Safety inspection method for smart power plant

Abstract

The invention discloses a safety inspection method for a smart power plant, and the method comprises the steps: collecting historical operation fault information, and classifying the collected historical operation fault information; constructing a network structure of the power generation system by determining nodes and edges according to equipment of the power plant; calculating a node degree and an edge weight according to the determined node and edge, simulating fault propagation according to the node degree and the edge weight in the network structure, and calculating a propagation speed and an influence range; adjusting the fault associated parameters according to the network structure; collecting fault time nodes, generating a fault association interval, determining influence association parameters based on the fault association interval, constructing a multi-level data model, inputting monitoring data into the multi-level data model, and obtaining a safety early warning result. According to the method, the thermal power generation system is specifically of a network structure, and the fault correlation parameters are adjusted, so that the parameters can capture the nonlinear propagation path of the fault in the system, and the interaction and dependency relationship among the components in the system can be more comprehensively considered.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical digital data processing, and in particular to a safety inspection method for a smart power plant. Background Art

[0002] In the operation and management of modern power plants, safety inspections are critical for ensuring stable operation and preventing accidents. With the continuous expansion of power plant scale and the increasing complexity of equipment, traditional inspection methods are no longer able to meet the needs of efficient and accurate safety management. This has led to the concept of the smart power plant, which leverages advanced information technology, the Internet of Things, big data analytics, and artificial intelligence to provide real-time monitoring, intelligent analysis, and early warning of power plant operating conditions, thereby improving safety and reliability.

[0003] For example, Chinese patent publication number CN115081926B discloses an operation safety early warning method and system applicable to smart power plants, which relates to the field of electrical digital data processing technology, wherein the method includes: classifying and integrating historical faults and dividing them according to the thermal power generation production process to obtain divided, classified and integrated data, determining fault-related parameters and related ratios based on the divided, classified and integrated data, generating fault-related intervals based on fault time nodes, determining a set of influencing related parameters based on the divided, classified and integrated data and the fault-related intervals, and determining the influence weight of each influencing related parameter, constructing a multi-level data model based on the fault-related parameters and the set of influencing related parameters, and inputting the monitoring data into the multi-level data model to obtain an operation safety early warning result.

[0004] In existing patented technologies, it is far from enough for power plant safety inspections to rely solely on simple analysis of historical fault information. It is also necessary to build a network structure model that can reflect the relationships between power plant equipment. Summary of the Invention

[0005] The present application provides a safety inspection method for a smart power plant. By concretizing the thermal power generation system into a network structure and adjusting the fault association parameters, these parameters can capture the nonlinear propagation path of the fault in the system and can more comprehensively consider the interactions and dependencies between components in the system.

[0006] This application provides a safety inspection method for a smart power plant, including:

[0007] S101, collecting historical operation fault information and classifying the collected historical operation fault information;

[0008] S102, constructing a network structure of a power generation system by determining nodes and edges based on the equipment in the power plant;

[0009] S103, calculating the node degrees and edge weights based on the determined nodes and edges, simulating fault propagation based on the node degrees and edge weights in the network structure, and calculating the propagation speed and impact range;

[0010] S104, adjusting fault-related parameters based on historical operating fault information and network structure, where the fault-related parameters include node degree, edge weight, propagation speed, and impact range;

[0011] S105, collecting fault time nodes, generating fault correlation intervals according to the fault time nodes, determining impact correlation parameters based on the fault correlation intervals, building a multi-level data model according to the fault correlation parameters and the impact correlation parameters, inputting the monitoring data of the power plant into the multi-level data model, and obtaining safety warning results.

[0012] Preferably, each node in the network structure is traversed, and the number of edges directly connected to the node is counted. The value of the number of edges is the degree of the node. Two nodes are connected by edges, and the edges represent the connection between the nodes.

[0013] Preferably, the historical operation fault information is divided according to the startup time sequence of the production process system in the power generation process, and the node degree is adjusted. The formula is: new node degree = original node degree × adjustment coefficient, where the original node degree is the original degree of the node in the network, and the adjustment coefficient is a number greater than 1, which is used to increase the weight of high node degrees. The adjustment coefficient is obtained through expert experience and historical fault information.

