A method and device for locating a fault in a soft direct current converter valve

Abstract

This invention provides a method and apparatus for fault location of a flexible DC converter valve. The method includes: performing real-time fault location of the flexible DC converter valve through a three-layer architecture of distributed fault monitoring and tracing in a cloud-edge-device collaborative environment, avoiding missed detections and false detections, reducing costs, and improving detection efficiency; through distributed monitoring, bandwidth utilization can be improved, avoiding network congestion or network latency, enabling rapid response to detection tasks, improving alarm timeliness, reducing computing power requirements, and saving computing resources.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a fault location method and apparatus for a flexible DC converter valve. Background Technology

[0002] Currently, flexible DC transmission projects are mainly demonstration projects. With the advancement of power electronics technology and the improvement of control technology, line voltage and transmission capacity are rapidly increasing, and the application space of flexible DC is constantly expanding. Among related technologies, for the application of various monitoring methods and various sensor devices, the status monitoring of various sensors mainly adopts traditional rule-based alarms and periodic inspections, which are costly and inefficient, and prone to problems such as missed detections and false detections, failing to guarantee the accuracy and reliability of monitoring data. Existing intelligent algorithm monitoring of flexible DC converter valve equipment adopts a centralized monitoring method, and various monitoring data occupy effective bandwidth, causing network congestion or network latency, failing to respond quickly to monitoring tasks, and resulting in poor alarm timeliness. Centralized computing exacerbates resource centralization, and a large amount of inefficient data wastes a lot of computing resources. Summary of the Invention

[0003] One objective of this invention is to provide a fault location method for a flexible DC-DC converter valve. A three-layer architecture of distributed fault monitoring and tracing in a cloud-edge-device collaborative environment enables real-time fault location of the flexible DC-DC converter valve, avoiding missed and false detections, reducing costs, and improving detection efficiency. Through distributed monitoring, bandwidth utilization can be improved, avoiding network congestion or latency, enabling rapid response to detection tasks, improving alarm timeliness, reducing computing power requirements, and saving computing resources. Another objective of this invention is to provide a fault location device for a flexible DC-DC converter valve. A further objective of this invention is to provide a computer-readable medium. A final objective of this invention is to provide a computer device.

[0004] To achieve the above objectives, this invention discloses a fault location method for a flexible DC converter valve, comprising:

[0005] Acquire sensor monitoring data from the subsystem of the flexible DC converter valve;

[0006] The residual results of the subsystem are generated by a pre-built residual generator based on sensor monitoring data and the obtained subsystem state estimates.

[0007] Based on the pre-generated detection threshold, residual results, and pre-acquired covariance matrix, fault detection is performed on each subsystem to obtain the system fault range, which includes multiple subsystems.

[0008] By using a pre-built probabilistic graphical model, local fault location is performed based on sensor monitoring data of subsystems within the acquired system fault range, thus obtaining the fault root node.

[0009] Preferably, the sensor monitoring data of the subsystem for acquiring the flexible DC converter valve includes:

[0010] Acquire raw sensor data;

[0011] According to the preset feature rules, the raw sensor data is preprocessed to obtain sensor monitoring data. The data preprocessing includes data sampling, data deduplication, missing value handling, and data normalization.

[0012] Preferably, the sensor monitoring data includes the output matrix of the local subsystem, the output matrix of the neighboring subsystem, and the input matrix of the local subsystem;

[0013] Before generating the residual results of the subsystem using a pre-built residual generator based on sensor monitoring data and acquired subsystem state estimates, the process also includes:

[0014] Acquire historical sensor data and simulation data of the flexible DC converter valve subsystem;

[0015] Based on sensor monitoring data, sensor historical data, and simulation data, the state of the subsystem is estimated to obtain the subsystem state estimate.

[0016] A residual generator is constructed based on the subsystem state estimate, the output matrix of the local subsystem, the output matrix of the neighboring subsystem, and the input matrix of the local subsystem.

[0017] Preferably, based on the pre-generated detection threshold, residual results, and pre-acquired covariance matrix, fault detection is performed on each subsystem to obtain the system fault range. The system fault range includes multiple subsystems, including:

[0018] Based on the residual results and covariance matrix, generate detection statistics;

[0019] If the detection statistic is less than or equal to the detection threshold, the system fault range is determined to be empty.

[0020] If the detection statistic is greater than the detection threshold, the system fault range is determined to be the local subsystem and the neighboring subsystem.

[0021] Preferably, before performing fault detection on each subsystem based on pre-generated detection thresholds, residual results, and pre-acquired covariance matrices to obtain the system fault range, and before the system fault range includes multiple subsystems, the following steps are also included:

[0022] Obtain the residual observation matrix for each subsystem;

[0023] Generate the covariance matrix based on the residual observation matrix.

[0024] Preferably, before performing fault detection on each subsystem based on pre-generated detection thresholds, residual results, and pre-acquired covariance matrices to obtain the system fault range, and before the system fault range includes multiple subsystems, the following steps are also included:

[0025] A chi-square test is performed on the residual observation matrix and the corresponding residual results of each subsystem to generate the detection threshold for the subsystem.

