Power station system evaluation method, system, equipment and medium

Abstract

The invention provides a power station system evaluation method, system, equipment and medium, and the method comprises the following steps: S1, obtaining the data information of a power station system, generating a power station system fault experience knowledge base which comprises a plurality of types of collected data and risk evaluation index sets, the risk evaluation index set is generated according to multiple types of collected data; s2, collecting actual data of the power station in a running state in real time, generating a weight vector set according to preprocessing and analysis of the actual data, and generating a fuzzy relation matrix according to a membership function; and S3, according to the weight vector set and the fuzzy relation matrix, calculating to obtain a comprehensive evaluation matrix, and according to a comprehensive evaluation matrix result, obtaining information corresponding to an evaluation result. According to the method, the safe and stable operation level and the operation and maintenance integrated intelligent level of the unit can be effectively improved, the risk forecast of system faults is given in advance, and sudden faults are avoided.

Description

Technical Field

[0001] The present invention belongs to the field of electrical automation technology, and in particular relates to a power station system evaluation method, system, equipment and medium. Background Art

[0002] With the rapid development of my country's economy and society, the power load has increased rapidly, the peak-to-valley difference has continued to increase, and the requirements for the stability of the power grid have become increasingly higher. Insufficient peak-shaving capacity will become a prominent problem restricting the development of the power system. Conventional hydropower stations, especially pumped-storage power stations, with their unique operating characteristics of peak-to-valley shifting, play the role of regulating loads, promoting energy conservation in power systems, and maintaining safe and stable operation of power grids, and have gradually become an effective and indispensable means of regulation for my country's power system. Against the background of the continuous increase in installed capacity of power stations, the structure of the units has become increasingly complex and the operating conditions have become worse, which has accelerated the deterioration of the units and increased the probability of failures, and put forward higher requirements for the management, maintenance, monitoring, and diagnosis of the units.

[0003] A pumped storage unit is a complex nonlinear dynamic system. The formation and development of faults during its operation are extremely random. However, the traditional fault diagnosis modeling theory and methods in the existing technology have long adopted time-frequency and time-space transformation analysis methods, which makes it difficult to accurately mathematically describe a large number of uncertain factors. As a result, it is difficult for the actual state maintenance system to draw more accurate diagnostic conclusions, which greatly restricts the application of fault diagnosis theory and methods in engineering practice. Summary of the invention

[0004] The first object of the present invention is to provide a power plant system evaluation method to improve the diagnosis level, enhance the operation stability, and enhance the precision of fault data classification.

[0005] To this end, the above-mentioned purpose of the present invention is achieved through the following technical solutions:

[0006] A power plant system evaluation method comprises the following steps:

[0007] S1. Acquire power plant system data information and generate a power plant system fault experience knowledge base, wherein the system fault experience knowledge base includes multiple types of collected data and a risk assessment indicator set, and the risk assessment indicator set is generated based on the multiple types of collected data;

[0008] S2, real-time collection of actual data of the power station under operation status, generating a weight vector set based on preprocessing and analysis of the actual data, and generating a fuzzy relationship matrix based on the membership function;

[0009] The membership function is a three-dimensional membership function constructed by a two-dimensional relationship between at least two types of collected data, wherein the at least two types of collected data have a transitive association relationship but do not belong to the data corresponding to the bottom events of the same cut set;

[0010] S3. Based on the weight vector set and the fuzzy relationship matrix, a comprehensive evaluation matrix is ​​calculated and obtained, and based on the comprehensive evaluation matrix result, information corresponding to the evaluation result is obtained.

[0011] While adopting the above technical solutions, the present invention may also adopt or combine the following technical solutions:

[0012] As a preferred technical solution of the present invention: in step S2, the membership function is:

[0013]

[0014] Among them, y lm is the absolute value of the collected value of the first type of data; x lm is the absolute value of the collected value of the second type of data; x min <x 1 <x 2 <x max y is a node in the threshold range corresponding to the collected value of the second type of data; min <y 1 <y 2 <y max k is a node in the threshold range corresponding to the collected value of the first type of data; 1 , k 2 is the correlation coefficient constant value of the first type of data and the second type of data in different threshold intervals; l is the dimension corresponding to the first type of data; m is the dimension corresponding to the second type of data, where a is the value in (x 2 ,x max ),(y 2 ,y max ) interval data fitting average slope constant.

