Power plant equipment fault control system and method based on big data analysis

Abstract

The invention discloses a power plant equipment fault control system and method based on big data analysis, and belongs to the technical field of fault control. Power plant equipment operation data is acquired, feature extraction is performed, an equipment feature vector and a fault database are constructed, and a fault positioning model is constructed; importing the equipment feature vector into a constructed fault positioning model, positioning a fault position through similarity calculation, constructing a fault evolution model, importing current equipment parameters into the fault evolution model to evaluate a fault evolution trend and a future fault risk score, constructing an optimal maintenance time evaluation model, and evaluating the fault evolution trend and the future fault risk score. The fault risk score, the equipment operation year number and the maintenance cost are imported into the optimal maintenance time assessment model to assess the maintenance duration and the optimal maintenance time, and efficient control and management of power plant equipment faults are achieved through data collection, feature extraction, fault positioning, fault evolution and optimal maintenance assessment multi-dimensional analysis.

Description

Technical Field

[0001] The present application belongs to the field of fault control technology, and more specifically, is a power plant equipment fault control system and method based on big data analysis. Background Art

[0002] Failures in power plant equipment not only lead to production stagnation and waste of resources, but can also have serious environmental impacts and even cause safety accidents. The current, commonly adopted maintenance strategy for mechanical equipment is to set fixed-cycle scheduled maintenance based on expert experience or manufacturer instructions. This is inflexible and difficult to set. This leads to untimely equipment maintenance and a lack of comprehensive monitoring and in-depth analysis of equipment status, resulting in low fault diagnosis accuracy, especially in the early stages of fault diagnosis. Furthermore, there is a lack of effective prediction methods for potential risks and fault evolution trends during equipment operation, making proactive maintenance impossible.

[0003] This application proposes a power plant equipment fault control system and method based on big data analysis. Through multi-dimensional analysis methods such as data collection, feature extraction, fault location, fault evolution prediction and optimal maintenance evaluation, it realizes the full life cycle management of equipment and effectively improves the accuracy and timeliness of equipment fault prediction. At the same time, by using big data analysis, it dynamically adjusts the maintenance strategy according to the real-time operating status of the equipment and the fault evolution trend, reduces equipment downtime and maintenance costs, and improves the overall production efficiency and economic benefits of the power plant. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, this application proposes a power plant equipment fault control system and method based on big data analysis.

[0005] To achieve the above objectives, this application provides the following technical solutions:

[0006] A power plant equipment fault control system and method based on big data analysis includes the following specific steps:

[0007] Obtain power plant equipment operating data, perform feature extraction, and build equipment feature vectors and fault databases;

[0008] Build a fault location model, import the device feature vector into the built fault location model and locate the fault position through similarity calculation;

[0009] Build a fault evolution model and import current equipment parameters into the fault evolution model to evaluate fault evolution trends and future fault risk scores;

[0010] Build an optimal maintenance time evaluation model, import the failure risk score, equipment operation years and maintenance cost into the optimal maintenance time evaluation model to evaluate the maintenance time and optimal maintenance time.

[0011] Preferably, the steps of acquiring power plant equipment operation data, performing feature extraction, and constructing equipment feature vectors and a fault database include the following specific steps:

[0012] S11. Deploy multiple types of sensors at power plant equipment nodes to collect various parameters of the power plant equipment through the sensors, and perform noise reduction and normalization processing on the collected data;

[0013] S12. Generate a device feature vector from the processed device parameters, where the device feature vector at time t is: x(t) = [x1(t), x2(t), ..., x i (t),...,x n (t)], where x i (t) is the i-th characteristic parameter of the device at time t, and n is the total number of characteristics;

[0014] S13. Obtain historical fault data, extract key features of equipment faults, and establish an equipment fault database, wherein the equipment fault database is: Among them, F (k) is the historical feature sample set of fault type k, K is the total number of fault types, where the historical feature sample set of fault type k is: in, is the jth historical feature vector belonging to fault type k, which contains multiple characteristic parameters under the equipment operation state, k is the kth fault type, j is the sample number, and m is the number of samples under the kth fault type. Among them, the jth historical feature vector belonging to fault type k is: Each historical fault type stored in the fault database corresponds to one or more feature vectors.