[0014] Preferably, the edge weight is adjusted according to the formula: new edge weight = original edge weight + fault propagation frequency × adjustment factor, where the original edge weight is the original weight of the edge, the fault propagation frequency is the frequency of fault propagation on this edge, and the adjustment factor is a coefficient used to adjust the impact of the fault propagation frequency on the edge weight.

[0015] Preferably, the propagation speed is adjusted according to the formula: new propagation speed = original propagation speed × (actual propagation speed / simulated propagation speed), where the original propagation speed is the initial propagation speed set in the simulation, the actual propagation speed is the propagation speed observed in the actual network due to the fault, and the simulated propagation speed is the propagation speed originally set in the simulation.

[0016] Preferably, the impact range is adjusted, and the formula is: new impact range = original impact range × (actual impact range coefficient / simulated impact range coefficient), where the original impact range is the initial impact range calculated in the simulation, the actual impact range coefficient is the coefficient of the actual impact range of the fault obtained based on actual data, and the simulated impact range coefficient is the impact range coefficient originally set in the simulation.

[0017] Preferably, the step of dynamically adjusting the fault-related interval is:

[0018] S201, based on historical fault data, collecting data fluctuation characteristics;

[0019] S202, dynamically calculating the time span based on the fluctuation characteristics of the data;

[0020] S203, classifying the fault types into fast-propagating faults, slowly developing faults, and local impact faults, and setting differentiated time spans based on the classified fault types;

[0021] S204 , combining the dynamically calculated time span with the differentiated time span to form a final dynamic time span, and adjusting the fault correlation interval according to the dynamic time span.

[0022] Preferably, the normal fluctuation range of the parameter is established by calculating the mean and standard deviation of each parameter data under normal conditions. The mean calculation formula is: Among them, μ represents the mean, N represents the number of samples, and x i Represents the i-th sample value, and the standard deviation calculation formula is Among them, σ represents the standard deviation, μ represents the mean, N represents the number of samples, and x i Represents the i-th sample value. The standard deviation reflects the degree of dispersion of the parameter value around the mean, that is, the fluctuation range of the parameter. The fluctuation range of the parameter is an interval centered on the mean and with a certain multiple of the standard deviation as the radius.

[0023] Preferably, the specific method for segmenting the fault-related interval is:

[0024] S301, dynamically adjusting the time span based on the collected system fault data. When the fault duration is long, the time span is expanded, i.e., the fault correlation interval is expanded. When the fault duration is short, the time span is shortened, i.e., the fault correlation interval is shortened.

[0025] S302, segmenting the fault correlation interval according to the adjusted fault correlation interval into a rapid change segment, a stable segment, and a recovery segment;

[0026] S303, calculating impact correlation parameters according to the rapid change segment, stable segment, and recovery segment in the fault correlation interval;

[0027] S304: Based on the segmented fault correlation intervals and impact correlation parameters, the fault location is located and predicted.

[0028] Preferably, the fault-related interval is divided into a rapid change segment, a stable segment and a recovery segment. In the rapid change segment, the parameter change rate increases and exceeds the high change rate threshold, causing the parameter to deviate from the normal value and the fluctuation amplitude to increase; the stable segment is when the parameter change rate is less than the change rate threshold and the fluctuation amplitude is less than the fluctuation size threshold; the recovery segment refers to the process in which the system parameters begin to gradually return to normal values.

[0029] One or more technical solutions provided in this application have at least the following technical effects or advantages: by refining the thermal power generation system into a network structure and adjusting the fault-related parameters, these parameters can capture the nonlinear propagation path of the fault in the system, more comprehensively consider the interactions and dependencies between components in the system, especially non-sequential, concurrent, and hidden fault factors, and more accurately reflect the propagation mechanism and impact range of the fault in the system, thereby improving the accuracy and efficiency of fault location, reducing fault troubleshooting time, and lowering maintenance costs;

[0030] By dynamically calculating and setting time spans differently, the fluctuations of key parameters before and after a fault can be captured more accurately, avoiding errors caused by excessively large or small correlation intervals. The time span can be flexibly adjusted for different types of faults, adapting to various complex situations and improving the accuracy and efficiency of fault analysis.