[0015] As a preferred technical solution of the present invention: 2 The absolute value of k is greater than 1 The absolute value of .

[0016] As a preferred technical solution of the present invention: the risk assessment index set includes a risk assessment index set determined in a first fault tree classification model trained by a system fault experience knowledge base;

[0017] The first fault tree classification model performs the following training steps:

[0018] S11, qualitative analysis, using the descending method to obtain the minimum cut set, the minimum cut set is a combination of bottom events that cannot be reduced in number causing the occurrence of the top event of the system;

[0019] S12, quantitative analysis to obtain failure probability and bottom event importance;

[0020] S13, determining the probability of a bottom event, wherein the bottom event probability determination is sequentially determining the failure distribution, estimating the distribution parameters based on the distribution model, and determining the occurrence probability according to the distribution model;

[0021] The top event is the most undesirable fault state;

[0022] The intermediate event is a sub-level fault factor that causes the top event to occur;

[0023] The bottom events are all direct factors that cause the intermediate events to occur, and the all direct factors are factors that do not need to be further investigated.

[0024] As a preferred technical solution of the present invention: in the quantitative analysis, the bottom event importance analysis includes bottom event probability importance analysis, bottom event structure importance analysis and bottom event key importance analysis;

[0025] The formula for analyzing the probability importance of bottom events is:

[0026]

[0027] The bottom event structure importance analysis formula is:

[0028]

[0029] The key importance analysis formula of the bottom event is:

[0030]

[0031] Among them, the probability of system failure when in the critical state of bottom event i is called probability importance

[0032] Q i (t) represents the failure probability of the i-th bottom event at time t, g[Q(t)] represents the failure probability of the top event at time t, and g[1 i , Q(t)] represents the failure probability of the top event at time t when the i-th bottom event fails, g[0 i , Q(t)] represents the failure probability of the top event at time t when the i-th bottom event is normal;

[0033] When the bottom event i changes from state 0 to 1, the ratio of the critical state number of the bottom event i to the total state number, the structural importance The effective calculation of is: The failure probability of all bottom events is Q k Set to 0.5;

[0034] The critical importance of the bottom event reflects the probability of the bottom event i triggering the system failure by using the relative value of the change rate of the system failure probability to the change rate of the bottom event i failure probability, where g(t) is the failure probability at the moment.

[0035] As a preferred technical solution of the present invention: the failure probability of the bottom event is the probability of failure of each link of the system calculated from bottom to top:

[0036]

[0037] Where Ei is the event that all the bottom events belonging to the minimum cut set Kj occur, where j is the event dimension of the minimum cut set, i is the event dimension of the bottom event, k is the total number of events in the minimum cut set, and F j is the probability expression of the union of the total number of k events, P r {E r} is the probability of each bottom event in the total number of k events, r is the calculation dimension of the probability of a single bottom event, P i {E i ∩E j} is the intersection probability of two events, is the probability of the intersection of all r events.

[0038] As a preferred technical solution of the present invention: the process of obtaining the minimum cut set is:

[0039] Starting from the top event of the fault tree, from top to bottom, replace the previous level event with the next level event in turn. When encountering an AND gate, write the input events horizontally in parallel. When encountering an OR gate, write the input events vertically in series until all logic gates are replaced with bottom events. At this time, the last column represents all cut sets. Then simplify the cut sets and absorb all the minimum cut sets Kj.

[0040] The second object of the present invention is to provide a power plant evaluation system, comprising the following modules:

[0041] System fault data information acquisition module, used to acquire power station system fault data information, compile and generate a fault experience knowledge base;

[0042] System fault tree construction module, used to classify the system fault diagnosis knowledge base, define top events, intermediate events and bottom events to construct a fault tree with a tree structure step by step;

[0043] The system evaluation module collects the actual data of the power station in operation in real time, generates a weight vector set based on the preprocessing and analysis of the actual data, and generates a fuzzy relationship matrix based on the membership function; calculates and obtains a comprehensive evaluation matrix based on the weight vector set and the fuzzy relationship matrix, and obtains information corresponding to the evaluation result based on the comprehensive evaluation matrix result;

[0044] The minimum cut set corresponding to the membership function is constructed based on a fault experience knowledge base, and the membership function is a three-dimensional membership function constructed by a two-dimensional relationship between at least two types of collected data, and the data corresponding to the bottom events that have a transitive association relationship between at least two types of collected data but do not belong to the same cut set.