[0015] Preferably, the constructing of the fault location model, importing the device feature vector into the constructed fault location model and locating the fault position by similarity calculation comprises the following specific steps:

[0016] S21. Substitute the device feature vector and the fault history feature vector in the fault database into the fault similarity calculation formula to evaluate the current device status, where the fault similarity calculation formula is: in, is the distance between the device feature vector and the fault history feature vector. The distance calculation formula between the device feature vector and the fault history feature vector is: Among them, w i is the weight, which indicates the importance of each feature to fault identification;

[0017] S22. Compare the maximum fault similarity with the similarity threshold. If it exceeds the threshold range, it is determined that the current equipment has a fault trend, and the ten groups of feature samples with the highest similarity are taken to perform fault type frequency statistics to determine the fault type with the highest frequency in the sample. Other groups with higher similarity are tracked and monitored.

[0018] Preferably, the construction of the fault evolution model and the importation of current equipment parameters into the fault evolution model to evaluate the fault evolution trend and future fault risk score include the following specific steps:

[0019] S31. Construct a historical equipment failure trajectory database, collect the operating parameter sequences of various equipment in the power plant within a period of time before and after the failure, and form a multi-dimensional time series trajectory. The fault evolution trajectory of the jth historical fault k is: in, The jth fault sample at t h The n-dimensional feature vector at time t, where the j-th fault sample is at t h The n-dimensional feature vector at the moment is: in, The jth fault sample at t h The eigenvalue at time i;

[0020] S32, the feature vectors collected by the current device at multiple times form the real-time running trajectory of the device, wherein the real-time running trajectory of the device is: X c (t)={x(t1),x(t2),...,x(t h ),...,x(t H )}, where x(t h ) is t h The eigenvector of the moment;

[0021] S33. Align trajectories of different lengths through dynamic time warping, substitute the real-time operation trajectory and historical fault evolution trajectory of the device into the fault evolution trend matching calculation formula, and select the one with the smallest difference between the real-time operation trajectory and the historical operation trajectory as the reference trajectory. The fault evolution trend matching calculation formula is: The dynamic time warping method is used to time-align the current fault evolution trajectory with the historical sample trajectory. The Euclidean distance is used to calculate the difference between points in the feature dimension, and the overall matching cost is calculated at the sequence level. Finally, the DTW distance corresponding to the minimum cost path is obtained, realizing the search for the optimal match between the current fault trajectory and the historical trajectory.

[0022] S34, based on the reference trajectory, a multi-step prediction model is constructed through the LSTM long short-term memory network, which inputs the feature vector sequence of the current moment and several past moments, and outputs the fault risk score R(tv ), the risk prediction values ​​of multiple time steps are used to generate a failure risk score curve.

[0023] Preferably, the construction of the optimal maintenance time evaluation model, importing the fault risk score, equipment operating years and maintenance cost into the optimal maintenance time evaluation model to evaluate the maintenance duration and optimal maintenance time includes the following specific steps:

[0024] S41. Substitute the equipment's operating years, basic equipment maintenance time, and worker proficiency into the fault repair time calculation formula to calculate the fault repair time. The fault repair time calculation formula is: Among them, E is the number of years the equipment has been in operation, L is the fixed life of the equipment, T b is the basic maintenance time of equipment, μ η is the worker's proficiency in fault repair, ε is the number of maintenance workers, R th (t) The optimal value of the fault risk score, where the basic equipment maintenance time is obtained through historical average data, and proficiency is evaluated by counting the number of times and effectiveness of workers handling specific faults through maintenance records;

[0025] S42. Calculate the losses of maintenance at each time point, select the maintenance time point with the least loss, and substitute the production loss and maintenance cost during the maintenance period into the loss calculation formula to calculate the losses during the maintenance period. The loss calculation formula is: L(t) = G a (t)+G p (t), where G a (t) is the production loss of the equipment during the maintenance period, G p (t) is the maintenance cost, including labor, materials, and downtime preparation. The production loss of equipment during maintenance is calculated as follows: Where P(t') is the economic output per unit time during equipment operation, and λ(t') is the functional reduction coefficient during maintenance. When the equipment is shut down, λ(t') = 1, and when it is degraded, λ(t') ∈ (0, 1).

[0026] The power plant equipment fault control system based on big data analysis is implemented based on the above-mentioned power plant equipment fault control method based on big data analysis, and specifically includes:

[0027] Data acquisition module, used to obtain power plant equipment operation data;

[0028] Database construction module, used to build a fault database;

[0029] The fault location module is used to locate the fault location by calculating the similarity between the device feature vector and the fault history feature vector in the fault database;

[0030] The fault evolution module is used to evaluate the fault evolution trend and future fault risk score based on the real-time operation trajectory and historical fault evolution trajectory of the equipment;

[0031] The optimal maintenance time evaluation module is used to evaluate the maintenance duration and optimal maintenance time based on the failure risk score, equipment operation years and maintenance cost.