[0031] Through segmented processing, the accuracy of fault location is significantly improved, the positioning error range is reduced, and the accuracy of fault prediction is improved based on the impact correlation parameters calculated in segments. Accurate fault location and effective preventive measures reduce unnecessary maintenance work and cost expenditures, and improve the accuracy of fault location and the effectiveness of fault prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flow chart of a safety inspection method for a smart power plant according to the present invention;

[0033] Figure 2 A schematic diagram of a process for dynamically adjusting fault-related intervals according to an embodiment of the present invention;

[0034] Figure 3 The figure is a flow chart of segmenting fault-related intervals according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] To facilitate understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0036] It should be noted that the terms “vertical”, “horizontal”, “up”, “down”, “left”, “right” and similar expressions used in this document are for illustrative purposes only and do not represent the only implementation method.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains; the terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more of the associated listed items.

[0038] Example 1: Figure 1 The figure is a flow chart of a safety inspection method for a smart power plant according to an embodiment of the present invention, which includes the following steps:

[0039] S101, collecting historical operation fault information and classifying the collected historical operation fault information;

[0040] Furthermore, operation logs are extracted from the automated monitoring system of the power generation system, including fault alarm records and equipment status changes, and historical maintenance records are collected, including fault descriptions, repair measures, and replacement parts. Based on the operation logs and historical maintenance records, the time of occurrence, fault type, fault location, impact range, processing process, and recovery time of faults in the power generation system over the past year are collected. The collected operation fault information is classified according to fault type, fault location, and impact degree, and the classified fault information is classified and stored in a database.

[0041] S102, constructing a network structure of a power generation system by determining nodes and edges based on the equipment in the power plant;

[0042] Specifically, the equipment of a power plant includes generators, boilers, transformers, transmission lines, control systems, and auxiliary equipment. Each device is marked with an identifier, and the equipment of the power plant is used as a node in the network structure. Energy transmission is the steam generated by the boiler driving the generator, and the electricity generated by the generator is transmitted through the transformer and the transmission line. Signal control is the control system sending fuel supply instructions to the boiler, or sending start / stop signals to the generator. The physical connection is the connection between the boiler and the steam pipe, and the connection between the transformer and the transmission line. Through energy transmission, signal control, and physical connection between nodes, edges are drawn between nodes to represent direct and indirect connections between nodes. Edges are directed (indicating the direction of energy or information flow) or undirected (indicating bidirectional interaction).

[0043] All nodes and edges are combined together according to their interaction relationships to form a network structure of a power generation system, and the network structure model is displayed using the graphical tool Visio.

[0044] S103, calculating the node degrees and edge weights based on the determined nodes and edges, simulating fault propagation based on the node degrees and edge weights in the network structure, and calculating the propagation speed and impact range;

[0045] Furthermore, each node in the network structure model is traversed, and the number of edges directly connected to the node is counted. This value is the degree of the node. A node degree threshold is set. When the degree of a node is greater than the preset node degree threshold, the node is regarded as a hub node, and the weight of the edges in the network structure is calculated based on energy transmission and expert advice.

[0046] The fault source node is set in the network structure model. The fault source node can be any device in the network, such as a generator, boiler, or transformer. These nodes will serve as the fault starting point in the simulation. After selecting the source node, the fault type is determined, including device failure, signal interruption, or physical connection break. Before the simulation begins, the simulation environment is initialized, including setting simulation parameters and initializing the network state. Based on the network structure model and edge weights, the process of fault propagation from the source node along the edges to other nodes is simulated. During the simulation, the speed, direction, and attenuation of fault propagation are considered. At the same time, the impact of redundancy and backup mechanisms in the network on fault propagation is also considered. Nodes or edges have backup or alternative paths. When the primary path fails, the fault will continue to propagate through these backup or alternative paths. During the simulation, the fault propagation status is recorded in real time, including the nodes to which the fault propagates and the propagation time or number of steps. Based on the recorded data during the simulation, the time or number of steps required for the fault to propagate from the source node to other nodes is calculated, and this time or number of steps is used as a measure of the fault propagation speed. All nodes in the network structure are traversed, the number of nodes affected by the fault during the simulation is counted, and the area of ​​the fault-affected area is calculated.