[0026] Preferably, the probabilistic graphical model is a directed acyclic graph, which includes multiple subsystems and the causal relationships between the subsystems;

[0027] By using a pre-constructed probabilistic graphical model, local fault location is performed based on sensor monitoring data of subsystems within the acquired system fault range, resulting in the fault root node, including:

[0028] Based on sensor detection data, generate a Bayesian information content index;

[0029] If the Bayesian information content index is greater than the preset index confidence line, start from the current subsystem and reason backward along the directed acyclic graph until the root node of the fault is located.

[0030] If the Bayesian information content index is less than or equal to the index confidence line, the current subsystem is determined to be fault-free.

[0031] This invention also discloses a fault location device for a flexible DC converter valve, comprising:

[0032] The monitoring data acquisition unit is used to acquire sensor monitoring data of the flexible DC converter valve subsystem;

[0033] The residual result generation unit is used to generate the residual results of the subsystem based on sensor monitoring data and the acquired subsystem state estimates using a pre-built residual generator.

[0034] The fault detection unit is used to perform fault detection on each subsystem based on the pre-generated detection threshold, residual results and pre-acquired covariance matrix, and obtain the system fault range, which includes multiple subsystems.

[0035] The fault location unit is used to locate local faults and obtain the fault root node by using a pre-built probabilistic graphical model based on sensor monitoring data of subsystems within the acquired system fault range.

[0036] Preferably, the monitoring data acquisition unit is specifically used to acquire raw sensor data; according to preset feature rules, it performs data preprocessing on the raw sensor data to obtain sensor monitoring data. The data preprocessing includes data sampling, data deduplication, missing value handling, and data normalization.

[0037] Preferably, the sensor monitoring data includes the output matrix of the local subsystem, the output matrix of the neighboring subsystem, and the input matrix of the local subsystem;

[0038] The device also includes:

[0039] The training data acquisition unit is used to acquire historical sensor data and simulation data of the flexible DC converter valve subsystem;

[0040] The system state estimation unit is used to estimate the state of the subsystem based on sensor monitoring data, sensor historical data and simulation data, and obtain the subsystem state estimate value;

[0041] The residual generator building unit is used to build a residual generator based on the subsystem state estimate, the output matrix of the local subsystem, the output matrix of the neighboring subsystem, and the input matrix of the local subsystem.

[0042] Preferably, the fault detection unit is specifically used to generate a detection statistic based on the residual result and the covariance matrix; if the detection statistic is less than or equal to the detection threshold, the system fault range is determined to be empty; if the detection statistic is greater than the detection threshold, the system fault range is determined to be the local subsystem and the neighboring subsystem.

[0043] Preferably, the device further includes:

[0044] The observation matrix acquisition unit is used to acquire the residual observation matrix for each subsystem;

[0045] The covariance matrix generation unit is used to generate the covariance matrix based on the residual observation matrix.

[0046] Preferably, the device further includes:

[0047] The chi-square test unit is used to perform a chi-square test on the residual observation matrix and the corresponding residual results of each subsystem, and generate the detection threshold of the subsystem.

[0048] Preferably, the probabilistic graphical model is a directed acyclic graph, which includes multiple subsystems and the causal relationships between the subsystems;

[0049] The fault location unit is specifically used to generate a Bayesian information content index based on sensor detection data. If the Bayesian information content index is greater than the preset index confidence line, it takes the current subsystem as the starting point and reverses the reasoning along the directed acyclic graph until the fault root node is located. If the Bayesian information content index is less than or equal to the index confidence line, it is determined that the current subsystem is fault-free.

[0050] The present invention also discloses a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.

[0051] The present invention also discloses a computer device, including a memory and a processor, wherein the memory is used to store information including program instructions, and the processor is used to control the execution of the program instructions, wherein the processor executes the program to implement the method described above.

[0052] The present invention also discloses a computer program product, including a computer program / instruction, which, when executed by a processor, implements the method described above.

[0053] This invention acquires sensor monitoring data from the subsystems of a flexible DC converter valve; generates residual results for the subsystems based on the sensor monitoring data and the acquired subsystem state estimates using a pre-built residual generator; performs fault detection on each subsystem based on the pre-generated detection threshold, residual results, and pre-acquired covariance matrix to obtain the system fault range, which includes multiple subsystems; uses a pre-built probabilistic graphical model to locate local faults based on the sensor monitoring data of the subsystems within the acquired system fault range, obtaining the fault root node; and employs a three-layer architecture of distributed fault monitoring and tracing in a cloud-edge-device collaborative environment to perform real-time fault location for the flexible DC converter valve, avoiding missed and false detections, reducing costs, and improving detection efficiency; the distributed monitoring method improves bandwidth utilization, avoids network congestion or latency, enables rapid response to detection tasks, improves alarm timeliness, reduces computing power requirements, and saves computing resources. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 A flowchart illustrating a fault location method for a flexible DC converter valve provided in an embodiment of the present invention;

[0056] Figure 2 A flowchart illustrating another fault location method for a flexible DC converter valve provided in an embodiment of the present invention;

[0057] Figure 3 This is a schematic diagram of the structure of a distributed subsystem provided in an embodiment of the present invention;

[0058] Figure 4 A schematic diagram of the structure of a fault location device for a flexible DC converter valve provided in an embodiment of the present invention;

[0059] Figure 5This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] To facilitate understanding of the technical solution provided in this application, the relevant content of the technical solution will be explained below. This invention constructs a three-layer architecture for distributed fault monitoring and tracing of flexible DC converter valves in a "cloud-edge-device" collaborative environment oriented towards complex manufacturing processes.

