A data analysis system and method based on a maintenance progress network diagram

Abstract

This invention provides a data analysis system and method based on a maintenance progress network diagram, relating to the fields of maintenance status assessment and task scheduling. The system divides the turbine into several functional zones, forming standardized maintenance task packages, each task package serving as a node in the maintenance progress network diagram. It acquires historical data of the refined equipment components and analyzes the impact characteristics of different components. Based on these impact characteristics, it determines whether an abnormality has occurred in the equipment component according to real-time impact characteristic values. For the components identified as abnormal, it performs cross-correlation analysis to obtain the influence relationships between them and calculates the time lag of the influence. This time lag is then used as a progress indicator in the maintenance progress network diagram for maintenance. By introducing cross-correlation analysis to calculate the influence relationships and time lags between equipment components and using the time lag as a maintenance priority indicator, this application prioritizes the repair of fault sources and prevents the spread of abnormalities.

Description

Technical Field

[0001] This invention relates to the field of maintenance status assessment and task scheduling technology, specifically to a data analysis system and method based on a maintenance progress network diagram. Background Technology

[0002] As the core equipment of a hydropower station, the operating status of the turbine directly affects the power generation efficiency and operational safety of the entire station. Due to its complex structure, the turbine typically consists of multiple functional zones, each containing numerous precision equipment components. During long-term operation, these components are subject to various factors such as water flow impact, mechanical wear, vibration, and temperature changes, which can lead to performance degradation or potential malfunctions.

[0003] Traditional maintenance plans rely on experience-based scheduling, which can lead to problems such as conflicting maintenance resources, unclear critical path identification, and inaccurate schedule forecasting. This can easily result in long-term equipment downtime or safety hazards. Furthermore, there are interrelationships between the equipment components being maintained. Therefore, there is an urgent need for a data analysis system and method based on maintenance schedule network diagrams to prioritize the maintenance of source equipment components and avoid large-scale damage to equipment components. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a data analysis system and method based on maintenance progress network diagram, so as to solve the problems raised in the prior art.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A data analysis method based on a maintenance schedule network diagram includes the following steps: The turbine is divided into several functional zones, and the equipment in each zone is further subdivided to form standardized maintenance task packages. Each task package serves as a node in the maintenance progress network diagram. Obtain historical data of the refined and segmented equipment components, and analyze the operating parameter values ​​under normal and abnormal conditions to obtain the impact characteristics of different equipment components; Based on the obtained impact characteristics of different equipment components, it is determined whether the equipment components are abnormal according to the real-time impact characteristic values; Cross-correlation analysis is performed on the equipment components that are judged to be abnormal to obtain the influence relationship between the equipment components, and the time delay of the influence is calculated. The time delay is used as the progress index of the maintenance progress network diagram for maintenance. It also includes a secondary assessment of anomalies in equipment components by influencing the volatility of eigenvalues ​​in real time.

[0006] The above-mentioned division of the turbine into several functional zones, further subdividing the equipment within each zone into standardized maintenance task packages, with each task package serving as a node in the maintenance progress network diagram, specifically: The turbine is divided into several functional zones according to its function. The functional zones include the main body, control system, sealing system, and protection system. The devices in each functional area are further broken down; Based on the detailed breakdown of equipment, standardized maintenance task packages are formed. Each task includes the name of the equipment component to be maintained, the type of test parameter, and the default range of the test parameter. Each standardized maintenance task package is treated as a node in the maintenance schedule network diagram, and a digital twin maintenance schedule network diagram is constructed.

[0007] The process of obtaining historical data for the refined and segmented equipment components, and analyzing the operating parameter values ​​under normal and abnormal conditions, yields the impact characteristics of different equipment components. This process specifically includes the following steps: Step s1-1: For any device component, obtain several historical operating parameter values, and classify and extract the operating parameter values ​​when normal operation occurs and those when abnormal operation occurs. For operating parameter type x, the parameter value of x when normal operation is recorded as [T]. x1 ,T x2 ,…,T xn When an anomaly occurs, the parameter value of x is denoted as [F]. x1 ,F x2 ,…,F xn ]; among which, T x1 ,T x2 ,…,T xn This represents the 1st, 2nd, ..., nth parameter value of the normal operating parameter x; F x1 ,F x2 ,…,F xn This represents the 1st, 2nd, ..., nth parameter value of the runtime parameter x when an exception occurs; Steps s1-2: Combine the parameter values ​​of the running parameter type x when it is normal and the parameter values ​​when it is abnormal into a numerical variable, and the set of numerical variables X = [T]. x1 ,T x2 ,…,T xn ,F x1 ,F x2 ,…,F xn The normal and abnormal states of the equipment are quantified into binary variables Y, with Y taking the value of 0 for the normal state and 1 for the abnormal state. The binary variables corresponding to the parameter values ​​of the operating parameter type x when they are normal and when they are abnormal are merged, and the binary variable set Y=[0,…,0,1,…,1] is formed. Among them, there are 2n values ​​in [0,…,0,1,…,1]. Steps s1-3: Using the numerical variable X and the binary variable Y representing the equipment status as a basis, perform correlation analysis, calculate the correlation coefficient, and characterize it as follows: Where r represents the correlation coefficient between the operating parameter type x and the device status; This represents an element in the set of numerical variables X, where i is the element identifier, and i∈[1,2n]. Represents the elements in the binary variable set Y; Represents the average value of a numerical variable; The average value of the binary variable Y representing the equipment status; Steps s1-4: Calculate the correlation coefficients between different operating parameter types and equipment status according to the method in step s3, and determine the operating parameter types with absolute values ​​of correlation coefficients greater than preset correlation thresholds as influencing features.