[0032] An electronic device comprises: a processor and a memory, wherein the memory stores a computer program that can be called by the processor;

[0033] The processor executes the above-mentioned power plant equipment fault control method based on big data analysis by calling the computer program stored in the memory.

[0034] A computer-readable storage medium is characterized in that it stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned power plant equipment fault control method based on big data analysis.

[0035] Compared with the prior art, the present invention has the following advantages:

[0036] This application obtains power plant equipment operation data, performs feature extraction and constructs equipment feature vectors and a fault database, constructs a fault location model, imports the equipment feature vectors into the constructed fault location model to locate the fault position through similarity calculation, constructs a fault evolution model, imports current equipment parameters into the fault evolution model to evaluate the fault evolution trend and future fault risk score, constructs an optimal maintenance time assessment model, imports the fault risk score, equipment operation years and maintenance cost into the optimal maintenance time assessment model to evaluate the maintenance time and optimal maintenance time. This application achieves efficient control and management of power plant equipment failures through multi-dimensional analysis of data collection, feature extraction, fault location, fault evolution and optimal maintenance assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a schematic diagram of the overall process of the power plant equipment fault control method based on big data analysis in this application;

[0038] Figure 2 This is the fault location flowchart for this application;

[0039] Figure 3 This is the fault evolution flow chart of this application;

[0040] Figure 4 This is a schematic diagram of the overall framework of the power plant equipment fault control system based on big data analysis in this application. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments.

[0042] Example 1

[0043] See also Figure 1 , this application provides an embodiment: a power plant equipment fault control method based on big data analysis, which includes the following specific steps:

[0044] Obtain power plant equipment operating data, perform feature extraction, and build equipment feature vectors and fault databases;

[0045] Build a fault location model, import the device feature vector into the built fault location model and locate the fault position through similarity calculation;

[0046] Build a fault evolution model and import current equipment parameters into the fault evolution model to evaluate fault evolution trends and future fault risk scores;

[0047] Build an optimal maintenance time evaluation model, import the failure risk score, equipment operation years and maintenance cost into the optimal maintenance time evaluation model to evaluate the maintenance time and optimal maintenance time.

[0048] In this embodiment, it should be specifically explained that obtaining power plant equipment operating data, performing feature extraction, and constructing equipment feature vectors and a fault database include the following specific steps:

[0049] S11. Deploy multiple types of sensors at power plant equipment nodes to collect various parameters of the power plant equipment through the sensors, and perform noise reduction and normalization processing on the collected data;

[0050] For example, when performing noise reduction and normalization processing on the collected device data, methods such as mean filtering or median filtering are used to reduce the noise of the data. The data is smoothed by taking the average or median of adjacent data points to eliminate high-frequency noise, and the values ​​of different features are adjusted to the same scale range. The data is linearly transformed according to the maximum and minimum value ranges through minimum-maximum normalization. After the noise reduction and normalization processing, the quality of the data is improved, and the eigenvalues ​​have a unified dimension.

[0051] S12. Generate a device feature vector from the processed device parameters, where the device feature vector at time t is: x(t) = [x1(t), x2(t), ..., x i (t),...,x n (t)], where x i (t) is the i-th characteristic parameter of the device at time t, and n is the total number of characteristics;

[0052] For example, the device feature vector at time t includes the device operating temperature, pressure, speed, probability, voltage, current, etc.;

[0053] S13. Obtain historical fault data, extract key features of equipment faults, and establish an equipment fault database, wherein the equipment fault database is: Among them, F (k) is the historical feature sample set of fault type k, K is the total number of fault types, where the historical feature sample set of fault type k is: in, is the jth historical feature vector belonging to fault type k, which contains multiple characteristic parameters under the equipment operation state, k is the kth fault type, j is the sample number, and m is the number of samples under the kth fault type. Among them, the jth historical feature vector belonging to fault type k is: Each historical fault type stored in the fault database corresponds to one or more feature vectors;

[0054] For example, the equipment failure database includes boiler equipment failure, generator equipment failure, compressor equipment failure, transformer equipment failure, etc., and the historical feature vector of fault type k includes temperature failure, vibration failure, pressure failure, production failure, bearing failure, etc.