[0062] This technology enables dimensionality reduction and decomposition of complex data on the edge, allowing for efficient processing and analysis of large-scale and complex data on resource-constrained devices. Dimensionality reduction and decomposition techniques help reduce the dimensionality of data, extract key features, and remove noise, thereby simplifying problems and accelerating computation.

[0063] Local process monitoring of the subsystem is performed at the "edge," collecting monitoring data generated by sensors and processing and analyzing it in real time on edge devices. A data acquisition system is used to collect data, and an edge computing platform is used to process the data, calculating key indicators and features. Based on the collected data, a predictive model or classifier for the performance of the monitoring subsystem is established.

[0064] By integrating information from multiple subsystems on the cloud side, a global integrated diagnosis can be achieved, enabling global performance monitoring, fault prediction, and diagnosis of the entire system.

[0065] The implementation process includes system decomposition based on monitoring performance optimization, construction of a causal probabilistic graphical model, and a distributed fault diagnosis and tracing process based on the probabilistic graphical model. First, with the goal of optimizing system monitoring performance, a process optimization objective function driven by the fusion of process data and knowledge is constructed. Constraints are determined based on the correspondence between variables and their respective subsystems, as determined by known physical analysis. Combining historical datasets and known power system mechanisms, an optimized decomposition is achieved to optimize the monitoring performance of the local subsystem at the equipment "end" side.

[0066] The following uses a fault location device for a flexible-DC converter valve as an example to illustrate the implementation process of the fault location method for a flexible-DC converter valve provided in this embodiment of the invention. It is understood that the execution subject of the fault location method for a flexible-DC converter valve provided in this embodiment of the invention includes, but is not limited to, a fault location device for a flexible-DC converter valve.

[0067] Figure 1 A flowchart illustrating a fault location method for a flexible DC converter valve provided in an embodiment of the present invention is shown below. Figure 1 As shown, the method includes:

[0068] Step 101: Obtain sensor monitoring data of the flexible DC converter valve subsystem.

[0069] Step 102: Using a pre-built residual generator, generate the residual results of the subsystem based on the sensor monitoring data and the obtained subsystem state estimates.

[0070] Step 103: Based on the pre-generated detection threshold, residual results and pre-acquired covariance matrix, perform fault detection on each subsystem to obtain the system fault range, which includes multiple subsystems.

[0071] Step 104: Using a pre-built probabilistic graphical model, local fault location is performed based on sensor monitoring data of subsystems within the acquired system fault range to obtain the fault root node.

[0072] The technical solution provided in this invention involves acquiring sensor monitoring data of a subsystem of a flexible DC converter valve; generating residual results for the subsystem based on the sensor monitoring data and the acquired subsystem state estimates using a pre-built residual generator; performing fault detection on each subsystem based on the pre-generated detection threshold, residual results, and pre-acquired covariance matrix to obtain the system fault range, which includes multiple subsystems; using a pre-built probabilistic graphical model, performing local fault location based on the sensor monitoring data of the subsystems within the acquired system fault range to obtain the fault root node; and employing a three-layer architecture of distributed fault monitoring and tracing in a cloud-edge-device collaborative environment to perform real-time fault location of the flexible DC converter valve, avoiding missed detections and false detections, reducing costs, and improving detection efficiency. The distributed monitoring method improves bandwidth utilization, avoids network congestion or latency, enables rapid response to detection tasks, improves alarm timeliness, reduces computing power requirements, and saves computing resources.

[0073] Figure 2 A flowchart illustrating another fault location method for a flexible DC converter valve provided in this embodiment of the invention is shown below. Figure 2 As shown, the method includes:

[0074] Step 201: Obtain raw sensor data.

[0075] In this embodiment of the invention, each step is performed by the fault location device of the flexible DC converter valve.

[0076] In this embodiment of the invention, the raw data monitored by the sensors at the end are acquired, including but not limited to optical CT data, vibration sensor data, and infrared thermometry data. The data format is numerical or image.

[0077] Step 202: According to the preset feature rules, perform data preprocessing on the raw sensor data to obtain sensor monitoring data. Data preprocessing includes data sampling, data deduplication, missing value handling, and data normalization.

[0078] In this embodiment of the invention, feature rules can be set according to system status and side requirements. For example, feature rules include a sampling strategy, which is to sparsely sample the original sensor data when the system is normal and oversample the original sensor data when the system is abnormal, thereby increasing the sampling frequency.

[0079] In this embodiment of the invention, the sensor monitoring data includes, but is not limited to, vibration sensor data and infrared temperature measurement data. The sensor monitoring data includes the output matrix of the local subsystem, the output matrix of the neighboring subsystem, and the input matrix of the local subsystem.

[0080] In this embodiment of the invention, data preprocessing of the original sensor data according to feature rules can improve data quality, better meet the computational needs of the side, and thus improve the accuracy of subsequent fault location.

[0081] Step 203: Obtain historical sensor data and simulation data of the flexible DC converter valve subsystem.

[0082] In this embodiment of the invention, the database stores historical sensor data of the flexible DC converter valve subsystem, and this historical sensor data can be retrieved from the database. The historical sensor data includes, but is not limited to, optical CT data, vibration sensor data, and infrared thermography data. The data format is numerical or image.

