Charging device fault positioning method, system and device

Abstract

The application provides a charging equipment fault positioning method, system and device, the method comprising: obtaining time dimension data and space dimension data of the charging equipment and preprocessing and feature extraction, obtaining time dimension features and space dimension features; constructing a space-time tensor, performing causal verification on the space-time tensor, obtaining a causal relationship and a corresponding causal probability; constructing a fault causal graph according to a hierarchical allocation result of the space-time tensor, the causal relationship and the corresponding causal probability; constructing a space-time causal tensor and performing layered tensor decomposition, obtaining a factor matrix and a factor weight; obtaining an abnormal score in the fault causal graph through the factor matrix, the factor weight and the space-time causal tensor, positioning a fault, obtaining a positioning path, and then obtaining a device fault cause in combination with the corresponding factor matrix and the factor weight. The method provided by the application can accurately perform causal decomposition, and then obtain a root cause of a charging equipment fault, thereby providing a basis for equipment operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of fault location technology, and specifically to a method, system and device for fault location of charging equipment. Background Technology

[0002] With the rapid development of the global new energy vehicle industry, the scale of charging equipment construction continues to expand. Typically, charging equipment is installed in densely populated public areas, allowing users to quickly charge their devices through direct interaction. However, due to high-frequency operation, environmental exposure, and frequent user interaction, any component may malfunction under certain conditions. Therefore, charging equipment faces a high risk of failure. Equipment failure not only affects user experience but may also threaten the safety of equipment and personnel. Efficient and accurate fault location methods for charging equipment are crucial for ensuring the safe and reliable operation of charging equipment.

[0003] Existing methods for locating charging equipment faults have the following problems: 1. They rely on historical operating status data, but the data dimensions are relatively singular. Fault diagnosis is based solely on the charging equipment's own data, easily overlooking external factors such as the geographical distribution of charging equipment or the behavioral habits of charging users. 2. They rely on simple fault codes and equipment models for isolated diagnosis, lacking joint judgment based on multi-dimensional information within the charging equipment's usage scenario. They cannot handle the cross-influence of multiple variables on faults, only providing superficial diagnostic results and failing to locate the source of the fault. This can easily lead to errors in fault cause location due to the singular data dimensions. For example, if frequent charging equipment failures are caused by temperature issues in a specific area, and since charging equipment in the same area is generally of the same model, traditional methods may misjudge the fault as being caused by a problem with that specific model of charging equipment, thus failing to locate the root cause of the charging equipment failure and providing incorrect maintenance guidance. 3. The post-fault playback diagnostic mode makes it impossible to analyze the cause of the fault, causing maintenance delays. By the time maintenance is performed, the equipment fault may have escalated, affecting equipment maintenance efficiency and user experience. Summary of the Invention

[0004] This invention addresses the shortcomings of existing fault location methods that cannot accurately locate faults by providing a method, system, and apparatus for fault location in charging equipment.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A method for locating faults in charging equipment includes the following steps: Obtain fault data of charging equipment, extract time dimension data and spatial dimension data from the fault data of charging equipment, and perform preprocessing and feature extraction to obtain time dimension features and spatial dimension features. Acquire basic information related to the charging equipment and fault description data from the fault data, construct a spatiotemporal tensor by combining the time dimension features and spatial dimension features, and determine the causal relationship and corresponding causal probability by combining the historical equipment fault dataset. The spatiotemporal tensor is hierarchically allocated using the causal relationships, and a fault causal graph is constructed by combining the causal relationships and causal probabilities. Calculate the path integral in the fault causal graph, obtain the spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and perform hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain the factor matrix and factor weights. Anomaly scores are obtained based on the factor matrix, factor weights, and spatiotemporal causal tensor. Faults are located based on the anomaly scores, fault description data, and fault causal graph, and the location path is obtained. The cause of equipment failure is determined by combining the factor matrix and factor weights.