[0008] The above methods also include: Based on the obtained impact characteristics, the real-time impact characteristic value corresponding to any device component is obtained. According to the preset standard operating range [a,b], the device components whose real-time impact characteristic value exceeds the standard operating range [a,b] are marked, and the number of markings z within the preset time period T is recorded. The device components whose number of markings z is greater than the preset first threshold are marked as abnormal and real-time warnings are issued.

[0009] The above-mentioned cross-correlation analysis of the equipment components judged to be abnormal is used to obtain the influence relationship between the equipment components and calculate the time delay of the influence. The time delay is used as the schedule index of the maintenance schedule network diagram for maintenance. The specific steps include the following: Step s2-1: Perform sequence sampling of the influence feature values ​​of different equipment components marked as abnormal. For different equipment components p and q, the sequence sampling results p(t) and q(t) of the influence feature values ​​are expressed as: p(t) = [p1, p2, ..., p m ] and q(t)=[q1,q2,…,q m ]; where m represents the number of sequence samples affecting the feature value, and m is a positive integer; t represents the time series identifier; Step s2-2: Calculate the cross-correlation coefficient r at lag k based on the sequence sampling results p(t) and q(t) that affect the eigenvalues. pq (k), Where k represents the lag index, k is an integer, k∈[-L,L], L represents the maximum lag, L<m; and Let p(t) and q(t) represent the elements in the sequence sampling results that affect the feature values, respectively, and j represent the element identifier, j∈[1,m]; This represents the average value of the sequence sampling result p(t) that affects the feature value; This represents the average value of the sequence sampling results q(t) that affect the feature values; Steps s2-3: For each lag index k, calculate the cross-correlation number using the method in step s2-2. For a preset cross-correlation threshold, different device components that are greater than the preset cross-correlation threshold are considered to have mutual influence. Steps s2-4: Arrange the cross-correlation coefficients greater than the preset cross-correlation threshold in ascending order, and record the corresponding time delay α, α=|k| Δt; where Δt represents the sampling interval; the time delay α is used as the progress indicator of the maintenance progress network diagram, and maintenance is carried out according to the time delay α.

[0010] The above methods also include: Based on the obtained impact characteristics, the real-time impact characteristic value corresponding to any device component is acquired. For device components whose real-time impact characteristic value is within the preset standard operating range [a,b], the real-time impact characteristic value [d1,d2,…,d] within a preset time period T is collected. l ]; where d1, d2, ..., d l This represents the 1st, 2nd, ..., lth real-time influencing feature values ​​collected, where l = T / f; f represents the sampling frequency. Calculate the volatility δ of the real-time influence on the eigenvalue within a preset time period T: Where μ represents the average value of the real-time impact feature within the time period T; j represents the identifier of the real-time impact feature, j∈[1,l]; Device components whose real-time volatility δ affecting the characteristic value exceeds a preset second threshold within a preset time period T are identified as abnormal and real-time warnings are issued.

[0011] The system using the above-mentioned data analysis method based on maintenance progress network diagram includes a partitioned data acquisition module, a feature selection module, an anomaly judgment module, and a maintenance progress network diagram module. The partition data acquisition module is used to divide the turbine into several functional partitions according to the equipment functions, further break down the equipment in each functional partition, and collect the historical and real-time operating parameter values ​​of different equipment components after the detailed breakdown. The feature selection module is used to obtain the impact characteristics of different equipment components by analyzing the historical operating parameter values ​​of the refined and segmented equipment components under normal and abnormal conditions. The anomaly detection module is used to determine whether a device component is abnormal based on the obtained impact characteristics of different device components and the real-time impact characteristic value. The maintenance progress network diagram module forms standardized maintenance task packages based on the detailed breakdown of equipment. Each task package includes the name of the maintenance equipment component, the type of detection parameter, and the qualified range of the preset detection parameters. Each standardized maintenance task package is used as a node in the maintenance progress network diagram. The time delay α is used as the progress indicator of the maintenance progress network diagram to construct a digital twin maintenance progress network diagram. Maintenance is carried out according to the maintenance progress network diagram.