[0055] In this embodiment, it should be specifically explained that Figure 2 As shown, building a fault location model, importing the device feature vector into the fault location model and locating the fault position through similarity calculation includes the following specific steps:

[0056] S21. Substitute the device feature vector and the fault history feature vector in the fault database into the fault similarity calculation formula to evaluate the current device status, where the fault similarity calculation formula is: in, is the distance between the device feature vector and the fault history feature vector. The distance calculation formula between the device feature vector and the fault history feature vector is: Among them, w i is the weight, which indicates the importance of each feature to fault identification. The similarity between the current equipment status and the historical fault samples is evaluated by weighting the distance between the features.

[0057] For example, in this embodiment, w i The frequency of historical fault occurrence is used to represent the number of times that temperature faults, vibration faults, voltage faults, and other faults have occurred in the history of overheating faults of generator equipment. The weighted fault similarity formula has the ability to specifically handle the differences in feature importance and can improve the accuracy and adaptability of fault detection.

[0058] S22. Compare the maximum fault similarity with the similarity threshold. If it exceeds the threshold, determine that the current device has a fault trend. Take the ten groups of feature samples with the highest similarity to perform fault type frequency statistics, determine the most frequent fault type in the samples, and track and monitor other groups with high similarity.

[0059] It should be noted here that the method for setting the similarity threshold in this embodiment is: obtaining representative historical power plant equipment failure data, while collecting the equipment's operating parameters and historical failure type information before and after the failure occurs, hiring engineering experts to manually determine the equipment failure type, and substituting the obtained historical data into the various steps in this embodiment for calculation, comparing the calculation results with actual failure cases, and using data fitting software to optimize the similarity threshold. By gradually adjusting the similarity threshold and matching it with the actual failure situation, the final output is the similarity threshold value that can provide the highest fault diagnosis accuracy.

[0060] In this embodiment, it should be specifically explained that Figure 3 As shown in the figure, building a fault evolution model and importing current device parameters into the fault evolution model to evaluate the fault evolution trend and future fault risk score includes the following specific steps:

[0061] S31. Construct a historical equipment failure trajectory database, collect the operating parameter sequences of various equipment in the power plant within a period of time before and after the failure, and form a multi-dimensional time series trajectory. The fault evolution trajectory of the jth historical fault k is: in, The jth fault sample at t h The n-dimensional feature vector at time t, where the j-th fault sample is at t h The n-dimensional feature vector at the moment is: in, The jth fault sample at t h The eigenvalue at time i;

[0062] For example, in this embodiment, we analyze overheating faults in generator samples. By recording parameters such as temperature, vibration, current, speed, and equipment load within 24 hours before and after the fault, we construct a time series trajectory containing multiple features. Before the fault occurs, the various parameters of the equipment gradually change. When the fault occurs, the temperature, current, and other indicators of the equipment exceed the normal range. After the fault occurs, the equipment parameters further deteriorate, indicating the continuous evolution of the fault. The multi-dimensional feature data constitutes the historical fault trajectory.

[0063] S32, the feature vectors collected by the current device at multiple times form the real-time running trajectory of the device, wherein the real-time running trajectory of the device is: X c(t)={x(t1),x(t2),...,x(t h ),...,x(t H )}, where x(t h ) is t h The eigenvector of the moment;

[0064] For example, relevant parameters at various time points within a certain period of time of the generator overheating fault are collected, including parameters such as temperature, vibration, current, speed, and equipment load;

[0065] S33. Align trajectories of different lengths through dynamic time warping, substitute the real-time operation trajectory and historical fault evolution trajectory of the device into the fault evolution trend matching calculation formula, and select the one with the smallest difference between the real-time operation trajectory and the historical operation trajectory as the reference trajectory. The fault evolution trend matching calculation formula is: The dynamic time warping method is used to time-align the current fault evolution trajectory with the historical sample trajectory. The Euclidean distance is used to calculate the difference between points in the feature dimension, and the overall matching cost is calculated at the sequence level. Finally, the DTW distance corresponding to the minimum cost path is obtained, realizing the search for the optimal match between the current fault trajectory and the historical trajectory.