[0007] As one possible implementation, the step of extracting time-dimensional and spatial-dimensional data from the fault data of the charging device, and performing preprocessing and feature extraction to obtain time-dimensional and spatial-dimensional features includes the following steps: Based on the fault data of the charging equipment, the fault occurrence time is obtained. A preset time window is established, and the equipment operation parameters during the operation of the equipment within the event window are extracted based on the fault occurrence time. Combined with the time series of equipment operation, time dimension data is formed. The time dimension data includes electrical parameters and equipment temperature parameters. Acquire spatial dimension data of the charging device, the spatial dimension data including device location data and time series; Preprocessing is performed on time-dimensional data and spatial-dimensional data to obtain preprocessed time-dimensional data and spatial-dimensional data. The preprocessing includes one or more of the following: deduplication, outlier removal, and missing value processing. By using the time series of device operation, the preprocessed time dimension data and spatial dimension data are aligned and correlated to obtain aligned time dimension data and spatial dimension data. Feature extraction is performed on the aligned time dimension data and spatial dimension data respectively to obtain time dimension features and spatial dimension data. The time dimension features include device statistical features, change trend features and data form features, and the spatial dimension features include device coding features and device density features.

[0008] As one possible implementation, the steps of acquiring basic information related to the charging device and fault description data from the fault data, constructing a spatiotemporal tensor by combining the time and spatial dimension features, and determining causal relationships and corresponding causal probabilities by combining historical device fault datasets include the following steps: Acquire basic information about the charging device, including charging user information, charging device attributes, and environmental parameters, and construct a spatiotemporal tensor by combining the time dimension features and spatial dimension features. Based on prior knowledge of the charging device, causal relationships between elements in each dimension of the spatiotemporal tensor are extracted. The prior knowledge includes thermal-related knowledge, electrical-related knowledge, usage-related knowledge, and device attribute-related knowledge. By obtaining the corresponding causal and outcome features through causal relationships, and then filtering based on the causal relationships between the causal and outcome features, mixed features are obtained. The sample number of result features under confounding features and causal features is obtained from historical equipment failure data to obtain the first sample number. The total sample number of confounding features and causal features is obtained to obtain the second sample number. The conditional probability is obtained based on the first sample number and the second sample number. The marginal probability is obtained by using the current number of promiscuous feature samples and the total number of promiscuous feature samples. The causal probability between the current causal feature and the result feature is obtained based on conditional probability and marginal probability. The causal probability is:

[0009] in, Represents causal probability. Represents conditional probability. Represents marginal probability. Indicates causal characteristics. Indicates the characteristics of the result. Indicates mixed characteristics, Indicates the first A hybrid feature, Indicates the number of hybrid features.

[0010] As one possible implementation, the hierarchical allocation of the spatiotemporal tensor through the causal relationship, and the construction of a fault causal graph by combining the causal relationship and causal probability, includes the following steps: The causal relationships are used to assign each dimension of the spatiotemporal tensor to a causal hierarchy, resulting in a hierarchy assignment result that includes a root cause layer, an attribute layer, a state layer, and a fault layer. Based on the hierarchical allocation results and causal relationships, a causal hierarchical path is formed, and a fault causal graph is formed based on the causal hierarchical path and the corresponding causal probability.

[0011] As one possible implementation, the path integral in the fault cause-effect graph is obtained through the following steps: Obtain the covariance and variance between causal features and outcome features of different causal paths in the fault cause-effect graph, and then obtain the effect coefficients between causal features and outcome features based on the covariance and variance; Based on the historical equipment failure dataset, the occurrence time of cause features and result features is obtained, response curves are constructed, and the response curves are evaluated to obtain time verification scores. Obtain the device location data corresponding to the cause characteristics, obtain the occurrence probability of the result characteristics within different preset distance threshold ranges based on the device location data, perform fitting based on the distance thresholds and occurrence probabilities, and obtain the spatial verification score through the fitting results; Intervention validation scores are obtained by analyzing historical maintenance data from historical equipment failure sets, identifying serious cause characteristics and outcome characteristics, and then validation scores are obtained by combining time-based and space-based validation scores. Finally, path integrals are derived based on effect coefficients and validation scores.