[0047] Based on the system characteristics and analysis requirements, set a safety threshold for the fault propagation speed and a maximum acceptable value for the fault impact range. When the fault propagation speed exceeds the safety threshold, the system faces serious risks or unacceptable performance degradation. When the fault impact range exceeds the maximum acceptable value, the system cannot operate normally.

[0048] S104, adjusting fault-related parameters based on historical operating fault information and network structure, where the fault-related parameters include node degree, edge weight, propagation speed, and impact range;

[0049] Specifically, the historical operation fault information is divided according to the start-up time sequence of the production process system in the power generation process, and the node degree is adjusted. The formula is: new node degree = original node degree × adjustment coefficient, where the original node degree is the original degree of the node in the network, and the adjustment coefficient is a number greater than 1, which is used to increase the weight of the high node degree. If the adjustment coefficient is 1.2, the new node degree will be 1.2 times the original node degree. The adjustment coefficient is obtained through expert experience and historical fault information; the edge weight is adjusted. The formula is: new edge weight = original edge weight + fault propagation frequency × adjustment factor, where the original edge weight is the original weight of the edge, the fault propagation frequency is the frequency of fault propagation on this edge, and the adjustment factor is a coefficient used to adjust the impact of the fault propagation frequency on the edge weight; the propagation speed is adjusted The formula is: New Propagation Speed ​​= Original Propagation Speed ​​× (Actual Propagation Speed ​​ / Simulated Propagation Speed), where the original propagation speed is the initial propagation speed set in the simulation, the actual propagation speed is the propagation speed observed when the fault occurs in the actual network, and the simulated propagation speed is the propagation speed originally set in the simulation. To adjust the impact range, the formula is: New Impact Range = Original Impact Range × (Actual Impact Range Coefficient / Simulated Impact Range Coefficient), where the original impact range is the initial impact range calculated in the simulation, the actual impact range coefficient is the coefficient of the actual impact range of the fault derived based on actual data or experience, and the simulated impact range coefficient is the impact range coefficient originally set in the simulation. If the fault propagation simulation results after adjustment deviate significantly from the actual situation, return to the original impact range and continue adjusting the parameters.

[0050] S105, collecting fault time nodes, generating fault correlation intervals according to the fault time nodes, determining impact correlation parameters based on the fault correlation intervals, constructing a multi-level data model according to the fault correlation parameters and the impact correlation parameters, inputting the monitoring data of the power plant into the multi-level data model, and obtaining an operation safety warning result.

[0051] Furthermore, in the thermal power generation process, each functional equipment and system is equipped with a parameter measurement device for real-time recording of operating data, and issuing an alarm when the operating data deviates from the preset safe operating value. The fault time node is the time when a safety accident occurred in the history of the current thermal power plant. Taking the fault time node as the center, a certain time span is pushed forward and backward to obtain the fault-related interval. The fault-related time interval includes the preventive maintenance stage where equipment abnormalities may cause fault abnormalities, the early warning and emergency repair stage where the accumulation of fault factors leads to the imminent occurrence or occurrence of a safety accident, and the stage of eliminating fault factors and accident hazards after the safety failure occurs. All parameter fluctuations within the fault-related time interval may cause a chain reaction, and all parameter fluctuations may lead to failures or safety accidents.

[0052] Based on the fault correlation interval, the parameters are divided into a set of influencing correlation parameters before the fault occurs, a set of influencing correlation parameters at the critical moment of fault occurrence and the moment of fault occurrence, and a set of influencing correlation parameters after the fault occurs. The weight of each influencing correlation parameter is obtained according to the risk factor and importance of the production process system to which the faulty equipment belongs in the thermal power generation process. A multi-level data model is constructed based on the fault correlation parameters and the set of influencing correlation parameters. The monitoring data of the power plant is input into the multi-level data model to obtain the operation safety early warning results.

[0053] The technical solutions in the above-mentioned embodiments of the present application have at least the following technical effects or advantages: by concretizing the thermal power generation system into a network structure and adjusting the fault-related parameters, these parameters can capture the nonlinear propagation path of the fault in the system, and can more comprehensively consider the interactions and dependencies between the components in the system, especially non-sequential, concurrent and hidden fault factors, and can more accurately reflect the propagation mechanism and impact range of the fault in the system, thereby improving the accuracy and efficiency of fault location, reducing the troubleshooting time, and reducing maintenance costs.