[0083] In this embodiment of the invention, the simulation data is generated by a simulation model established using the converter valve mechanism and a hardware-in-the-loop simulation model.

[0084] It is worth noting that the subsystem is obtained by decomposing the local system at the device end. The objective function is constructed by optimizing the process using real-time monitoring data and historical data, with the goal of achieving optimal system monitoring performance. Constraints are determined based on the correspondence between variables and their respective subsystems, as determined by known physical analysis. Combining historical datasets and known power system mechanisms, the optimal decomposition for monitoring the local subsystem at the device end is achieved. That is, the large objective function is decomposed into smaller sub-objective functions based on historical datasets and known power system mechanisms. This is accomplished by modeling different variables and constraints separately and then combining them into a unified overall objective function.

[0085] Step 204: Based on sensor monitoring data, sensor historical data, and simulation data, estimate the state of the subsystem to obtain the subsystem state estimate.

[0086] In this embodiment of the invention, sensor monitoring data, sensor historical data, and simulation data are actual data covering the operation of the subsystem under different conditions. Based on the sensor monitoring data, sensor historical data, and simulation data, the state of the subsystem is identified, and an estimated state value of the subsystem is obtained.

[0087] Step 205: Construct a residual generator based on the subsystem state estimate, the output matrix of the local subsystem, the output matrix of the neighboring subsystem, and the input matrix of the local subsystem.

[0088] Figure 3 This is a schematic diagram of the structure of a distributed subsystem provided in an embodiment of the present invention, such as... Figure 3 As shown, taking two subsystems as an example, namely system j and system i, system j corresponds to sensor j, system i corresponds to sensor i, sensor j corresponds to detection unit j, and sensor i corresponds to detection unit i. Figure 3 As shown, x j Let x be the process state matrix of system j. i Let u be the process state matrix of system i. j Let y be the input matrix of system j. j Let u be the output matrix of system j. i Let y be the input matrix of system i. i Let be the output matrix of system i.

[0089] Monitoring methods based on subspace identification take system dynamics into account, enabling them to handle highly dynamic and closed-loop systems without requiring complex and cumbersome detection system designs. A distributed anomaly detection algorithm based on subspace identification is constructed, but to establish the connection between the mechanistic model and process data, it does not consider… Figure 3 To neutralize output noise in the system, assuming the system is a stochastic linear system, the input-output (IO) data model can be built on the i-th subsystem as shown below:

[0090]

[0091] x(t+1)=Ax(t)+Bu(t)

[0092] y(t)=Cx(t)+Du(t)

[0093]

[0094]

[0095] Where A, B, C, and D are system matrices representing the response dimensions. The output Hankel matrix of the local subsystem. For augmented matrix, This is the process state matrix of the local subsystem. All are Toeplitz matrices. The output Hankel matrix of the neighboring subsystem. The output noise Hankel matrix of the neighboring subsystem. The Hankel matrix represents the completely unknown portion of the neighboring subsystem. The Hankel matrix is ​​the input to the local subsystem. This is the output noise Hankel matrix of the local subsystem.

[0096] Based on the above IO model, the residual generator of the fault detection unit corresponding to the local subsystem is derived:

[0097]

[0098] Where, r i (k) is the residual generator corresponding to local subsystem i. Let i be the state estimate of subsystem i. The output Hankel matrix of the local subsystem. The output Hankel matrix of the neighboring subsystem. The input Hankel matrix is ​​the local subsystem.

[0099] Step 206: Using the residual generator, generate the residual results of the subsystem based on the sensor monitoring data and the obtained subsystem state estimates.

[0100] In this embodiment of the invention, the sensor monitoring data is the actual data collected during system operation, including but not limited to the state estimates, output matrices, input matrices, and output matrices of neighboring subsystems of each subsystem.

[0101] Specifically, each subsystem sends the output matrix of its local subsystem, the output matrix of its neighboring subsystem, and the input matrix of its local subsystem to the edge side; the detection unit on the edge side inputs the subsystem state estimate, the received output matrix of the local subsystem of a certain subsystem, the output matrix of the neighboring subsystem, and the input matrix of the local subsystem into the residual generator to perform residual calculation and generate the residual result of the subsystem.

[0102] Step 207: Obtain the residual observation matrix for each subsystem.

[0103] In this embodiment of the invention, the residual signal of each subsystem under fault-free conditions is acquired, and a residual observation matrix R is generated. The residual observation matrix has n observations, each observation has p variables, and the residual observation matrix R is an n-row p-column matrix.

[0104] Step 208: Generate the covariance matrix based on the residual observation matrix.

[0105] Specifically, through The covariance matrix is ​​generated by calculating the residual observation matrix. Where ∑res i Let R be the covariance matrix of the local residual signal under fault-free conditions of subsystem i, and let R be the residual observation matrix.

[0106] Step 209: Perform a chi-square test on the residual observation matrix and the corresponding residual results for each subsystem to generate the detection threshold for the subsystem.

[0107] In this embodiment of the invention, J ith =χ 2 (i) Perform a chi-square test on the residual observation matrix and corresponding residual results of each subsystem to generate the detection threshold for the subsystem. Where J ith χ is the detection threshold for subsystem i. 2 (i) represents the residual observation matrix of subsystem i and the chi-square test value of the corresponding residual results.