[0012] As one possible implementation, the process of obtaining a spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and then performing hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain a factor matrix and factor weights, includes the following steps: The path integral is mapped to the spacetime tensor to obtain the spacetime causality tensor; Based on the result of the hierarchical allocation, the decomposition direction of the spatiotemporal causal tensor is constrained, and the environmental parameters and spatial dimension features of the spatiotemporal causal tensor are decomposed to obtain the environmental factor matrix, the spatial factor matrix and the corresponding factor weights. Based on the environmental factor matrix, spatial factor matrix and corresponding factor weights, the dimensions of charging user information and charging equipment attributes are decomposed to obtain the customer factor matrix, attribute factor matrix and corresponding factor weights. The dimensions of the time dimension features are decomposed based on the environmental factor matrix, spatial factor matrix, customer factor matrix, attribute factor matrix and corresponding factor weights to obtain the time factor matrix and corresponding factor weights. The dimensions of the fault description data are decomposed based on the environmental factor matrix, spatial factor matrix, customer factor matrix, attribute factor matrix, and time factor matrix to obtain the fault factor matrix and the corresponding factor weights.

[0013] As one possible implementation, the method of obtaining anomaly scores based on the factor matrix, factor weights, and spatiotemporal causal tensor, locating faults based on the anomaly scores, fault description data, and fault causal graphs, obtaining a location path, and determining the cause of equipment faults by combining the factor matrix and factor weights includes the following steps: Based on the factor matrix and the corresponding factor weights, tensor reconstruction is performed to obtain the reconstructed causal tensor. Statistical analysis is performed by reconstructing the causal tensor and the spatiotemporal causal tensor to obtain statistical residual results. Anomaly scores are obtained based on the statistical residual results. The statistical residual results include residuals, residual mean, and residual standard deviation. Based on the anomaly score, the fault location path is obtained through fault location and path backtracking in the fault cause-effect graph; Calculate the comprehensive weight of the positioning path, combine it with the anomaly score to obtain a comprehensive score, and sort the data based on the comprehensive score to determine the cause of the equipment failure.

[0014] A fault location system for charging equipment, comprising: The feature extraction module acquires fault data of the charging equipment, extracts time dimension data and spatial dimension data based on the fault data of the charging equipment, and performs preprocessing and feature extraction to obtain time dimension features and spatial dimension features. The causal verification module acquires basic information related to the charging equipment and fault description data from the fault data, constructs a spatiotemporal tensor by combining the time dimension features and spatial dimension features, and determines the causal relationship and corresponding causal probability by combining the historical equipment fault dataset. The causal graph construction module allocates the spatiotemporal tensor hierarchically through the causal relationships and constructs a fault causal graph by combining the causal relationships and causal probabilities. The tensor decomposition module calculates the path integral in the fault causal graph, obtains the spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and performs hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain the factor matrix and factor weights. The fault location module obtains anomaly scores based on the factor matrix, factor weights, and spatiotemporal causal tensor. It then locates the fault based on the anomaly scores, fault description data, and fault causal graph, obtains the location path, and determines the cause of the equipment fault by combining the factor matrix and factor weights.

[0015] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in any one of the following ways: Obtain fault data of charging equipment, extract time dimension data and spatial dimension data from the fault data of charging equipment, and perform preprocessing and feature extraction to obtain time dimension features and spatial dimension features. Acquire basic information related to the charging equipment and fault description data from the fault data, construct a spatiotemporal tensor by combining the time dimension features and spatial dimension features, and determine the causal relationship and corresponding causal probability by combining the historical equipment fault dataset. The spatiotemporal tensor is hierarchically allocated using the causal relationships, and a fault causal graph is constructed by combining the causal relationships and causal probabilities. Calculate the path integral in the fault causal graph, obtain the spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and perform hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain the factor matrix and factor weights. Anomaly scores are obtained based on the factor matrix, factor weights, and spatiotemporal causal tensor. Faults are located based on the anomaly scores, fault description data, and fault causal graph, and the location path is obtained. The cause of equipment failure is determined by combining the factor matrix and factor weights.