[0012] The aforementioned anomaly detection module includes a detection unit and a secondary detection unit; The judgment unit, based on the real-time impact characteristic values ​​of different device components and according to the preset standard operating range [a,b], marks the device components whose real-time impact characteristic values ​​exceed the standard operating range [a,b], records the number of markings z within a preset time period T, and records the device components whose number of markings z is greater than a preset first threshold as abnormal. The secondary judgment unit records equipment components whose real-time impact on the characteristic value is greater than a preset second threshold within a preset time period T based on the volatility of the characteristic value within the preset time period T.

[0013] The system also includes an early warning device that provides real-time warnings for devices marked as abnormal. The early warning device is connected to the anomaly judgment module. When the anomaly judgment module determines that a device component is abnormal, the early warning device immediately triggers a real-time warning. The warning information includes the name of the abnormal device component, the anomaly judgment type, the real-time impact characteristic value / volatility data, and the corresponding threshold range.

[0014] The aforementioned maintenance progress network diagram module also includes a maintenance scheduling unit. The maintenance scheduling unit prioritizes the maintenance of abnormal equipment components according to the time delay α, generates maintenance scheduling instructions, and updates the maintenance status of each node in the maintenance progress network diagram in real time, thereby realizing visualized monitoring and dynamic scheduling of the maintenance process.

[0015] The aforementioned partitioned data acquisition module includes a historical data storage unit and a real-time data acquisition unit. The historical data storage unit is used to store and classify the historical operating parameter values ​​of the device components throughout their entire lifecycle. The real-time data acquisition unit collects the operating parameters of the device components in real time through various sensors, and the acquisition frequency can be flexibly adjusted according to the importance of the device components.

[0016] The aforementioned feature selection module incorporates a correlation analysis algorithm and a threshold setting unit. The threshold setting unit can dynamically adjust relevant thresholds based on changes in equipment operating conditions and historical data updates to ensure the accuracy and adaptability of feature extraction.

[0017] The data analysis system and method based on maintenance progress network diagram mentioned in this invention have the following advantages: This application achieves standardized, structured, and visualized management of maintenance tasks by breaking down the turbine into functional zones, forming standardized maintenance task packages, and constructing a digital twin maintenance progress network diagram. It utilizes historical data analysis to extract key influencing features, combines real-time operating parameters with preset ranges for anomaly detection, and improves the accuracy and comprehensiveness of anomaly identification through a dual judgment mechanism of marking frequency and volatility. Furthermore, it introduces cross-correlation analysis to calculate the influence relationships and time delays between equipment components, using time delay as a maintenance priority indicator to prioritize the repair of fault sources and prevent the spread of anomalies. This solution not only achieves real-time monitoring and rapid early warning but also significantly shortens the maintenance cycle, reduces resource waste and maintenance costs, and improves the operational safety and reliability of the turbine. Attached Figure Description

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a schematic diagram of the method flow for the maintenance data storage method based on encryption technology of the present invention. Detailed Implementation

[0019] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0020] Example: like Figure 1 As shown, the present invention provides a technical solution, a data analysis method based on a maintenance progress network diagram, which specifically includes the following steps: The turbine is divided into several functional zones, and the equipment in each zone is further subdivided to form standardized maintenance task packages. Each task package serves as a node in the maintenance progress network diagram. Furthermore, the turbine is divided into several functional zones, and the equipment in each zone is further subdivided to form standardized maintenance task packages. Each task package serves as a node in the maintenance progress network diagram; specifically: The turbine is divided into several functional zones according to its function. The functional zones include the main body, control section, sealing section and protection section. The devices in each functional area are further broken down; Based on the detailed breakdown of equipment, standardized maintenance task packages are formed. Each task package includes the name of the equipment component to be maintained, the type of test parameter, and the default range of the test parameter. Each standardized maintenance task package is treated as a node in the maintenance schedule network diagram, and a digital twin maintenance schedule network diagram is constructed.