[0054] Example 2: Based on the determination of the fault correlation interval based on the time span in Example 1, the selection of the time span may be too subjective, resulting in the correlation interval being too large or too small, and unable to accurately capture the fluctuations of key parameters before and after the fault. This embodiment dynamically adjusts the fault correlation interval, which can accurately capture the fluctuations of key parameters before and after the fault. When generating the fault correlation interval, the corresponding time span is selected according to the fault type.

[0055] like Figure 2 As shown in the figure, the specific steps for dynamically adjusting the fault correlation interval are as follows:

[0056] S201, based on historical fault data, collecting data fluctuation characteristics;

[0057] Furthermore, by calculating the mean and standard deviation of each parameter data under normal conditions, the normal fluctuation range of the parameter is established. The mean calculation formula is: Among them, μ represents the mean, N represents the number of samples, and x i Represents the i-th sample value, and the standard deviation calculation formula is Among them, σ represents the standard deviation, μ represents the mean, N represents the number of samples, and x i represents the i-th sample value. The standard deviation reflects the degree of dispersion of the parameter value around the mean, that is, the fluctuation range of the parameter. The fluctuation range of the parameter is an interval centered on the mean and with a certain multiple of the standard deviation as the radius. For example, if the mean of the parameter is 10 and the standard deviation is 2, the normal fluctuation range of the parameter is roughly between 8 and 12. If the value exceeds the fluctuation range, it indicates that a fault has occurred or is about to occur, thereby triggering further fault analysis or early warning mechanism.

[0058] Using time series analysis technology and long short-term memory networks, we can identify the changing trends and periodic fluctuations of parameters over time. We can use historical parameter data to train the selected time series model, learn the changing patterns of parameters over time, use the trained model to predict future parameter values, and evaluate the accuracy of the predictions. We can use clustering algorithms to classify parameter fluctuation patterns, identify specific fluctuation patterns related to specific fault types, extract features that can reflect the fluctuation patterns from the parameter data, and use the selected clustering algorithm to perform cluster analysis on the extracted features. The identified specific fluctuation patterns are verified and confirmed in combination with historical fault data and expert experience.

[0059] S202, dynamically calculating the time span based on the fluctuation characteristics of the data;

[0060] Specifically, based on the standard deviation calculated in step S201, twice the standard deviation of the data is used as the upper and lower bounds of the time span. For example, the standard deviation is σ, the current time is t, then the time span range is [t-2σ, t+2σ]. The time span is further optimized using a machine learning algorithm. The time span is optimized using a machine learning algorithm support vector machine (SVM) to construct features. The features include parameter values, statistical indicators (such as mean, standard deviation), and time series characteristics (such as trends and periodicity) within a preliminarily determined time span. The constructed features are input into the selected machine learning algorithm for model training. Through training, the model will learn how to predict the occurrence of faults based on the features, and find the time span that can best capture the fluctuations of key parameters before and after the fault.

[0061] S203, classifying the fault types into fast-propagating faults, slowly developing faults, and local impact faults, and setting differentiated time spans based on the classified fault types;

[0062] Furthermore, according to the impact range, propagation speed and duration of the fault, the fault types are classified into fast-propagation faults, slow-development faults and local-impact faults. Fast-propagation faults usually have a wide impact range and a fast propagation speed, and may have a significant impact on the entire system in a short period of time, for example, a short-circuit fault in the power system; slow-development faults usually develop slowly and may take a long time to gradually show their impact, for example, the wear or aging of mechanical parts; the impact range of local-impact faults is relatively small, and is usually limited to a certain device or component, for example, the failure of a single sensor. According to the fault classification results, a different time span is set for each type of fault. The time span of fast-propagation faults is set to 30 minutes to 1 hour, and the time span of slow-development faults is set to 5 minutes to 15 minutes, so as to monitor the changes in parameters more frequently. The appropriate time span for local-impact faults is set according to the characteristics of the specific fault and monitoring requirements. The setting of differentiated time spans is based on experience and data analysis.