[0108] Step 210: Generate detection statistics based on the residual results and covariance matrix.

[0109] Specifically, through J i =r i T (k)∑ -1 res i r i (k), where J i For the detection statistic, r i (k) represents the residual result, r i T (k) is the transpose matrix of the residual results, ∑ -1 res i is the inverse of the covariance matrix.

[0110] Step 211: Determine whether the detection statistic is less than or equal to the detection threshold. If yes, proceed to step 212; otherwise, proceed to step 213.

[0111] Specifically, if the detection statistic is less than or equal to the detection threshold, it indicates that subsystem i and its neighboring subsystems are fault-free, and step 212 is executed; if the detection statistic is greater than the detection threshold, it indicates that subsystem i or its neighboring subsystems are faulty, and step 213 is executed.

[0112] Step 212: Determine that the system fault range is empty, and continue to execute step 201.

[0113] In this embodiment of the invention, if the subsystem and its neighboring subsystems are fault-free, the system fault range is determined to be empty, and step 201 is continued to achieve real-time monitoring.

[0114] Step 213: Determine the scope of the system fault as the local subsystem and the neighboring subsystem.

[0115] In this embodiment of the invention, if a subsystem has a fault or a neighboring subsystem of the subsystem has a fault, the current local subsystem and its neighboring subsystems are classified as system fault ranges.

[0116] Step 214: Using a pre-built probabilistic graphical model, local fault location is performed based on sensor monitoring data of subsystems within the acquired system fault range to obtain the fault root node.

[0117] After optimizing and decomposing the large-scale system, a multivariate causal analysis method is designed to model the causal relationships of monitored process variables in local subsystems. A directed acyclic graph (DAG) model is constructed, and the quantitative relationships between nodes are described by estimating the conditional probability density between process variables, thus establishing a subsystem model. After completing the probabilistic graphical model of the subsystem, local fault monitoring of the subsystem can be performed.

[0118] Bayesian networks combine probability and graphical models, enabling probabilistic graphical models to directly represent probabilistic relationships using graphs. For a set of data, given a distribution P(x1, x2, ..., x...)... n If modeling is used, then according to the chain rule of the conditional law, the distribution can be decomposed into: P(x1,x2,...,x...). n )=P(x1)Π i P(x i |x i-1 Therefore, we first make a basic assumption: adjacent points in a DAG have dependencies. Using this assumption, we can deduce that: P(x1,x2,...,x...). n )=P(x1)P(x2|x1)P(x3|x2,x1)P(x4|x3,x2,x1).

[0119] After completing the probabilistic graphical model of the subsystem, local fault monitoring of the subsystem can be performed; further, information fusion and system fault monitoring can be carried out in the cloud to complete fault tracing, and information fusion and fault diagnosis algorithms can be applied to locate the source of the fault.

[0120] In this embodiment of the invention, the probabilistic graphical model is a directed acyclic graph (DAG). The DAG includes multiple subsystems and the causal relationships between the subsystems. Each subsystem is represented by a node, and the causal relationships between the subsystems are represented by directed edges.

[0121] In this embodiment of the invention, the sensor monitoring data is standardized and denoted as D. norm ; using D norm Based on performance optimization, the known mechanism is decomposed into M subsystems; a causal probabilistic graphical model of the subsystems is constructed: x j =f j,2 (f i,1 (x i )+e j ), x i It is x j The parent node, e j Let f be the noise interference, and f be the conditional probability density.

[0122] Furthermore, statistical information is calculated from the sensor monitoring data using a probabilistic graphical model to obtain the corresponding statistical information. It is then determined whether the statistical information falls within a pre-generated confidence interval. If so, the sensor monitoring data is considered reliable, and the Bayesian information content index is calculated. If not, the sensor monitoring data is considered unreliable, and the process ends. The statistical information S monitored by the design subsystem is then considered. new for:

[0123]

[0124]

[0125]

[0126] Among them, C - For statistical information S new The lower confidence limit, C + For statistical information S new The upper confidence limit, Kernel density estimation represents the statistical information corresponding to the sensor monitoring data. This represents the mean value corresponding to the sensor monitoring data. The variance corresponding to the sensor monitoring data is z, which is the standard score. The value of z can be obtained by looking up the score table based on the confidence level (usually set to 95%). n is the number of samples, which can be used to monitor system anomalies based on changes in probability density.

[0127] In this embodiment of the invention, step 214 specifically includes:

[0128] Step 2141: Generate Bayesian Information Content (BIC) index based on sensor monitoring data.

[0129] In this embodiment of the invention, the sensor monitoring data is real-time operational data collected online.

[0130] Specifically, through The BIC index is obtained by calculating the sensor monitoring data. Specifically, P(x) i,j |F) represents the conditional probability of the subsystem under fault conditions, x i,j | represents the i-th variable in the j-th subsystem, P(F|x) i,j ) represents the probability of the i-th variable being abnormal or malfunctioning in the j-th subsystem.

[0131] Step 2142: Determine whether the BIC indicator is greater than the preset indicator confidence line. If yes, proceed to step 2143; otherwise, proceed to step 2144.

[0132] In this embodiment of the invention, the confidence line of the indicator can be set according to actual needs, and this embodiment of the invention does not limit it.