[0021] In this embodiment, the functional areas include a main body, a control section, a sealing section, and a protection section. The equipment in each area is further subdivided into standardized maintenance task packages. The main body includes the runner, main shaft, and blades. The main body is the core component of the turbine, composed of several blades that directly interact with the water flow, converting the water's energy into rotational mechanical energy. The blades regulate the direction and flow rate of the water entering the runner, controlling power and efficiency. The control section uniformly introduces water into the runner from the pressure pipe and includes an inlet pipe, a tailrace pipe, and a governor. The governor automatically adjusts the guide vane opening according to load changes to maintain stable rotational speed. The sealing section includes bearings and sealing devices. The sealing devices prevent water from entering the bearings or the machine base along the shaft. The protection section includes a speed control system hydraulic device, a cooling and lubrication system, and a measurement and monitoring system. The measurement and monitoring system includes sensors for speed, vibration, water pressure, and temperature.

[0022] Obtain historical data of the refined and segmented equipment components, and analyze the operating parameter values ​​under normal and abnormal conditions to obtain the impact characteristics of different equipment components; Furthermore, by acquiring historical data of the refined and segmented equipment components and analyzing the operating parameter values ​​under normal and abnormal conditions, the impact characteristics of different equipment components are obtained, specifically: Step s1-1: For any device component, obtain several historical operating parameter values, and classify and extract the operating parameter values ​​when normal operation occurs and those when abnormal operation occurs. For operating parameter type x, the parameter value of x when normal operation is recorded as [T]. x1 ,T x2 ,…,T xn When an anomaly occurs, the parameter value of x is denoted as [F]. x1 ,F x2 ,…,F xn ]; among which, T x1 ,T x2 ,…,T xn This represents the 1st, 2nd, ..., nth parameter value of the normal operating parameter x; F x1 ,F x2 ,…,F xn This represents the 1st, 2nd, ..., nth parameter value of the runtime parameter x when an exception occurs; Steps s1-2: Combine the parameter values ​​of the running parameter type x when it is normal and the parameter values ​​when it is abnormal into a numerical variable, and the set of numerical variables X = [T]. x1 ,T x2 ,…,T xn ,F x1 ,Fx2 ,…,F xn The normal and abnormal states of the equipment are quantified into binary variables Y, with Y taking the value of 0 for the normal state and 1 for the abnormal state. The binary variables corresponding to the parameter values ​​of the operating parameter type x when they are normal and when they are abnormal are merged, and the binary variable set Y=[0,…,0,1,…,1] is formed. Among them, there are 2n values ​​in [0,…,0,1,…,1]. Steps s1-3: Using the numerical variable X and the binary variable Y representing the equipment status as a basis, perform correlation analysis, calculate the correlation coefficient, and characterize it as follows: Where r represents the correlation coefficient between the operating parameter type x and the device status; This represents an element in the set of numerical variables X, where i is the element identifier, and i∈[1,2n]. Represents the elements in the binary variable set Y; Represents the average value of a numerical variable; The average value of the binary variable Y representing the equipment status; Steps s1-4: Calculate the correlation coefficients between different operating parameter types and equipment status according to the method in step s3, and determine the operating parameter types with absolute values ​​of correlation coefficients greater than preset correlation thresholds as influencing features.

[0023] It should be noted that in this embodiment, the normal state Y is set to 0, and the abnormal state Y is set to 1. Therefore, r>0 indicates that the parameter value is positively correlated with the abnormal state, and r<0 indicates that the parameter value is negatively correlated with the abnormal state. The closer |r| is to 1, the stronger the correlation; the closer |r| is to 0, the weaker the correlation. Therefore, the operating parameter type with an absolute value of the correlation coefficient greater than the preset correlation threshold is determined as the influencing feature. The method of setting the correlation threshold is not limited. Due to different calculation levels in practice, if the correlation threshold is set high, there will be fewer influencing features, which can reduce the amount of calculation. However, in the equipment anomaly assessment, the result will have a large error. Conversely, if the correlation threshold is set low, the amount of calculation will increase significantly. If the correlation threshold is set very low, overfitting may occur. Therefore, in actual implementation, the threshold is set according to the actual calculation level. In this embodiment, a normal distribution is approximated by a large amount of historical data, and the correlation threshold is set according to the three standard deviations principle in the normal distribution.

[0024] Based on the obtained impact characteristics of different equipment components, it is determined whether the equipment components are abnormal according to the real-time impact characteristic values; Furthermore, based on the obtained impact characteristics of different equipment components, the system determines whether the equipment components are malfunctioning based on the real-time impact characteristic values; specifically: Based on the obtained impact characteristics, the real-time impact characteristic value corresponding to any device component is obtained. According to the preset standard operating range [a,b], the device components whose real-time impact characteristic value exceeds the standard operating range [a,b] are marked, and the number of markings z within the preset time period T is recorded. The device components whose number of markings z is greater than the preset first threshold are marked as abnormal and real-time warnings are issued.