[0063] S204: combining the dynamically calculated time span with the differentiated time span to form a final dynamic time span, and adjusting the fault correlation interval according to the dynamic time span;

[0064] Specifically, the time span dynamically calculated in step S202 is used as the basic time span. For fast-propagating faults, the propagation speed of this type of fault is fast. Time is added on the basis of the basic time span. By extending the time span, the propagation path and impact range of the fault can be more comprehensively understood. For slowly developing faults, the parameter changes are relatively smooth. By reducing the time span, the starting point and development trajectory of the fault can be more accurately located. Local impact faults are determined according to specific circumstances. If the fault impact range is small, a relatively short time span is used for monitoring and analysis. However, if the fault causes larger system problems, the time span is extended. According to the characteristics of the fault type and the differentiated adjustment strategy, the basic time span is adjusted to form an adjusted dynamic time span, and the fault association interval is adjusted according to the dynamic time span.

[0065] The technical solutions in the above-mentioned embodiments of the present application have at least the following technical effects or advantages: through dynamic calculation and differentiated setting of time spans, it is possible to more accurately capture the fluctuations of key parameters before and after the fault, avoid errors caused by too large or too small correlation intervals, and flexibly adjust the time span for different types of faults to adapt to various complex situations and improve the accuracy and efficiency of fault analysis.

[0066] Example 3: Based on the dynamic adjustment of the fault-related interval in Example 2, the dynamically adjusted fault-related interval cannot accurately predict the location of the subsequent fault and cannot locate the fault location. This example segments the fault-related interval and accurately locates the fault location. The segmented fault-related interval can provide a clearer view of the fault development process.

[0067] like Figure 3 As shown in Figure 2, the specific method for segmenting the fault-related interval is as follows:

[0068] S301, dynamically adjust the time span based on the collected system fault data. When the fault duration is long, expand the time span, that is, expand the fault correlation interval. When the fault duration is short, shorten the time span, that is, shorten the fault correlation interval.

[0069] S302, segmenting the fault correlation interval according to the adjusted fault correlation interval into a rapid change segment, a stable segment, and a recovery segment;

[0070] Specifically, the basis for segmenting the fault-related interval is the parameter's change rate and fluctuation size. The parameter's change rate reflects how fast the parameter changes. The faster change rate corresponds to the drastic change at the initial stage of the fault, while the slower change rate corresponds to the stable period or recovery period of the fault. The parameter's fluctuation size reflects the stability. The larger fluctuation part contains fault information, while the smaller fluctuation part represents the normal state of the system or the stable state of the fault. A high change rate threshold and a low change rate threshold are set. The high change rate threshold is used to identify time periods where the parameters change extremely drastically, that is, those parts where the parameter values ​​change significantly in a short period of time. The low change rate threshold is used to define the interval where the parameter changes are relatively stable. A fluctuation size threshold is set. The fluctuation size threshold is used to measure the amplitude of the parameter fluctuation, that is, the range of fluctuations in the parameter value within a certain time period.

[0071] The fault-related interval is divided into a rapid change segment, a stable segment, and a recovery segment. The rapid change segment corresponds to the initial stage of a system fault or when the fault escalates, when the parameter change rate increases sharply and exceeds the high change rate threshold. During this stage, the system is subject to strong external interference, causing key parameters to rapidly deviate from normal values ​​and increase the fluctuation amplitude. The stable segment occurs when the parameter change rate drops below the low change rate threshold and the fluctuation amplitude is less than the fluctuation size threshold. The system enters the stable segment. This stage occurs for a period of time after the fault occurs. The system temporarily reaches a relatively stable state. Although it still deviates from the normal operating state, the parameter changes are no longer as drastic. Identifying the stable segment helps analyze the scope of the fault impact and evaluate the current state of the system. The recovery segment refers to the process in which system parameters begin to gradually recover or approach normal values. The judgment of this stage is mainly based on whether the parameter value is close to or has reached the preset normal operating range. The identification of the recovery segment indicates the effectiveness of the system fault mitigation or repair work and is an important indicator for evaluating the system recovery capability and maintenance effect.