[0133] In this embodiment of the invention, if the BIC index is greater than the index confidence line, it indicates that a system fault has been detected, and step 2143 is executed; if the BIC index is less than or equal to the index confidence line, it indicates that no system fault has occurred, and step 2144 is executed.

[0134] Step 2143: Starting from the current subsystem, reverse reasoning along the directed acyclic graph until the root node of the fault is located, and the process ends.

[0135] Specifically, starting from the root node variable of the current subsystem, reverse reasoning and monitoring are performed from bottom to top along the directed acyclic graph. The root node variables on all fault propagation paths are sequentially calculated to determine if they are in a fault state, until the fault root node x is located. root .

[0136] In this embodiment of the invention, the presence or absence of a fault is determined based on the Integrated Bayesian Inference Index (BIC), and the status of the monitoring system is identified based on changes in the BIC index to determine a comprehensive governance decision.

[0137] Step 2144: Determine that the current subsystem is fault-free.

[0138] In this embodiment of the invention, if no system failure occurs, the operation of the flexible DC converter valve continues to be monitored in real time.

[0139] This invention, based on existing intelligent monitoring algorithms, implements algorithm pruning techniques and edge deployment methods. Edge optimization enables system decomposition based on monitoring performance optimization, construction of causal probabilistic graphical models, and distributed fault diagnosis and tracing processes based on probabilistic graphical models, improving monitoring efficiency. Different diagnostic models are deployed at different spatial locations, effectively compressing the algorithm size and facilitating embedded deployment. Pruning, quantization, and decomposition techniques reduce the number and size of model parameters, improving computational efficiency. Through cloud-edge-device optimized collaborative algorithms, data processing and analysis speed can be improved. Based on task complexity and performance requirements, cloud subsystem fusion decisions are made to select the most suitable edge device for computation, maximizing computational speed.

[0140] It is worth noting that the acquisition, storage, use, and processing of data in the technical solution of this application all comply with relevant laws and regulations. The user information in the embodiments of this application was obtained through legal and compliant means, and the acquisition, storage, use, and processing of user information have been authorized and agreed upon by the client.

[0141] The technical solution of the fault location method for a flexible DC converter valve provided in this invention involves: acquiring sensor monitoring data of the subsystems of the flexible DC converter valve; generating residual results for the subsystems based on the sensor monitoring data and the acquired subsystem state estimates using a pre-built residual generator; performing fault detection on each subsystem based on the pre-generated detection threshold, residual results, and pre-acquired covariance matrix to obtain the system fault range, which includes multiple subsystems; performing local fault location based on the sensor monitoring data of the subsystems within the acquired system fault range using a pre-built probabilistic graphical model to obtain the fault root node; and using a three-layer architecture of distributed fault monitoring and tracing in a cloud-edge-device collaborative environment to perform real-time fault location for the flexible DC converter valve, avoiding missed detections and false detections, reducing costs, and improving detection efficiency. The distributed monitoring method improves bandwidth utilization, avoids network congestion or latency, enables rapid response to detection tasks, improves alarm timeliness, reduces computing power requirements, and saves computing resources.

[0142] Figure 4 This is a schematic diagram of a fault location device for a flexible DC converter valve provided in an embodiment of the present invention. This device is used to execute the aforementioned fault location method for the flexible DC converter valve. Figure 4 As shown, the device includes: a monitoring data acquisition unit 11, a residual result generation unit 12, a fault detection unit 13, and a fault location unit 14.

[0143] The monitoring data acquisition unit 11 is used to acquire sensor monitoring data of the subsystem of the flexible DC converter valve.

[0144] The residual result generation unit 12 is used to generate the residual result of the subsystem based on the sensor monitoring data and the obtained subsystem state estimate by using a pre-built residual generator.

[0145] The fault detection unit 13 is used to perform fault detection on each subsystem based on the pre-generated detection threshold, residual results and pre-acquired covariance matrix, and obtain the system fault range, which includes multiple subsystems.

[0146] The fault location unit 14 is used to locate local faults and obtain the fault root node by using a pre-built probabilistic graphical model based on the sensor monitoring data of the subsystems within the acquired system fault range.

[0147] In this embodiment of the invention, the monitoring data acquisition unit 11 is specifically used to acquire raw sensor data; according to preset feature rules, the raw sensor data is preprocessed to obtain sensor monitoring data. The data preprocessing includes data sampling, data deduplication, missing value processing, and data normalization.

[0148] In this embodiment of the invention, the sensor monitoring data includes the output matrix of the local subsystem, the output matrix of the neighboring subsystem, and the input matrix of the local subsystem; the device also includes: a training data acquisition unit 15, a system state estimation unit 16, and a residual generator construction unit 17.

[0149] The training data acquisition unit 15 is used to acquire historical sensor data and simulation data of the flexible DC converter valve subsystem.

[0150] The system state estimation unit 16 is used to estimate the state of the subsystem based on sensor monitoring data, sensor historical data and simulation data, and obtain the subsystem state estimate value.

[0151] The residual generator building unit 17 is used to build a residual generator based on the subsystem state estimate, the output matrix of the local subsystem, the output matrix of the neighboring subsystem, and the input matrix of the local subsystem.

