually perform fault identification diagnosis and fault assessment. After the manual evaluation processing is completed, the reliability index model and the evaluation and diagnosis model are iteratively updated according to the fault identification diagnosis results and the fault assessment results.

Images