[0045] The third object of the present invention is to provide an electronic device, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus, and wherein:

[0046] A memory, the memory being used to store a computer program;

[0047] A processor is used to execute a computer program stored in a memory to implement the steps of the power plant system evaluation method as described above.

[0048] Another object of the present invention is to provide a non-transitory readable storage medium, characterized in that: the non-transitory readable storage medium stores a program, and when the program is executed by a processor, it implements the power plant system evaluation method steps as described above.

[0049] The present invention provides a power plant system evaluation method, system, equipment and medium, which have the following beneficial effects:

[0050] 1) Based on the power plant fault diagnosis method of the present invention that takes into account the membership function of multi-dimensional data, the possible degradation of the system is grasped, the safe and stable operation level of the unit and the intelligent level of integrated operation and maintenance are effectively improved, and the risk forecast of system failures can be given in advance to avoid the occurrence of sudden failures.

[0051] 2) It is proposed to consider the correlation between the two cut set data in the membership function, so as to form a membership function set with data space, which can more accurately feedback the correlation and transitivity of the data between the cut sets, form the data correlation analysis between the cut sets in the evaluation system, and improve the evaluation accuracy of the system, so as to facilitate the evaluation of the power station system with higher accuracy and output a maintenance strategy with higher security.

[0052] 3) It can effectively improve the safe and stable operation level of the unit and the intelligent level of integrated operation and maintenance, give risk forecasts of system failures in advance, and avoid the occurrence of sudden failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flow chart of the power plant system evaluation method provided by the present invention.

[0054] Figure 2 A schematic diagram of the sub-process steps of the method for obtaining the cut set of the fault tree and the fault logic.

[0055] Figure 3 Schematic diagram of the composition space of the membership function. DETAILED DESCRIPTION

[0056] The present invention will be described in further detail with reference to the accompanying drawings and specific examples.

[0057] like Figure 1 As shown, a power plant evaluation method includes:

[0058] S1. Acquire power plant system data information and generate a system failure experience knowledge base, where the system failure experience knowledge base includes multiple types of collected data and a risk assessment indicator set generated based on the multiple types of collected data;

[0059] S2. Real-time collection of actual data of the power station in operation, generation of a weight vector set based on preprocessing and analysis of the actual data, and generation of a fuzzy relationship matrix based on a membership function;

[0060] S3. Based on the weight vector set and the fuzzy relationship matrix, a comprehensive evaluation matrix is ​​calculated and obtained, and based on the comprehensive evaluation matrix result, information corresponding to the evaluation result is obtained; the information corresponding to the evaluation result in the present invention refers to the grade classification information of the fault information based on the evaluation matrix.

[0061] The membership function is a three-dimensional membership function constructed by a two-dimensional relationship between at least two types of collected data, and the at least two types of collected data have a transitive association relationship but do not belong to the data corresponding to the bottom events of the same cut set.

[0062] The membership function is:

[0063]

[0064] Among them, y lm is the absolute value of the collected value of the first type of data; x lm is the absolute value of the collected value of the second type of data, and both are related data of two types preliminarily screened out through deep learning; x min <x1 <x 2 <x max y is a node in the threshold range corresponding to the collected value of the second type of data; min <y 1 <y 2 <y max is a node in the threshold range interval corresponding to the collected value of the first type of data; where a is the node in (x 2 , x max ),(y 2 ,y max ) interval data fitting average slope constant. k 1 , k 2 is the correlation coefficient constant value of the first type of data and the second type of data in different threshold intervals. In a specific implementation, in different data threshold ranges, k 1 , k 2 The value can be assigned through empirical values, or obtained by performing data correlation fitting calculation in actual calculation operation, where l is the dimension corresponding to the first type of data; and m is the dimension corresponding to the second type of data. 2 The absolute value of k is greater than 1 The absolute value of the correlation coefficient constant of the data generally changes in the classification interval of the data, and the correlation becomes larger. The risk assessment index set includes the risk assessment index set determined in the first fault tree classification model trained by the system fault experience knowledge base.