[0152] In this embodiment of the invention, the fault detection unit 13 is specifically used to generate a detection statistic based on the residual result and the covariance matrix; if the detection statistic is less than or equal to the detection threshold, the system fault range is determined to be empty; if the detection statistic is greater than the detection threshold, the system fault range is determined to be the local subsystem and the neighboring subsystem.

[0153] In this embodiment of the invention, the device further includes: an observation matrix acquisition unit 18 and a covariance matrix generation unit 19.

[0154] The observation matrix acquisition unit 18 is used to acquire the residual observation matrix of each subsystem.

[0155] The covariance matrix generation unit 19 is used to generate a covariance matrix based on the residual observation matrix.

[0156] In this embodiment of the invention, the device further includes a chi-square test unit 20.

[0157] Chi-square test unit 20 is used to perform chi-square test on the residual observation matrix and the corresponding residual results of each subsystem to generate the detection threshold of the subsystem.

[0158] In this embodiment of the invention, the probabilistic graphical model is a directed acyclic graph, which includes multiple subsystems and the causal relationships between the subsystems;

[0159] The fault location unit 14 is specifically used to generate a Bayesian information content index based on sensor detection data; if the Bayesian information content index is greater than the preset index confidence line, it takes the current subsystem as the starting point and reverses the reasoning along the directed acyclic graph until the fault root node is located; if the Bayesian information content index is less than or equal to the index confidence line, it is determined that the current subsystem is fault-free.

[0160] In this embodiment of the invention, sensor monitoring data of the flexible DC converter valve subsystem is acquired; a residual generator is used to generate residual results for the subsystem based on the sensor monitoring data and the acquired subsystem state estimates; fault detection is performed on each subsystem based on the pre-generated detection threshold, residual results, and pre-acquired covariance matrix to obtain the system fault range, which includes multiple subsystems; a pre-built probabilistic graphical model is used to locate local faults based on the sensor monitoring data of the subsystems within the acquired system fault range, obtaining the fault root node; a three-layer architecture of distributed fault monitoring and tracing in a cloud-edge-device collaborative environment is used to perform real-time fault location of the flexible DC converter valve, avoiding missed detections and false detections, reducing costs, and improving detection efficiency; the distributed monitoring method can improve bandwidth utilization, avoid network congestion or network latency, quickly respond to detection tasks, improve alarm timeliness, reduce computing power requirements, and save computing resources.

[0161] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer device, specifically, a computer device can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0162] This invention provides a computer device, including a memory and a processor. The memory is used to store information including program instructions, and the processor is used to control the execution of the program instructions. When the program instructions are loaded and executed by the processor, they implement the steps of the above-described embodiment of the fault location method for a flexible DC converter valve. For a detailed description, please refer to the above-described embodiment of the fault location method for a flexible DC converter valve.

[0163] The following is for reference. Figure 5 It shows a schematic diagram of the structure of a computer device 600 suitable for implementing the embodiments of this application.

[0164] like Figure 5 As shown, the computer device 600 includes a central processing unit (CPU) 601, which can perform various appropriate tasks and processes based on programs stored in read-only memory (ROM) 602 or programs loaded from storage section 608 into random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of the computer device 600. The CPU 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0165] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal feedback (LCD), etc., and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed in storage section 608 as needed.

[0166] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program including program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611.

[0167] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0168] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0169] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0170] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0171] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0172] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0173] The acquisition, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.

[0174] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0175] This application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0176] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0177] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method of fault location for a LCC valve, characterized in that, The method comprises: obtaining sensor monitoring data of a subsystem of a flexible direct current valve; generating a residual result of the subsystem by a pre-constructed residual generator according to the sensor monitoring data and an obtained subsystem state estimation value; performing fault detection on each subsystem according to a pre-generated detection threshold, the residual result and a pre-obtained covariance matrix to obtain a system fault range, the system fault range comprising a plurality of subsystems; performing local fault positioning on each subsystem in the system fault range according to the obtained sensor monitoring data of the subsystem by a pre-constructed probabilistic graph model to obtain a fault root node, the probabilistic graph model being a directed acyclic graph comprising a plurality of subsystems and causal relationships between each subsystem, each subsystem being represented by a node and the causal relationships between each subsystem being represented by a directed edge; the sensor monitoring data comprising an output matrix of a local subsystem, an output matrix of a neighbor subsystem and an input matrix of the local subsystem; before the step of generating the residual result of the subsystem by the pre-constructed residual generator according to the sensor monitoring data and the obtained subsystem state estimation value, the method further comprises: obtaining sensor historical data and simulation data of the subsystem of the flexible direct current valve; estimating a state of the subsystem according to the sensor monitoring data, the sensor historical data and the simulation data to obtain a subsystem state estimation value; constructing the residual generator according to the subsystem state estimation value, the output matrix of the local subsystem, the output matrix of the neighbor subsystem and the input matrix of the local subsystem.

2. The method of fault locating of a LCC valve according to claim 1, characterized in that, the step of obtaining the sensor monitoring data of the subsystem of the flexible direct current valve comprises: obtaining sensor raw data; performing data preprocessing on the sensor raw data according to a pre-set feature rule to obtain the sensor monitoring data, the data preprocessing comprising data sampling, data deduplication, missing value processing and data normalization.