[0072] The rapid change segment usually contains key information at the initial stage of the fault, such as the time when the fault occurred and the initial fault impact; the stable segment provides data during the stable period of the fault, which helps to analyze the cause and mechanism of the fault; the recovery segment reflects the recovery process of the fault, which helps to evaluate the effectiveness of fault handling measures and predict system recovery time.

[0073] S303, calculating impact correlation parameters according to the rapid change segment, stable segment, and recovery segment in the fault correlation interval;

[0074] Furthermore, relevant parameter data is extracted from the rapidly changing segment of the fault-related interval. A high-sensitivity algorithm (short-time window moving average) with fast-changing characteristics is selected. The short-time window moving average algorithm is used to calculate the parameter change rate and fluctuation amplitude within the rapidly changing segment. Based on the characteristics of the algorithm output, combined with system characteristics and historical fault data, the initial fault impact parameter P1 is calculated. P1 represents the direct impact of the fault on the system in the early stage of the fault, such as the degree of equipment damage and the proportion of system performance degradation. Relevant parameter data is extracted from the stable segment of the fault-related interval. The median filter algorithm is used to calculate the mean and standard deviation of the parameters within the stable segment. Combined with the performance indicators of the system during stable operation and the fault impact model, the results of the median filter algorithm are used to calculate P2. P2 reflects the magnitude of the fault's continued impact on the system during the stable period, including the percentage of system efficiency reduction and the duration of the fault. Relevant parameter data is extracted from the recovery segment of the fault-related interval. Trend analysis is used to calculate the recovery rate and recovery time of the parameter during the parameter recovery process. Combined with the performance indicators of the system during recovery and the fault impact model, the results of the recovery evaluation algorithm are used to calculate P3. P3 represents the impact on system performance after recovery, such as the performance loss and recovery efficiency after recovery.

[0075] S304, locating and predicting the fault location based on the segmented fault correlation intervals and impact correlation parameters;

[0076] Specifically, real-time data or historical data of P1 (initial fault influencing parameter), P2 (stable period fault influencing parameter) and P3 (recovery period fault influencing parameter) are obtained from the monitoring system, and the values ​​of P1, P2, and P3 at different time points are plotted on a scatter plot. The trend of P1, P2, and P3 changing over time is observed based on the drawn scatter plot, and prominent parameter values ​​are identified based on the changing trend. The prominent parameter values ​​are the sudden increase, decrease, or fluctuation of the parameters. A known fault model is established based on the historical fault data, and the calculated P1, P2, and P3 parameter characteristics are compared with the characteristics in the fault model. According to the comparison results, the fault type and fault location are determined, the probability of the fault location is calculated through a Bayesian network, and a distribution map is drawn according to the calculated probability.

[0077] Observe the changing trends of P1, P2, and P3 over time, and combine the historical data and current status of system operation to predict the possible future development direction of the fault.

[0078] The technical solutions in the above-mentioned embodiments of the present application have at least the following technical effects or advantages: through segmented processing, the accuracy of fault location is significantly improved, the positioning error range is narrowed, the accuracy of fault prediction is improved based on the impact correlation parameters calculated by segmentation, and accurate fault location and effective preventive measures reduce unnecessary maintenance work and cost expenditure, thereby improving the accuracy of fault location and the effectiveness of fault prediction.

[0079] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Various modifications and variations are readily apparent to those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A safety inspection method for a smart power plant, characterized in that: include: S101, collecting historical operation fault information and classifying the collected historical operation fault information; S102, constructing a network structure of a power generation system by determining nodes and edges based on the equipment in the power plant; S103, calculating the node degrees and edge weights based on the determined nodes and edges, simulating fault propagation based on the node degrees and edge weights in the network structure, and calculating the propagation speed and impact range; S104, adjusting fault-related parameters based on historical operating fault information and network structure, where the fault-related parameters include node degree, edge weight, propagation speed, and impact range; S105, collecting fault time nodes, generating fault correlation intervals according to the fault time nodes, determining impact correlation parameters based on the fault correlation intervals, building a multi-level data model according to the fault correlation parameters and the impact correlation parameters, inputting the monitoring data of the power plant into the multi-level data model, and obtaining safety warning results.

2. A safety inspection method for a smart power plant according to claim 1, characterized in that: Traverse each node in the network structure and count the number of edges directly connected to the node. The value of the edge number is the degree of the node. Connect two nodes with edges, and the edges represent the connection between the nodes.