[0065] According to the principle of the membership function implemented by the present invention, as shown in Table 1 below, in the cut set corresponding to each fault tree, each cut set has its own corresponding monitoring parameters, and the membership matrix is ​​based on the actual measured value of the data, or the normalized coefficient of the data, and the fault tree cut set corresponds to a corresponding evaluation level, for example, the four levels of A, B, C, and D correspond to several different types of states such as serious, maintenance, key monitoring, and healthy operation. The membership function of the present invention considers the correlation between two parameters between different cut sets, so that when constructing the parameter matrix between each fault cut set, the original independent and separate membership relationship is considered, but based on the parameter relationship between the two cut sets, a correction deviation of the membership function is formed, so that the membership function can link the relationship between the two cut sets, and the distribution of the coefficients and the acquisition of the function are adjusted according to the training relationship between the two data sets under different data states, such as Figure 3As shown in , the spatial membership function established in this way expands the dimension of the granularity of the data correlation in the three-dimensional space in the two-dimensional space, and improves the data acquisition accuracy by the data correlation under the three-dimensional space structure. With the development of the current big data processing technology, it can improve the granularity of the collected data analysis, obtain more accurate fuzzy classification results, and obtain more accurate evaluation results.

[0066] Table 1

[0067]

[0068] Obtain power plant system fault data information and generate a system fault knowledge base;

[0069] Classify the system fault data information according to the fault knowledge base and perform training to generate a first fault tree classification model;

[0070] A first input data set is obtained and input into a first fault tree classification model to complete fault warning and intelligent diagnosis of the system; the first input data set includes system body data, unit operation data and oil pressure device data.

[0071] The first fault tree classification model performs the following steps:

[0072] S11. Qualitative analysis: Use the descending method to obtain the minimum cut set, which is the combination of bottom events that cannot be reduced in number to cause the top event of the system to occur.

[0073] S12, quantitative analysis to obtain failure probability and bottom event importance;

[0074] S13, determining the probability of a bottom event, which includes determining the failure distribution, estimating the distribution parameters based on the distribution model, and determining the occurrence probability based on the distribution model;

[0075] The top event is the most undesirable fault state;

[0076] The intermediate event is the sub-level fault factor that causes the top event;

[0077] The base event is all the direct factors that cause the intermediate event to occur, and all direct factors are factors that do not need to be further investigated.

[0078] The acquisition of the fault knowledge base according to the present invention includes but is not limited to collecting data based on the actual operating status of the system, or using a deep learning model to accumulate and process the relevant data of the fault in the fault library to achieve early warning and diagnosis.

[0079] It includes obtaining power station fault data information and compiling a system fault knowledge base; classifying system faults and building a system fault tree based on the obtained system fault knowledge base; based on the obtained system fault tree, combining system body data, unit operation data and oil pressure device data, using a system fault tree model to perform fault analysis on the system operation status, and completing fault warning and intelligent diagnosis of the power station system.

[0080] Specifically, the power plant system diagnosis model research takes the power plant system fault data information in the existing technology as the knowledge basis, and compiles the fault types of each subsystem and subcomponent of the existing fault compilation and collection system under different working conditions into a system fault knowledge base.

[0081] On the basis of the above system fault knowledge base, the system operation fault categories are divided into two categories: main body faults and oil pressure system faults according to the common fault types of power station systems, and the sub-events that may cause one of the above faults are listed, thus forming a system fault tree.

[0082] On the basis of the established system fault tree, a system fault tree and system model are established for all possible fault conditions in the system. Combined with the system body data, unit operation data and oil pressure device data, the system fault tree model is used to perform fault analysis on the system operation status to achieve power station system fault early warning and intelligent diagnosis.

[0083] Among them, the fault tree model establishes a tree structure based on the power station system fault, takes the most undesirable fault state as the top event, finds out all the fault factors that directly cause this fault, and uses it as the intermediate event of the tree, and then finds out all the direct factors that cause the next event, and keeps tracing until there is no need to further investigate the factors, which is the bottom event. The design of each node structure in the fault tree includes: node number; node name: fault type name; node type: AND, OR, NOT logic gate; occurrence probability: the bottom event is the set value, and the probability of occurrence of the top event and the intermediate event is determined by the probability of occurrence of the bottom event; parent node: the node number directly affected by the current node; child node: the node number set directly affected by the current node, in the form of an array.

[0084] All faults or events in the above fault tree model are connected through set logic gates, thereby reflecting the logical structural relationship between specific events of the system or equipment and the fault events of its various subsystems or components.