[0016] A fault location device for charging equipment includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the method described in any one of the following ways: Obtain fault data of charging equipment, extract time dimension data and spatial dimension data from the fault data of charging equipment, and perform preprocessing and feature extraction to obtain time dimension features and spatial dimension features. Acquire basic information related to the charging equipment and fault description data from the fault data, construct a spatiotemporal tensor by combining the time dimension features and spatial dimension features, and determine the causal relationship and corresponding causal probability by combining the historical equipment fault dataset. The spatiotemporal tensor is hierarchically allocated using the causal relationships, and a fault causal graph is constructed by combining the causal relationships and causal probabilities. Calculate the path integral in the fault causal graph, obtain the spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and perform hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain the factor matrix and factor weights. Anomaly scores are obtained based on the factor matrix, factor weights, and spatiotemporal causal tensor. Faults are located based on the anomaly scores, fault description data, and fault causal graph, and the location path is obtained. The cause of equipment failure is determined by combining the factor matrix and factor weights.

[0017] This invention, by adopting the above technical solutions, has significant technical effects: This invention extracts time-dimensional and spatial-dimensional data within a preset time window from the fault data of the device to be located, performs preprocessing and feature extraction to obtain time-dimensional and spatial-dimensional features, and then constructs a spatiotemporal tensor. Causal verification is used to obtain causal relationships and causal probabilities, and a fault causal graph is constructed. Based on the fault causal graph and hierarchical allocation results, hierarchical tensor decomposition is performed to obtain factor matrices and factor weights. Through tensor reconstruction and fault localization, the localization path is obtained, thereby determining the cause of the device fault. This application first avoids the problem of biased fault localization results caused by a single data dimension by collecting multidimensional data affecting charging equipment faults and forming a spatiotemporal tensor. Second, by constructing a fault causal graph, this application comprehensively analyzes the causal relationships between multidimensional data to uncover the root cause of charging equipment faults, rather than merely focusing on superficial fault diagnosis. Finally, this application responds in real time when a fault occurs, quickly locating the cause of the fault, improving operation and maintenance efficiency and user experience. Attached Figure Description

[0018] 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.

[0019] Figure 1 This is a schematic flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the modules of the system of the present invention. Detailed Implementation

[0020] 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, not all, of the embodiments of the present invention. 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.

[0021] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0022] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0023] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0024] Example 1: A method for locating faults in charging equipment, such as Figure 1 As shown, it includes the following steps: S100: Obtain fault data of the charging equipment, extract time dimension data and spatial dimension data based on the fault data of the charging equipment, and perform preprocessing and feature extraction to obtain time dimension features and spatial dimension features. S200: Obtain basic information related to the charging equipment and fault description data in the fault data; construct a spatiotemporal tensor by combining the time dimension features and spatial dimension features; and determine the causal relationship and corresponding causal probability by combining the historical equipment fault dataset. S300. The spatiotemporal tensor is hierarchically allocated according to the causal relationship, and a fault causal graph is constructed by combining the causal relationship and causal probability. S400. Calculate the path integral in the fault causal graph, obtain the spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and perform hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain the factor matrix and factor weights. S500. Based on the factor matrix, factor weights, and spatiotemporal causal tensor, anomaly scores are obtained. Based on the anomaly scores, fault description data, and fault causal graph, the fault is located, the location path is obtained, and the cause of the equipment fault is determined by combining the factor matrix and factor weights.

[0025] In this embodiment, by preprocessing the time-dimensional and spatial-dimensional data of the equipment operation process within a preset time window of the equipment failure event, time-dimensional features and spatial-dimensional features are obtained, historical equipment failure datasets are acquired, causal relationships and corresponding causal probabilities are obtained through spatiotemporal causal verification, a spatiotemporal tensor is constructed, and the spatiotemporal tensor is hierarchically allocated based on the causal relationships to construct a failure causal graph. Based on the path integral and spatiotemporal tensor of the failure causal graph, hierarchical tensor decomposition is performed to obtain the factor matrix and factor weights, thereby obtaining the anomaly score and locating the cause of the equipment failure.