[0066] In this embodiment, it is important to note that by calculating the Euclidean distance on the feature dimension and evaluating the differences between different time points, DTW further calculates the overall matching cost and finds the minimum cost path. The DTW distance corresponding to this minimum cost path is the similarity between the real-time trajectory and the historical trajectory. The historical trajectory that best matches the current fault trajectory is obtained. The evolution pattern of the historical fault is used to predict the evolution trend of the current equipment fault.

[0067] S34, based on the reference trajectory, a multi-step prediction model is constructed through the LSTM long short-term memory network, which inputs the feature vector sequence of the current moment and several past moments, and outputs the fault risk score R(t v ), the risk prediction values ​​of multiple time steps are used to generate a failure risk score curve.

[0068] What needs to be specifically explained in this embodiment is that, through the LSTM (long short-term memory network) model, multi-step fault risk prediction is performed on the basis of the reference trajectory, with the device feature vector sequence at the current moment and several past moments as input. The feature vector includes multiple parameters of the device and comprehensively reflects the operating status of the device. The LSTM network captures the long-term dependencies in the time series and uses the dynamic change patterns in the historical data to predict fault risks. The output of the LSTM network is the fault risk score at several future moments, which indicates the failure probability or severity of the device at each moment. As the time step increases, the risk value gradually predicts and reflects the potential changes in the device status, and the fault risk scores of multiple time steps are connected into a fault risk score curve.

[0069] In this embodiment, it should be specifically explained that constructing an optimal maintenance time evaluation model, importing the failure risk score, equipment operating years, and maintenance cost into the optimal maintenance time evaluation model to evaluate the maintenance duration and optimal maintenance time includes the following specific steps:

[0070] S41. Substitute the equipment's operating years, basic equipment maintenance time, and worker proficiency into the fault repair time calculation formula to calculate the fault repair time. The fault repair time calculation formula is: Among them, E is the number of years the equipment has been in operation since it was put into operation, L is the fixed life of the equipment, T b is the basic maintenance time of equipment, μ η is the worker's proficiency in fault repair, ε is the number of maintenance workers, R th (t) The optimal value of the failure risk score, where the basic maintenance time of the equipment is obtained through historical average data, and proficiency is evaluated by counting the number of times and effectiveness of workers in handling specific failures through maintenance records. The evaluation of maintenance time integrates the age of the equipment, maintenance history, proficiency of the maintenance personnel, and the severity of the failure;

[0071] For example, a pump equipment in a power plant fails. The equipment has been in operation for 8 years, has a fixed life of 16 years, has a basic maintenance time of 10 hours, has maintenance workers with proficiency levels of 0.8 and 0.5, has 2 maintenance workers, and a failure risk score of 40. The optimal failure risk score is 100. Substituting these values ​​into the failure repair time calculation formula yields the following repair time:

[0072] S42. Calculate the losses of maintenance at each time point, select the maintenance time point with the least loss, and substitute the production loss and maintenance cost during the maintenance period into the loss calculation formula to calculate the losses during the maintenance period. The loss calculation formula is: L(t) = G a (t)+G p (t), where G a(t) is the production loss of the equipment during the maintenance period, G p (t) is the maintenance cost, including labor, materials, and downtime preparation. The production loss of equipment during maintenance is calculated as follows: Where P(t') is the economic output per unit time during equipment operation, and λ(t') is the functional impairment coefficient during maintenance. When the equipment is shut down, λ(t') = 1, and when it is operating in a degraded state, λ(t') ∈ (0, 1). The output per unit time can be obtained from the output and operating condition information recorded by the production management system. The functional impairment coefficient can be obtained by comparing the historical equipment performance with the current status and based on expert evaluation scores.

[0073] For example, historical data shows that when a certain type of failure occurs in the equipment, the output per unit time drops by 30%. Based on this historical data, the function reduction coefficient of the equipment in a failure or degraded state can be preliminarily set, and this coefficient can be further adjusted by real-time monitoring of the equipment's operating status. If the equipment's operating status is close to normal, the function reduction coefficient is low. If the equipment fails or is in a degraded operating state, the function reduction coefficient will increase accordingly.

[0074] When the equipment is in a minor fault state, the function reduction coefficient is set to 0.8, indicating an output loss of 20%. If the equipment is shut down, the function reduction coefficient is set to 1, indicating an output loss of 100%. Experts have found that a certain type of fault can cause the equipment to shut down for 24 hours, which has a greater impact. Therefore, the function reduction coefficient of this fault type is increased.