[0085] like Figure 2 As shown in , the fault tree analysis of the system fault tree model mainly includes the following three parts: qualitative analysis, quantitative analysis and determination of the bottom event probability.

[0086] The qualitative analysis of the fault tree implemented according to the present invention is mainly to obtain the minimum cut set, that is, the combination of bottom events that cause the top event of the fault tree to occur. The minimum cut set is a combination of bottom events that cannot be reduced in number to cause the top event of the fault tree to occur. It represents a fault mode that causes the top event of the fault tree to occur. Any fault tree is composed of a finite number of minimum cut sets, which are unique to a given top event of the fault tree. The minimum cut set composed of a single event means that the top event will occur once the event occurs. The minimum cut set composed of a double event means that the two events will cause the top event to occur only when they occur together. For the minimum cut set composed of N events, these N events must occur simultaneously for the top event to occur.

[0087] Generally, the downward method is used to find the minimum cut set, that is, starting from the top event of the fault tree, from top to bottom, the previous level event is replaced with the next level event in turn. When encountering an AND gate, the input event is written out horizontally in parallel. When encountering an OR gate, the input event is written out vertically in series until all logic gates are replaced with bottom events. At this time, the last column represents all cut sets, and then the cut sets are simplified to absorb all the minimum cut sets.

[0088] The quantitative fault tree analysis implemented according to the present invention mainly includes failure probability analysis and bottom event importance analysis. The failure probability analysis calculates the failure probability of each link of the system through a bottom-up method.

[0089] Assume that the minimum cut set expression of the fault tree is K j (X), then the minimum cut set structure function is:

[0090] In the above formula, k is the total number of minimum cut sets, K j (X) is defined as

[0091] To find the probability of the top event, that is, the probability of making θ(X) = 1, we just need to take the mathematical expectation of both ends of the above equation, and the left side is the probability of the top event:

[0092] Let Ei be the event that all the bottom events belonging to the minimum cut set Kj occur, then the top event is the event that at least one of the k Ei occurs, so

[0093] If the event and probability are written as Fj, then F j =Σ 1＜j1＜…＜k Pr{E i1 ∩E i2 ∩…E ij}, which is used to compute the probability of the union of multiple events, first calculate the sum of the probabilities of all individual events, subtract the sum of the probabilities of the intersection of all two events, add the sum of the probabilities of the intersection of all three events, and so on, until the probabilities of the intersection of all k events have been added or subtracted.

[0094] By expanding the above formula, we can get:

[0095]

[0096] Where Ei is the event that all the bottom events belonging to the minimum cut set Kj occur, where j is the event dimension of the minimum cut set, i is the event dimension of the bottom event, k is the total number of events in the minimum cut set, and F j is the probability expression of the union of the total number of k events, P r {E r} is the probability of each bottom event in the total number of k events, r is the calculation dimension of the probability of a single bottom event,

[0097] P i {E i ∩E j} is the intersection probability of two events, is the probability of the intersection of all r events.

[0098] Through the above method, the failure probability analysis of the top event can be completed. At the same time, similar methods can also be used to perform failure probability analysis on each link of the number of failures. Similar methods will not be described in detail here.

[0099] The bottom event importance analysis in the fault tree model implemented according to the present invention includes analysis of the bottom event probability importance, structural importance and critical importance.

[0100] Among them, probability importance: When the system is in the critical state of component i (when event i fails, the state that causes system failure is called the critical state. Among the 2n-1 situations in which event i fails, only those situations that cause system failure are the critical states of component i), the probability of system failure is called probability importance I i Pr (t).

[0101] If Q i (t) represents the failure probability of the i-th bottom event at time t, g[Q(t)] represents the failure probability of the top event at time t, and g[1 i , Q(t)] represents the failure probability of the top event at time t when the i-th bottom event fails, g[0 i , Q(t)] represents the failure probability of the top event at time t when the i-th bottom event is normal. Then the probability importance for:

[0102] Structural importance: When the bottom event i changes from state 0 to 1, the ratio of the critical state number of bottom event i to the total state number. When calculating the structural importance, in one specific implementation, in the probability importance expression of event i, the failure probability of all bottom events is set to 0.5. Therefore, the structural importance The effective calculation of is:

[0103] Critical importance: The relative value of the rate of change of the system failure probability to the rate of change of the failure probability of the bottom event i is used to reflect the probability of the bottom event i triggering the system failure. Its definition is expressed as: For system fault diagnosis and inspection, determining the critical importance of bottom events has very important guiding significance, because once a system failure occurs, maintenance personnel have reason to first suspect that the bottom event with the greatest critical importance has triggered the system failure.