3. The safety inspection method for a smart power plant according to claim 1, characterized in that: The historical operation fault information is divided according to the startup time sequence of the production process system in the power generation process, and the node degree is adjusted. The formula is: new node degree = original node degree × adjustment coefficient, where the original node degree is the original degree of the node in the network, and the adjustment coefficient is a number greater than 1, which is used to increase the weight of high node degrees. The adjustment coefficient is derived from expert experience and historical fault information.

4. The safety inspection method for a smart power plant according to claim 1, characterized in that: The edge weight is adjusted using the following formula: New edge weight = Original edge weight + Fault propagation frequency × Adjustment factor, where Original edge weight is the original weight of the edge, Fault propagation frequency is the frequency at which faults propagate along this edge, and Adjustment factor is a coefficient used to adjust the impact of Fault propagation frequency on the edge weight.

5. The safety inspection method for a smart power plant according to claim 1, characterized in that: The propagation speed is adjusted using the formula: New propagation speed = Original propagation speed × (Actual propagation speed / Simulated propagation speed), where the Original propagation speed is the initial propagation speed set in the simulation, the Actual propagation speed is the propagation speed observed in the actual network, and the Simulated propagation speed is the propagation speed originally set in the simulation.

6. The safety inspection method for a smart power plant according to claim 1, characterized in that: The impact range is adjusted according to the following formula: New impact range = original impact range × (actual impact range coefficient / simulated impact range coefficient), where the original impact range is the initial impact range calculated in the simulation, the actual impact range coefficient is the coefficient of the actual impact range of the fault obtained based on the actual data, and the simulated impact range coefficient is the impact range coefficient originally set in the simulation.

7. The safety inspection method for a smart power plant according to claim 1, characterized in that: Steps for dynamically adjusting the fault-related interval: S201, based on historical fault data, collecting data fluctuation characteristics; S202, dynamically calculating the time span based on the fluctuation characteristics of the data; S203, classifying the fault types into fast-propagating faults, slowly developing faults, and local impact faults, and setting differentiated time spans based on the classified fault types; S204 , combining the dynamically calculated time span with the differentiated time span to form a final dynamic time span, and adjusting the fault correlation interval according to the dynamic time span.

8. The safety inspection method for a smart power plant according to claim 7, characterized in that: By calculating the mean and standard deviation of each parameter data under normal conditions, the normal fluctuation range of the parameter is established. The mean calculation formula is: Among them, μ represents the mean, N represents the number of samples, and x i Represents the i-th sample value, and the standard deviation calculation formula is σ Among them, σ represents the standard deviation, μ represents the mean, N represents the number of samples, and x i Represents the i-th sample value. The standard deviation reflects the degree of dispersion of the parameter value around the mean, that is, the fluctuation range of the parameter. The fluctuation range of the parameter is an interval centered on the mean and with a certain multiple of the standard deviation as the radius.

9. The safety inspection method for a smart power plant according to claim 7, characterized in that: Specific method for segmenting fault-related intervals: S301, dynamically adjusting the time span based on the collected system fault data. When the fault duration is long, the time span is expanded, i.e., the fault correlation interval is expanded. When the fault duration is short, the time span is shortened, i.e., the fault correlation interval is shortened. S302, segmenting the fault correlation interval according to the adjusted fault correlation interval into a rapid change segment, a stable segment, and a recovery segment; S303, calculating impact correlation parameters according to the rapid change segment, stable segment, and recovery segment in the fault correlation interval; S304: Based on the segmented fault correlation intervals and impact correlation parameters, the fault location is located and predicted.

10. A safety inspection method for a smart power plant according to claim 9, characterized in that: The fault-related interval is divided into a rapid change segment, a stable segment, and a recovery segment. In the rapid change segment, the parameter change rate increases and exceeds the high change rate threshold, causing the parameter to deviate from the normal value and the fluctuation amplitude to increase. The stable segment is when the parameter change rate is less than the change rate threshold and the fluctuation amplitude is less than the fluctuation size threshold. The recovery segment refers to the process in which the system parameters begin to gradually return to normal values.

Images