[0075] The advantages of this embodiment over the prior art are:

[0076] This application obtains power plant equipment operation data, performs feature extraction and constructs equipment feature vectors and a fault database, constructs a fault location model, imports the equipment feature vectors into the constructed fault location model to locate the fault position through similarity calculation, constructs a fault evolution model, imports current equipment parameters into the fault evolution model to evaluate the fault evolution trend and future fault risk score, constructs an optimal maintenance time assessment model, imports the fault risk score, equipment operation years and maintenance cost into the optimal maintenance time assessment model to evaluate the maintenance time and optimal maintenance time. This application achieves efficient control and management of power plant equipment failures through multi-dimensional analysis of data collection, feature extraction, fault location, fault evolution and optimal maintenance assessment.

[0077] Example 2

[0078] like Figure 4As shown, a power plant equipment fault control system based on big data analysis is implemented based on the above-mentioned power plant equipment fault control method based on big data analysis, and specifically includes a data acquisition module, a database construction module, a fault location module, a fault evolution module and an optimal maintenance time evaluation module. The data acquisition module is used to obtain power plant equipment operation data; the database construction module is used to establish a fault database; the fault location module is used to locate the fault position by calculating the similarity between the equipment feature vector and the fault history feature vector in the fault database; the fault evolution module is used to evaluate the fault evolution trend and future fault risk score through the real-time operation trajectory and historical fault evolution trajectory of the equipment; the optimal maintenance time evaluation module is used to evaluate the maintenance time and the optimal maintenance time through the fault risk score, the equipment operation years and the maintenance cost.

[0079] Example 3

[0080] This embodiment provides an electronic device, comprising: a processor and a memory, wherein the memory stores a computer program that can be called by the processor;

[0081] The processor executes the above-mentioned power plant equipment fault control method based on big data analysis by calling the computer program stored in the memory.

[0082] The electronic device may have relatively large differences due to different configurations or performances, and may include one or more processors (Central Processing Units, CPU) and one or more memories, wherein the memory stores at least one computer program, which is loaded and executed by the processor to implement the power plant equipment fault control method based on big data analysis provided by the above method embodiment. The electronic device may also include other components for realizing the functions of the device. For example, the electronic device may also have components such as a wired or wireless network interface and an input and output interface to input and output data. This embodiment will not be described in detail here.

[0083] Example 4

[0084] This embodiment provides a computer-readable storage medium having a rewritable computer program stored thereon;

[0085] When the computer program runs on a computer device, the computer device executes the above-mentioned power plant equipment fault control method based on big data analysis.

[0086] For example, computer-readable storage media can be read-only memory (ROM), random access memory (RAM), compact disc read-only memory (CD-ROM), magnetic tape, floppy disk, and optical data storage device.

[0087] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function according to the embodiments of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired network or / and a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

Claims

1. A power plant equipment fault control method based on big data analysis, characterized in that: It includes the following specific steps: Obtain power plant equipment operating data, perform feature extraction, and build equipment feature vectors and fault databases; Build a fault location model, import the device feature vector into the built fault location model and locate the fault position through similarity calculation; Build a fault evolution model and import current equipment parameters into the fault evolution model to evaluate fault evolution trends and future fault risk scores; Build an optimal maintenance time evaluation model, import the failure risk score, equipment operation years and maintenance cost into the optimal maintenance time evaluation model to evaluate the maintenance time and optimal maintenance time.

2. The power plant equipment fault control method based on big data analysis according to claim 1, characterized in that: The steps of obtaining power plant equipment operation data, extracting features, and constructing equipment feature vectors and a fault database include the following: S11. Deploy multiple types of sensors at power plant equipment nodes to collect various parameters of the power plant equipment through the sensors, and perform noise reduction and normalization processing on the collected data; S12. Generate a device feature vector from the processed device parameters, where the device feature vector at time t is: x(t) = [x1(t), x2(t), ..., x i (t),...,x n (t)], where x i (t) is the i-th characteristic parameter of the device at time t, and n is the total number of characteristics; S13. Obtain historical fault data, extract key features of equipment faults, and establish an equipment fault database, wherein the equipment fault database is: Among them, F (k) is the historical feature sample set of fault type k, K is the total number of fault types, where the historical feature sample set of fault type k is: in, is the jth historical feature vector belonging to fault type k, k is the kth fault type, j is the sample number, and m is the number of samples under the kth fault type.