[0104] Among them, the quantitative analysis based on the fault tree is based on the probability of occurrence of the bottom event, so the determination of the probability of the bottom event is crucial to the entire analysis.

[0105] The determination of the probability of occurrence of bottom events can be mainly divided into three steps: determination of failure distribution, estimation of distribution parameters based on distribution model, and determination of occurrence probability based on distribution model. In practice, the failure distribution curve of each bottom event can be obtained according to the occurrence of each fault during an overhaul period, and then the standard probability distribution model that conforms to the failure probability distribution can be determined. Based on the obtained distribution model, the distribution curve of the probability of occurrence of bottom events is obtained by using parameter identification method. Using this curve, combined with the time from the last overhaul, the probability of occurrence of each event can be determined.

[0106] Through the above method, an open and extensible fault diagnosis knowledge base framework is established, the fault diagnosis knowledge base and fault sample standard library are accumulated and updated in real time, and a fault tree diagnosis model based on fault reasoning is constructed, which realizes early warning and intelligent diagnosis of potential faults of power station units, provides theoretical guidance and technical support for the formulation of system maintenance strategies, and ensures the safety and reliability of system operation.

[0107] The present invention also provides a power station system fault diagnosis system. The system for implementing the power station fault diagnosis method disclosed by the present invention includes a system fault data information acquisition module, a system fault tree construction module, and a system fault tree analysis module; the system fault data information acquisition module, the system fault tree construction module, and the system fault tree analysis module are connected in series in sequence; the system fault data information acquisition module is used to acquire power station system fault data information and compile a system fault knowledge base; the system fault tree construction module is used to classify the system fault diagnosis knowledge base, define top events, intermediate events, and bottom events, and construct a tree-structured system fault tree step by step; the system fault tree analysis module is used to perform qualitative analysis, quantitative analysis, and bottom event probability determination on the system fault tree in combination with system body data, unit operation data, and oil pressure device data to determine the system fault warning and intelligent diagnosis.

[0108] The contents described in this specification are merely examples of the present invention. Those skilled in the art to which the present invention belongs may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the contents of the present specification or exceed the scope defined by the claims, they shall fall within the protection scope of the present invention.

Claims

1. A power plant system evaluation method, characterized in that: The steps include: S1. Acquire power plant system data information and generate a power plant system fault experience knowledge base, wherein the system fault experience knowledge base includes multiple types of collected data and a risk assessment indicator set, and the risk assessment indicator set is generated based on the multiple types of collected data; S2, real-time collection of actual data of the power station under operation status, generating a weight vector set based on preprocessing and analysis of the actual data, and generating a fuzzy relationship matrix based on the membership function; The membership function is a three-dimensional membership function constructed by a two-dimensional relationship between at least two types of collected data, wherein the at least two types of collected data have a transitive association relationship but do not belong to the data corresponding to the bottom events of the same cut set; S3. Based on the weight vector set and the fuzzy relationship matrix, a comprehensive evaluation matrix is ​​calculated and obtained, and based on the comprehensive evaluation matrix result, information corresponding to the evaluation result is obtained.

2. The method according to claim 1, characterized in that: In step S2, the membership function is: Among them, y lm is the absolute value of the collected value of the first type of data; x lm is the absolute value of the collected value of the second type of data; x min <x1<x2<x max y is a node in the threshold range corresponding to the collected value of the second type of data; min <y1<y2<y max is the node of the threshold range interval corresponding to the collection value of the first type of data; k1 and k2 are the correlation coefficient constant values ​​of the first type of data and the second type of data in different threshold intervals; l is the dimension corresponding to the first type of data; m is the dimension corresponding to the second type of data, where a is the dimension of (x2, x max ),(y2,y max ) interval data fitting average slope constant.

3. The method according to claim 2, characterized in that: The absolute value of k2 is greater than the absolute value of k1.