[0026] S100. Obtain fault data of the charging equipment, extract time dimension data and spatial dimension data based on the fault data of the charging equipment, and perform preprocessing and feature extraction to obtain time dimension features and spatial dimension features, including the following steps: S1001: Based on the fault occurrence time in the fault data, obtain the equipment operating parameters within a preset time window to form time-dimensional data; In this embodiment, the fault data is a simple description of the fault event submitted by the user to the charging equipment platform after discovering the fault. The fault data includes the fault occurrence time and fault description data. The charging equipment platform is used to monitor and record all equipment operating parameters during the operation of the charging equipment. First, the fault occurrence time is extracted based on the fault data of the charging equipment. A preset time window is defined, and the equipment operating parameters recorded by the charging equipment platform before or after the fault occurrence time are extracted with the fault occurrence time as the time origin. A time series is formed by the timestamps of the equipment operating parameters. Based on the equipment operating parameters and the corresponding time series, time dimension data is formed. The equipment operating parameters include electrical parameters and equipment temperature parameters. In this embodiment, the electrical parameters include voltage parameters, current parameters, and power parameters, etc. The time series is the occurrence time of the corresponding equipment operating parameters recorded in the charging equipment platform.

[0027] S1002: Obtain spatial dimension data of the charging device based on its location; In real-world scenarios, the location of a charging device can be one of the causes of its malfunction. For example, high temperatures can cause the charging device to overheat, leading to a malfunction. Therefore, in this embodiment, spatial dimension data is extracted based on the location of the charging device. This spatial dimension data includes device location data and a time series formed by corresponding timestamps.

[0028] S1003: Preprocess and align the time dimension data and spatial dimension data to obtain aligned time dimension data and spatial dimension data; In this embodiment, preprocessing includes one or more of the following: deduplication, outlier removal, and missing value processing. In this embodiment, if a timestamp is empty, it indicates an error occurred during data acquisition and is considered an outlier, requiring removal. Additionally, if device operating parameters show abrupt changes, these are also considered outliers and need to be removed. The removed time-dimension and spatial-dimension data are then processed for missing values. Missing value processing in this embodiment includes linear interpolation and data supplementation based on historical periods. The preprocessed time dimension data and spatial dimension data are aligned and associated to obtain aligned time dimension data and spatial dimension data. In the data alignment and association process of this embodiment, the time series of the preprocessed time dimension data and spatial dimension data are first extracted respectively. The preprocessed time dimension data and spatial dimension data with the same timestamp in the time series are aligned and associated by timestamp to obtain aligned time dimension data and spatial dimension data.

[0029] S1004: Calculate time dimension features based on the aligned time dimension data, wherein the time dimension features include device statistical features, change trend features and data form features; This embodiment calculates device statistical characteristics and trend characteristics based on aligned time-dimensional data within a preset time window. The preset time window is divided using a preset sliding window. The device statistical characteristics include the mean, standard deviation, maximum, and minimum values ​​of device operating parameters within the time-dimensional data of several sliding windows. The trend characteristics include voltage change rate and current change rate. Furthermore, this embodiment constructs change curves for each parameter based on the aligned time-dimensional data and obtains data morphology characteristics. The device statistical characteristics, trend characteristics, and data morphology characteristics together form the time-dimensional characteristics.

[0030] S1005: Calculate the spatial dimension features based on the aligned spatial dimension data. The spatial dimension features include device coding features and device density features. This embodiment obtains device location data from the aligned spatial dimension data, and encodes the device location data and device identifiers to obtain device coding features. Further, device density features are calculated using the device location data. In this embodiment, the device density feature represents the deployment density of charging devices, obtained by considering the area of ​​a defined region and the number of charging devices within that region. The spatial dimension features are formed based on the device coding features and the device density features.

[0031] S200. Obtain basic information related to the charging equipment and fault description data from the fault data; construct a spatiotemporal tensor by combining the time dimension features and spatial dimension features; and determine causal relationships and corresponding causal probabilities by combining historical equipment fault datasets. This specifically includes the following steps: S2001: Obtain basic information related to the charging equipment and fault description data from the fault data; In this embodiment of the invention, the basic information related to the charging device includes charging user information, charging device attributes, and environmental parameters. Specifically, the charging user information includes at least user type and user order information; the device attributes include at least device model; and the environmental parameters include at least temperature and humidity. Fault description data is extracted based on fault data submitted by the user on the charging device platform, and the fault description data includes at least the fault symptoms.