3. The method of fault locating of a LCC valve according to claim 1, characterized in that, performing fault detection on each subsystem according to a pre-generated detection threshold, the residual result and a pre-obtained covariance matrix to obtain a system fault range, the system fault range comprising a plurality of subsystems, which comprises: generating a detection statistic according to the residual result and the covariance matrix; if the detection statistic is less than or equal to the detection threshold, determining that the system fault range is empty; if the detection statistic is greater than the detection threshold, determining that the system fault range comprises the local subsystem and the neighbor subsystem.

4. The method of fault locating of a LCC valve according to claim 1, characterized in that, before the step of performing fault detection on each subsystem according to a pre-generated detection threshold, the residual result and a pre-obtained covariance matrix to obtain a system fault range, the system fault range comprising a plurality of subsystems, the method further comprises: obtaining a residual observation matrix of each subsystem; generating a covariance matrix according to the residual observation matrix.

5. The method of fault locating of a LCC valve according to claim 4, characterized in that, before the step of performing fault detection on each subsystem according to a pre-generated detection threshold, the residual result and a pre-obtained covariance matrix to obtain a system fault range, the system fault range comprising a plurality of subsystems, the method further comprises: The residual observation matrix of each subsystem and the corresponding residual result are subjected to a chi-square test to generate a detection threshold of the subsystem. 6.The method of claim 1, wherein, The local fault location is performed according to the sensor monitoring data of the subsystems in the system fault range by using the pre-constructed probabilistic graph model to obtain a fault root node, and the method comprises: A Bayesian information quantity index is generated according to the sensor monitoring data; If the Bayesian information quantity index is greater than a preset index confidence line, the current subsystem is taken as a starting point, and the directed acyclic graph is reversely reasoned until the fault root node is located; If the Bayesian information quantity index is less than or equal to the index confidence line, it is determined that the current subsystem is fault-free.

7. A fault locating device for a Litz converter valve, characterized in that The device comprises: A monitoring data acquisition unit configured to acquire sensor monitoring data of subsystems of the HVDC valve; A residual result generation unit configured to generate residual results of the subsystems by using a pre-constructed residual generator according to the sensor monitoring data and acquired state estimation values of the subsystems; A fault detection unit configured to perform fault detection on each subsystem according to a pre-generated detection threshold, the residual results and a pre-acquired covariance matrix to obtain a system fault range, wherein the system fault range comprises a plurality of subsystems; A fault location unit configured to perform local fault location according to the sensor monitoring data of the subsystems in the system fault range by using a pre-constructed probabilistic graph model to obtain a fault root node, wherein the probabilistic graph model is a directed acyclic graph, the directed acyclic graph comprises a plurality of subsystems and causal relationships between the subsystems, each subsystem is represented by a node, and the causal relationships between the subsystems are represented by directed edges. The sensor monitoring data comprises an output matrix of a local subsystem, an output matrix of a neighbor subsystem and an input matrix of the local subsystem. The device further comprises: A training data acquisition unit configured to acquire sensor historical data and simulation data of the subsystems of the HVDC valve; A system state estimation unit configured to estimate the state of the subsystems according to the sensor monitoring data, the sensor historical data and the simulation data to obtain state estimation values of the subsystems; A residual generator construction unit configured to construct the residual generator according to the state estimation values of the subsystems, the output matrix of the local subsystem, the output matrix of the neighbor subsystem and the input matrix of the local subsystem. 8.The device of claim 7, wherein The monitoring data acquisition unit is specifically configured to acquire sensor raw data, perform data preprocessing on the sensor raw data according to a preset feature rule to obtain the sensor monitoring data, and the data preprocessing comprises data sampling, data deduplication, missing value processing and data normalization. 9.The device of claim 7, wherein The fault detection unit is specifically configured to generate a detection statistic according to the residual result and a covariance matrix; and if the detection statistic is less than or equal to the detection quantity threshold, determine that the system fault range is empty. If the detection statistic is greater than the detection quantity threshold, determine that the system fault range is the local subsystem and the neighbor subsystem.

10. The fault locating device of a LIT valve according to claim 7, characterized in that, The device further comprises: An observation matrix acquisition unit configured to acquire a residual observation matrix of each subsystem; A covariance matrix generation unit configured to generate a covariance matrix according to the residual observation matrix.

11. The fault locating device of a LIT valve according to claim 10, characterized in that, The device further comprises: A chi-square test unit configured to perform chi-square test on the residual observation matrix of each subsystem and the corresponding residual result, and generate a detection quantity threshold of the subsystem.

12. The fault locating device of the LIGBT according to claim 7, wherein The fault locating unit is specifically configured to generate a Bayesian information quantity index according to the sensor monitoring data; if the Bayesian information quantity index is greater than a preset index confidence line, take the current subsystem as a starting point, and perform reverse reasoning according to the directed acyclic graph until a fault root node is located; and if the Bayesian information quantity index is less than or equal to the index confidence line, determine that the current subsystem is fault-free.

13. A computer readable medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the fault locating method of the LIGBT according to any one of claims 1 to 6.

14. A computer device comprising a memory for storing information including program instructions, and a processor for controlling execution of the program instructions, characterized in that, The program instructions are loaded and executed by the processor to implement the fault locating method of the LIGBT according to any one of claims 1 to 6.

15. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instructions are executed by the processor to implement the fault locating method of the LIGBT according to any one of claims 1 to 6.

Images