4. The method according to claim 1, characterized in that: The risk assessment index set includes a risk assessment index set determined in a first fault tree classification model trained by a system failure experience knowledge base; The first fault tree classification model performs the following training steps: S11, qualitative analysis, using the descending method to obtain the minimum cut set, the minimum cut set is a combination of bottom events that cannot be reduced in number causing the occurrence of the top event of the system; S12, quantitative analysis to obtain failure probability and bottom event importance; S13, determining the probability of a bottom event, wherein the bottom event probability determination is sequentially determining the failure distribution, estimating the distribution parameters based on the distribution model, and determining the occurrence probability according to the distribution model; The top event is the most undesirable fault state; The intermediate event is a sub-level fault factor that causes the top event to occur; The bottom events are all direct factors that cause the intermediate events to occur, and the all direct factors are factors that do not need to be further investigated.

5. The method according to claim 4, characterized in that: In quantitative analysis, bottom event importance analysis includes bottom event probability importance analysis, bottom event structure importance analysis and bottom event key importance analysis; The formula for analyzing the probability importance of bottom events is: The bottom event structure importance analysis formula is: The key importance analysis formula of the bottom event is: Among them, the probability of system failure when in the critical state of bottom event i is called probability importance Q i (t) represents the failure probability of the i-th bottom event at time t, g[Q(t)] represents the failure probability of the top event at time t, and g[1 i , Q(t)] represents the failure probability of the top event at time t when the i-th bottom event fails, g[0 i , Q(t)] represents the failure probability of the top event at time t when the i-th bottom event is normal; When the bottom event i changes from state 0 to 1, the ratio of the critical state number of the bottom event i to the total state number, the structural importance The effective calculation of is: The failure probability of all bottom events is Q k Set to 0.5; The critical importance of the bottom event reflects the probability of the bottom event i triggering the system failure by using the relative value of the change rate of the system failure probability to the change rate of the bottom event i failure probability, where g(t) is the failure probability at the moment.

6. The method according to claim 5, characterized in that: The failure probability of the bottom event is the probability of failure of each link in the system calculated from bottom to top: Where Ei is the event that all the bottom events belonging to the minimum cut set Kj occur, where j is the event dimension of the minimum cut set, i is the event dimension of the bottom event, k is the total number of events in the minimum cut set, and F j is the probability expression of the union of the total number of k events, P r {E r } is the probability of each bottom event in the total number of k events, r is the calculation dimension of the probability of a single bottom event, P i {E i ∩E j } is the intersection probability of two events, is the probability of the intersection of all r events.

7. The method according to claim 6, characterized in that: The process of obtaining the minimum cut set is: Starting from the top event of the fault tree, from top to bottom, replace the previous level event with the next level event in turn. When encountering an AND gate, write the input events horizontally in parallel. When encountering an OR gate, write the input events vertically in series until all logic gates are replaced with bottom events. At this time, the last column represents all cut sets. Then simplify the cut sets and absorb all the minimum cut sets Kj.

8. A power plant evaluation system, characterized in that: Includes the following modules: System fault data information acquisition module, used to acquire power station system fault data information, compile and generate a fault experience knowledge base; System fault tree construction module, used to classify the system fault diagnosis knowledge base, define top events, intermediate events and bottom events to construct a fault tree with a tree structure step by step; The system evaluation module collects the actual data of the power station in real time, generates a weight vector set based on the preprocessing and analysis of the actual data, and generates a fuzzy relationship matrix based on the membership function; According to the weight vector set and the fuzzy relationship matrix, a comprehensive evaluation matrix is ​​calculated and obtained, and according to the result of the comprehensive evaluation matrix, information corresponding to the evaluation result is obtained; The minimum cut set corresponding to the membership function is constructed based on a fault experience knowledge base, and the membership function is a three-dimensional membership function constructed by a two-dimensional relationship between at least two types of collected data, and the data corresponding to the bottom events that have a transitive association relationship between at least two types of collected data but do not belong to the same cut set.

9. An electronic device, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus, characterized in that: A memory, the memory being used to store a computer program; A processor, wherein the processor is used to execute a computer program stored in a memory to implement the steps of the power plant system evaluation method according to any one of claims 1 to 7.

10. A non-transitory readable storage medium, characterized in that: The non-transitory readable storage medium stores a program, and when the program is executed by the processor, the steps of the power plant system evaluation method as described in any one of claims 1-7 are implemented.

Images