[0025] In this embodiment, the preset time period T is one cycle of device operation. In actual situations, it can be set according to different devices, and there is no restriction here. Secondly, a first threshold is set according to historical data. In this embodiment, the method for setting the first threshold is to use the minimum number of times the device operation is abnormally collected within one cycle in history as the first threshold.

[0026] Cross-correlation analysis is performed on the equipment components that are judged to be abnormal to obtain the influence relationship between the equipment components, and the time delay of the influence is calculated. The time delay is used as the progress index of the maintenance progress network diagram for maintenance. Furthermore, cross-correlation analysis is performed on the equipment components identified as abnormal to obtain the influence relationships between the equipment components, and the time lag of the influence is calculated. The time lag is used as the progress index of the maintenance progress network diagram for maintenance. Specifically: Step s2-1: Perform sequence sampling of the influence feature values ​​of different equipment components marked as abnormal. For different equipment components p and q, the sequence sampling results p(t) and q(t) of the influence feature values ​​are expressed as: p(t) = [p1, p2, ..., p m ] and q(t)=[q1,q2,…,q m ]; where m represents the number of sequence samples affecting the feature value, and m is a positive integer; t represents the time series identifier; Step s2-2: Calculate the cross-correlation coefficient r at lag k based on the sequence sampling results p(t) and q(t) that affect the eigenvalues. pq (k), Where k represents the lag index, k is an integer, k∈[-L,L], L represents the maximum lag, L<m; and Let p(t) and q(t) represent the elements in the sequence sampling results that affect the feature values, respectively, and j represent the element identifier, j∈[1,m]; This represents the average value of the sequence sampling result p(t) that affects the feature value; This represents the average value of the sequence sampling results q(t) that affect the feature values; It should be noted that, This represents the number of valid data pairs with a lag of k. The sampling interval for sequence sampling is determined based on the specific equipment. Before calculating the cross-correlation coefficient, the sequence sampling results p(t) and q(t) that affect the feature values ​​are standardized to prevent the influence of extreme data and eliminate the dimensions. The standardization method can be the Z-score method, and there are no restrictions here. Steps s2-3: For each lag index k, calculate the cross-correlation number using the method in step s2-2. For a preset cross-correlation threshold, different device components that are greater than the preset cross-correlation threshold are considered to have mutual influence. Steps s2-4: Arrange the cross-correlation coefficients greater than the preset cross-correlation threshold in ascending order, and record the corresponding time delay α, α=|k| Δt; where Δt represents the sampling interval; the time delay α is used as the progress indicator of the maintenance progress network diagram, and maintenance is carried out according to the time delay α.

[0027] It should be noted that the principle of calculating the cross-correlation coefficient in this application is based on the mutual influence and correlation between equipment components. In reality, a water turbine is a mechanical structure composed of multiple equipment components working together. An abnormality in one equipment component will cause abnormalities in one or more other equipment components. If the equipment component causing the abnormality is not detected and resolved in time, it will lead to large-scale abnormalities of equipment components, resulting in huge waste of resources and maintenance costs. What this application aims to solve is to carry out maintenance based on the length of time delay, so as to ensure that the source of abnormality can be repaired first and avoid causing large-scale abnormalities. Therefore, time delay α is used as the progress indicator of the maintenance progress network diagram to ensure that the equipment maintenance is completed before time delay α. It should be noted that the cross-correlation threshold is preset to 0.6, meaning that when the cross-correlation coefficient is greater than 0.6, it is considered that there is an influence between the device components. Secondly, positive lag k>0: p leads q, and an anomaly of p may lead to an anomaly of q; negative lag k<0: p lags q, and an anomaly of q may lead to an anomaly of p. It should be noted that when k=0, both equipment component p and equipment component q need to be inspected simultaneously. In this application, the influence relationship is determined by the value of k, and the response time is set by the time delay to complete the inspection before the maximum response (i.e., the time delay).

[0028] Furthermore, in practice, even if the influencing feature value operates within the preset standard operating range [a,b] or the number of markings z does not exceed the preset first threshold, there are still cases of equipment component malfunctions during random inspections. This application provides a solution to this problem: Preferred options also include: Based on the obtained impact characteristics, the real-time impact characteristic value corresponding to any device component is acquired. For device components whose real-time impact characteristic value is within the preset standard operating range [a,b], the real-time impact characteristic value [d1,d2,…,d] within a preset time period T is collected. l ]; where d1, d2, ..., d l This represents the 1st, 2nd, ..., lth real-time influencing feature values ​​collected, where l = T / f; f represents the sampling frequency. Calculate the volatility δ of the real-time influence on the eigenvalue within a preset time period T: Where μ represents the average value of the real-time impact feature within the time period T; j represents the identifier of the real-time impact feature, j∈[1,l]; Device components whose real-time volatility δ affecting the characteristic value exceeds a preset second threshold within a preset time period T are identified as abnormal and real-time warnings are issued.