[0032] S2002: Construct a multidimensional tensor based on the basic information, fault description data, time dimension features and spatial dimension features, obtain the equipment failure rate under the intersection of the corresponding positions in the multidimensional tensor, and form a spatiotemporal tensor; In this embodiment, a multidimensional tensor is constructed based on charging user information, charging device attributes, environmental parameters, fault description data, time dimension features, and spatial dimension features. The device failure rate at the intersection of the corresponding positions in the multidimensional tensor is obtained to form a spatiotemporal tensor. The multidimensional tensor constitutes the dimensions of the spatiotemporal tensor, and the device failure rate is the element value of the spatiotemporal tensor.

[0033] S2003: Construct thermal-related causal chains, electrical-related causal chains, usage-related causal chains, and device attribute causal chains by combining prior knowledge of charging equipment; This invention constructs a heat-related causal chain based on prior knowledge of the working principle and fault mechanism of charging devices, such as "temperature". Equipment temperature Power module temperature Cooling fan speed "Fan failure rate," for example, if the temperature rises due to weather conditions, the charging equipment temperature rises, and consequently the power module temperature of the charging equipment rises. To better dissipate heat, the cooling fan speed will increase, leading to an increased fan failure rate. This involves constructing an electrical causal chain based on prior electrical knowledge, such as "input voltage fluctuations." Power module stress "Power module failure rate", "Input current fluctuation" Charging gun current Charging gun temperature "Charging gun failure rate" refers to factors such as power grid quality fluctuations causing input voltage or current fluctuations. The power module of the charging equipment is more sensitive to input voltage fluctuations, potentially leading to overloaded voltage or frequent charging and discharging, increasing stress on the power module and consequently increasing the failure rate over time. Input current fluctuations directly affect the charging gun's current, causing instability. Higher current and more frequent fluctuations result in more severe heating of the charging gun, leading to higher temperatures and, over time, a higher failure rate. This involves constructing a causal chain based on prior knowledge of usage-related information, such as "charging user information." Frequency of use Charging time equipment cumulative running time Component aging "Component failure rate", "usage frequency" Number of charging cycles Number of times the charging gun is plugged in and unplugged Charging gun wear "Charging gun failure rate" refers to the charging frequency and duration of ride-hailing users, which are higher than those of ordinary users. This results in greater operating time for the charging equipment and increased wear and tear on the charging gun compared to ordinary users, leading to a higher failure rate for both components and the charging gun itself. A causal chain of equipment attributes is constructed based on prior knowledge of these attributes, such as "equipment model." Design parameters of each component Inherent reliability of each component "Failure rate of each component", "Equipment model" Production batch Manufacturing quality "Early failure rate" refers to the fact that different equipment models have different module designs, production batches, and manufacturing quality, resulting in different reliability of each module and thus different possibilities for charging equipment failure.

[0034] S2004: Determine the causal relationships between the various dimensions in the spatiotemporal tensor based on the thermal causal chain, electrical causal chain, usage causal chain, and device attribute causal chain; In this embodiment of the invention, the causal relationships between various dimensions in the spatiotemporal tensor are determined based on the constructed relevant causal chains. Such causal relationships include the influence of temperature and the heat dissipation capacity of the device on the device temperature, the influence of device model on the heat dissipation capacity of the device, and the influence of user type on the frequency of device use.

[0035] S2005: Based on the causal relationship, determine the corresponding causal characteristics, result characteristics, and confounding characteristics; In this embodiment of the invention, the causal features and result features are elements of different dimensions in a spatiotemporal tensor. After allocating causal features and result features to all elements in the spatiotemporal tensor, a selection is made based on the causal relationship between the causal features and result features to obtain hybrid features. The hybrid features are elements that simultaneously affect both causal features and result features. For example, the location of the charging device determines the region where the device is located. The region where the device is located affects the distribution of device models, such as due to procurement reasons. Furthermore, the location of the charging device also affects the failure rate of the device due to environmental factors. Since there is a causal relationship between the device model and the failure rate of the device, the region where the charging device is located is a hybrid factor. That is, the region where the charging device is located affects both the causal features and result features in the causal relationship.