3. The power plant equipment fault control method based on big data analysis according to claim 2, characterized in that: The construction of the fault location model and the importation of the device feature vector into the constructed fault location model and the location of the fault by similarity calculation include the following specific steps: S21. Substitute the device feature vector and the fault history feature vector in the fault database into the fault similarity calculation formula to evaluate the current device status, where the fault similarity calculation formula is: in, is the distance between the device feature vector and the fault history feature vector. The distance calculation formula between the device feature vector and the fault history feature vector is: Among them, w i is the weight; S22. Compare the maximum fault similarity with the similarity threshold. If it exceeds the threshold range, it is determined that the current device has a fault trend, and the ten groups of feature samples with the highest similarity are taken to perform fault type frequency statistics to determine the fault type with the highest frequency in the samples.

4. The power plant equipment fault control method based on big data analysis according to claim 3, characterized in that: The construction of the fault evolution model and the importation of current device parameters into the fault evolution model to evaluate the fault evolution trend and future fault risk score include the following specific steps: S31. Construct a historical equipment failure trajectory database, collect the operating parameter sequences of various equipment in the power plant within a period of time before and after the failure, and form a multi-dimensional time series trajectory. The fault evolution trajectory of the jth historical fault k is: in, The jth fault sample at t h n-dimensional feature vector at the moment; S32, the feature vectors collected by the current device at multiple times form the real-time running trajectory of the device, wherein the real-time running trajectory of the device is: X c (t)={x(t1),x(t2),...,x(t h ),...,x(t H )}, where x(t h ) is t h The eigenvector of the moment; S33. Align trajectories of different lengths through dynamic time warping, substitute the real-time operation trajectory and historical fault evolution trajectory of the device into the fault evolution trend matching calculation formula, and select the one with the smallest difference between the real-time operation trajectory and the historical operation trajectory as the reference trajectory. The fault evolution trend matching calculation formula is: S34, based on the reference trajectory, a multi-step prediction model is constructed through the LSTM long short-term memory network, which inputs the feature vector sequence of the current moment and several past moments, and outputs the fault risk score R(t v ), the risk prediction values ​​of multiple time steps are used to generate a failure risk score curve.

5. The power plant equipment fault control method based on big data analysis according to claim 4, characterized in that: The construction of the optimal maintenance time evaluation model, importing the fault risk score, equipment operating years and maintenance cost into the optimal maintenance time evaluation model to evaluate the maintenance duration and optimal maintenance time, includes the following specific steps: S41. Substitute the equipment's operating years, basic equipment maintenance time, and worker proficiency into the fault repair time calculation formula to calculate the fault repair time. The fault repair time calculation formula is: Among them, E is the number of years the equipment has been in operation, L is the fixed life of the equipment, T b is the basic maintenance time of equipment, μ η is the worker's proficiency in fault repair, ε is the number of maintenance workers, R th (t) optimal value of failure risk score; S42. Calculate the losses of maintenance at each time point, select the maintenance time point with the least loss, and substitute the production loss and maintenance cost during the maintenance period into the loss calculation formula to calculate the losses during the maintenance period. The loss calculation formula is: L(t) = G a (t)+G p (t), where G a (t) is the production loss of the equipment during the maintenance period, G p (t) is the maintenance cost, where the production loss of equipment during maintenance is calculated as: Where P(t') is the economic output per unit time during equipment operation, and λ(t') is the functional impairment coefficient during maintenance.

6. A power plant equipment fault control system based on big data analysis, which is implemented based on the power plant equipment fault control method based on big data analysis according to any one of claims 1 to 6, characterized in that: Specifically include: Data acquisition module, used to obtain power plant equipment operation data; Database construction module, used to build a fault database; The fault location module is used to locate the fault location by calculating the similarity between the device feature vector and the fault history feature vector in the fault database; The fault evolution module is used to evaluate the fault evolution trend and future fault risk score based on the real-time operation trajectory and historical fault evolution trajectory of the equipment; The optimal maintenance time evaluation module is used to evaluate the maintenance duration and optimal maintenance time based on the failure risk score, equipment operation years and maintenance cost.

7. An electronic device comprising: A processor and a memory, wherein the memory stores a computer program that can be called by the processor; It is characterized in that the processor executes the power plant equipment fault control method based on big data analysis as described in any one of claims 1 to 5 by calling the computer program stored in the memory.

8. A computer-readable storage medium, characterized in that Instructions are stored, and when the instructions are run on a computer, the computer is caused to execute the power plant equipment fault control method based on big data analysis as described in any one of claims 1 to 5.

Images