[0029] In this embodiment, a second threshold is set based on historical data. The method for setting the second threshold is to use the minimum value of the fluctuation rate of the real-time impact feature value when the equipment operation is abnormal within a period in history as the second threshold. Since the goal of this application is to carry out preventive maintenance in advance and ensure that the equipment components are potential abnormalities rather than already abnormal, the minimum value is used as the lower limit of the warning when setting the second threshold or the first threshold, which can predict the abnormality of the equipment components to a greater extent.

[0030] A data analysis system based on a maintenance progress network diagram includes a partitioned data acquisition module, a feature selection module, an anomaly detection module, and a maintenance progress network diagram module. The partition data acquisition module is used to divide the turbine into several functional partitions according to the equipment functions, further break down the equipment in each functional partition, and collect the historical and real-time operating parameter values ​​of different equipment components after the detailed breakdown. The feature selection module is used to obtain the impact characteristics of different equipment components by analyzing the historical operating parameter values ​​of the refined and segmented equipment components under normal and abnormal conditions. The anomaly detection module is used to determine whether a device component is abnormal based on the obtained impact characteristics of different device components and the real-time impact characteristic values. The maintenance progress network diagram module forms standardized maintenance task packages based on the detailed breakdown of equipment. Each task package includes the name of the maintenance equipment component, the type of detection parameter, and the qualified range of the preset detection parameters. Each standardized maintenance task package is used as a node in the maintenance progress network diagram. The time delay α is used as the progress indicator of the maintenance progress network diagram to construct a digital twin maintenance progress network diagram. Maintenance is carried out according to the maintenance progress network diagram.

[0031] Furthermore, the anomaly detection module includes a detection unit and a secondary detection unit; The judgment unit, based on the real-time impact characteristic values ​​of different device components and according to the preset standard operating range [a,b], marks the device components whose real-time impact characteristic values ​​exceed the standard operating range [a,b], records the number of markings z within a preset time period T, and records the device components whose number of markings z is greater than a preset first threshold as abnormal. The secondary judgment unit records equipment components whose real-time impact on the characteristic value is greater than a preset second threshold within a preset time period T based on the volatility of the characteristic value within the preset time period T.

[0032] Furthermore, it also includes an early warning device that provides real-time warnings for devices identified as abnormal. The early warning device is connected to the anomaly judgment module. When the anomaly judgment module determines that a device component is abnormal, the early warning device immediately triggers a real-time warning. The warning information includes the name of the abnormal device component, the anomaly judgment type, real-time impact characteristic value / volatility data, and the corresponding threshold range.

[0033] The aforementioned maintenance progress network diagram module also includes a maintenance scheduling unit. The maintenance scheduling unit prioritizes the maintenance of abnormal equipment components according to the time delay α, generates maintenance scheduling instructions, and updates the maintenance status of each node in the maintenance progress network diagram in real time, thereby realizing visualized monitoring and dynamic scheduling of the maintenance process.

[0034] The aforementioned partitioned data acquisition module includes a historical data storage unit and a real-time data acquisition unit. The historical data storage unit is used to store and classify the historical operating parameter values ​​of the device components throughout their entire lifecycle. The real-time data acquisition unit collects the operating parameters of the device components in real time through various sensors, and the acquisition frequency can be flexibly adjusted according to the importance of the device components.

[0035] The aforementioned feature selection module incorporates a correlation analysis algorithm and a threshold setting unit. The threshold setting unit can dynamically adjust relevant thresholds based on changes in equipment operating conditions and historical data updates to ensure the accuracy and adaptability of feature extraction.

[0036] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A data analysis method based on a maintenance schedule network diagram, characterized in that, Includes the following steps: The turbine is divided into several functional zones, and the equipment in each zone is further subdivided to form standardized maintenance task packages. Each task package serves as a node in the maintenance progress network diagram. Obtain historical data of the refined and segmented equipment components, and analyze the operating parameter values ​​under normal and abnormal conditions to obtain the impact characteristics of different equipment components; Based on the obtained impact characteristics of different equipment components, it is determined whether the equipment components are abnormal according to the real-time impact characteristic values; Cross-correlation analysis is performed on the equipment components that are judged to be abnormal to obtain the influence relationship between the equipment components, and the time delay of the influence is calculated. The time delay is used as the progress index of the maintenance progress network diagram for maintenance. It also includes a secondary assessment of anomalies in equipment components by influencing the volatility of eigenvalues ​​in real time.

2. The data analysis method based on a maintenance progress network diagram according to claim 1, characterized in that, The process involves dividing the turbine into several functional zones, further subdividing the equipment within each zone to form standardized maintenance task packages. Each task package serves as a node in the maintenance progress network diagram. Specifically: The turbine is divided into several functional zones according to its function. The functional zones include the main body, control system, sealing system, and protection system. The devices in each functional area are further broken down; Based on the detailed breakdown of equipment, standardized maintenance task packages are formed. Each task package includes the name of the equipment component to be maintained, the type of test parameter, and the default range of the test parameter. Each standardized maintenance task package is treated as a node in the maintenance schedule network diagram, and a digital twin maintenance schedule network diagram is constructed.

3. The data analysis method based on a maintenance progress network diagram according to claim 2, characterized in that, The process of obtaining historical data of the refined and segmented equipment components, and analyzing the operating parameter values ​​under normal and abnormal states to obtain the impact characteristics of different equipment components, specifically includes the following steps: Step s1-1: For any device component, obtain several historical operating parameter values, and classify and extract the operating parameter values ​​when normal operation occurs and those when abnormal operation occurs. For operating parameter type x, the parameter value of x when normal operation is recorded as [T]. x1 ,T x2 ,…,T xn When an anomaly occurs, the parameter value of x is denoted as [F]. x1 ,F x2 ,…,F xn ]; among which, T x1 ,T x2 ,…,T xn This represents the 1st, 2nd, ..., nth parameter value of the normal operating parameter x; F x1 ,F x2 ,…,F xn This represents the 1st, 2nd, ..., nth parameter value of the runtime parameter x when an exception occurs; Steps s1-2: Combine the parameter values ​​of the running parameter type x when it is normal and the parameter values ​​when it is abnormal into a numerical variable, and the set of numerical variables X = [T]. x1 ,T x2 ,…,T xn ,F x1 ,F x2 ,…,F xn The normal and abnormal states of the equipment are quantified into binary variables Y, with Y taking the value of 0 for the normal state and 1 for the abnormal state. The binary variables corresponding to the parameter values ​​of the operating parameter type x when they are normal and when they are abnormal are merged, and the binary variable set Y=[0,…,0,1,…,1] is formed. Among them, there are 2n values ​​in [0,…,0,1,…,1]. Steps s1-3: Using the numerical variable X and the binary variable Y representing the equipment status as a basis, perform correlation analysis, calculate the correlation coefficient, and characterize it as follows: Where r represents the correlation coefficient between the operating parameter type x and the device status; This represents an element in the set of numerical variables X, where i is the element identifier, and i∈[1,2n]. Represents the elements in the binary variable set Y; Represents the average value of a numerical variable; The average value of the binary variable Y representing the equipment status; Steps s1-4: Calculate the correlation coefficients between different operating parameter types and equipment status according to the method in step s3, and determine the operating parameter types with absolute values ​​of correlation coefficients greater than preset correlation thresholds as influencing features.

4. The data analysis method based on a maintenance progress network diagram according to claim 3, characterized in that, The method further includes: Based on the obtained impact characteristics, the real-time impact characteristic value corresponding to any device component is obtained. According to the preset standard operating range [a,b], the device components whose real-time impact characteristic value exceeds the standard operating range [a,b] are marked, and the number of markings z within the preset time period T is recorded. The device components whose number of markings z is greater than the preset first threshold are marked as abnormal and real-time warnings are issued.

5. The data analysis method based on a maintenance progress network diagram according to claim 4, characterized in that, The aforementioned method of performing cross-correlation analysis on equipment components identified as abnormal to obtain the influence relationships between equipment components, calculating the time lag of the influence, and using the time lag as the progress index of the maintenance progress network diagram for maintenance, specifically includes the following steps: Step s2-1: Perform sequence sampling of the influence feature values ​​of different equipment components marked as abnormal. For different equipment components p and q, the sequence sampling results p(t) and q(t) of the influence feature values ​​are expressed as: p(t) = [p1, p2, ..., p m ] and q(t)=[q1,q2,…,q m ]; where m represents the number of sequence samples affecting the feature value, and m is a positive integer; t represents the time series identifier; Step s2-2: Calculate the cross-correlation coefficient r at lag k based on the sequence sampling results p(t) and q(t) that affect the eigenvalues. pq (k), Where k represents the lag index, k is an integer, k∈[-L,L], L represents the maximum lag, L<m; and Let p(t) and q(t) represent the elements in the sequence sampling results that affect the feature values, respectively, and j represent the element identifier, j∈[1,m]; This represents the average value of the sequence sampling result p(t) that affects the feature value; This represents the average value of the sequence sampling results q(t) that affect the feature values; Steps s2-3: For each lag index k, calculate the cross-correlation number using the method in step s2-2. For a preset cross-correlation threshold, different device components that are greater than the preset cross-correlation threshold are considered to have mutual influence. Steps s2-4: Arrange the cross-correlation coefficients greater than the preset cross-correlation threshold in ascending order, and record the corresponding time delay α, α=|k| Δt; where Δt represents the sampling interval; the time delay α is used as the progress indicator of the maintenance progress network diagram, and maintenance is carried out according to the time delay α.

6. The data analysis method based on a maintenance progress network diagram according to claim 3, characterized in that, The method further includes: Based on the obtained impact characteristics, the real-time impact characteristic value corresponding to any device component is acquired. For device components whose real-time impact characteristic value is within the preset standard operating range [a,b], the real-time impact characteristic value [d1,d2,…,d] within a preset time period T is collected. l ]; where d1, d2, ..., d l This represents the 1st, 2nd, ..., lth real-time influencing feature values ​​collected, where l = T / f; f represents the sampling frequency. Calculate the volatility δ of the real-time influence on the eigenvalue within a preset time period T: Where μ represents the average value of the real-time impact feature within the time period T; j represents the identifier of the real-time impact feature, j∈[1,l]; Device components whose real-time volatility δ affecting the feature value exceeds a preset second threshold within a preset time period T are identified as abnormal and real-time warnings are issued.

7. A system using the data analysis method based on a maintenance progress network diagram as described in any one of claims 1-6, characterized in that, It includes a partitioned data acquisition module, a feature selection module, an anomaly detection module, and a maintenance progress network diagram module; The partition data acquisition module is used to divide the turbine into several functional partitions according to the equipment functions, further subdivide the equipment in each functional partition, and collect the historical and real-time operating parameter values ​​of different equipment components after the subdivision. The feature selection module is used to obtain the influence characteristics of different equipment components by analyzing the historical operating parameter values ​​of the refined and segmented equipment components under normal and abnormal conditions. The anomaly detection module is used to determine whether a device component is abnormal based on the obtained impact characteristics of different device components and the real-time impact characteristic value. The maintenance progress network diagram module forms standardized maintenance task packages based on the detailed breakdown of equipment. Each task package includes the name of the maintenance equipment component, the type of detection parameter, and the qualified range of the preset detection parameters. Each standardized maintenance task package is used as a node in the maintenance progress network diagram. The time delay α is used as the progress indicator of the maintenance progress network diagram to construct a digital twin maintenance progress network diagram. Maintenance is carried out according to the maintenance progress network diagram.

8. A data analysis system based on a maintenance progress network diagram according to claim 7, characterized in that, The anomaly detection module includes a detection unit and a secondary detection unit; The judgment unit, based on the real-time impact feature values ​​of different device components and according to the preset standard operating range [a,b], marks the device components whose real-time impact feature values ​​exceed the standard operating range [a,b], records the number of markings z within a preset time period T, and records the device components whose number of markings z is greater than a preset first threshold as abnormal. The secondary judgment unit records equipment components whose real-time impact on the feature value is greater than a preset second threshold within a preset time period T based on the volatility of the feature value within the preset time period T.

9. A data analysis system based on a maintenance progress network diagram according to claim 8, characterized in that, It also includes an early warning device that provides real-time warnings for devices marked as abnormal. The early warning device is connected to the anomaly judgment module. When the anomaly judgment module determines that a device component is abnormal, the early warning device immediately triggers a real-time warning. The warning information includes the name of the abnormal device component, the anomaly judgment type, the real-time impact characteristic value / volatility data, and the corresponding threshold range.

10. A data analysis system based on a maintenance progress network diagram according to claim 7, characterized in that, The maintenance progress network diagram module also includes a maintenance scheduling unit. The maintenance scheduling unit prioritizes the maintenance of abnormal equipment components according to the time delay α, generates maintenance scheduling instructions, and updates the maintenance status of each node in the maintenance progress network diagram in real time, thereby realizing visualized monitoring and dynamic scheduling of the maintenance process.

11. A data analysis system based on a maintenance progress network diagram according to claim 7, characterized in that, The partitioned data acquisition module includes a historical data storage unit and a real-time data acquisition unit. The historical data storage unit is used to store and classify the historical operating parameter values ​​of the device components throughout their entire life cycle. The real-time data acquisition unit collects the operating parameters of the device components in real time through various sensors, and the acquisition frequency can be flexibly adjusted according to the importance of the device components.

12. A data analysis system based on a maintenance progress network diagram according to claim 7, characterized in that, The feature selection module has a built-in correlation analysis algorithm and a threshold setting unit. The threshold setting unit can dynamically adjust the relevant thresholds according to changes in equipment operating conditions and historical data updates to ensure the accuracy and adaptability of feature extraction.

Images