[0036] S2006: Based on historical equipment failure data, the number of samples of the result features under the current cause features and confounding features is taken as the first sample number, and the total number of samples of the confounding features and cause features is taken as the second sample number. The conditional probability is obtained based on the first sample number and the second sample number. In this embodiment of the invention, the number of samples of outcome features occurring under the influence of causal features and confounding features in historical equipment fault data is counted to obtain a first sample number. The total number of samples of confounding features and causal features is counted as a second sample number. A conditional probability is obtained based on the first sample number and the second sample number. The conditional probability is... ,in For conditional probability, The first sample size, This represents the second sample size.

[0037] S2007: Based on the current number of samples with confounding features and the total number of samples with all confounding features, obtain the marginal probability; In this embodiment of the invention, the number of samples for a certain confounding feature and the total number of samples for all confounding features in a historical equipment fault dataset are statistically analyzed, such as the number of samples in region A and the number of samples in all regions, to obtain the marginal probability. The marginal probability is... ,in Represents marginal probability. Indicates the first The number of samples with confounding features. This represents the total number of samples with all confounding features.

[0038] S2008: Obtain the causal probability between the current cause feature and the result feature through the conditional probability and the marginal probability; In this embodiment of the invention, the conditional probability and marginal probability of different confounding features are calculated, and the causal probability is obtained by summing them:

[0039] in, Represents causal probability. Represents conditional probability. Represents marginal probability. Indicates causal characteristics. Indicates the characteristics of the result. Indicates mixed characteristics, Indicates the first A hybrid feature, Indicates the number of hybrid features.

[0040] S300. The spatiotemporal tensor is hierarchically allocated according to the causal relationship, and a fault causal graph is constructed by combining the causal relationship and causal probability. In this embodiment of the invention, the position of each element in the fault causality graph is determined by hierarchically allocating the spatiotemporal tensor, which guides the order and constraint direction of tensor decomposition, including the following steps: S3001: Construct constraints based on the causal relationship and perform hierarchical allocation on the spatiotemporal tensor; Specifically, in this embodiment of the invention, the spatiotemporal tensor is hierarchically allocated according to causal relationships and feature types. The resulting hierarchically allocated spatiotemporal tensor contains spatial dimension features and environmental parameters that are unaffected by variables, thus belonging to the root cause layer (L1). Charging user information and device attributes are inherent attributes of the device and do not change with the device's operating state, thus belonging to the attribute layer (L2). Time dimension data represents the operating state parameters of the charging device and is affected by the environment and device attributes, thus belonging to the state layer (L3). Fault description data represents the final fault result of the charging device, belonging to the fault layer (L4). This hierarchical allocation stipulates that causality can only point from a lower level to a higher level (L1). L2 L3 L4, in the tensor decomposition process, can only proceed from lower levels to higher levels (L1). L2 L3 The L4 process proceeds sequentially, ultimately tracing the fault back to its root cause.

[0041] S3002: Based on the spatiotemporal tensor after hierarchical allocation, a causal hierarchical path is formed, and a fault causal graph is formed through the causal hierarchical path and the corresponding causal probability. In this embodiment of the invention, based on the spatiotemporal tensor after hierarchical allocation, the hierarchy to which all elements belong in the causal relationship are determined. Through the hierarchy to which the elements belong and the causal relationship between the elements, a causal hierarchical path is formed. Combined with the causal probability between the cause features and the result features in the causal relationship, a fault causal graph is formed. The fault causal graph includes elements, edges, causal probabilities, and the hierarchy to which they belong.

[0042] S400. Calculate the path integral in the fault causality graph, obtain a spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and perform hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain the factor matrix and factor weights; including the following steps: In this embodiment of the invention, the path integral is used to assess the impact of the root cause on the final charging equipment failure. Specifically, calculating the path integral in the fault causal graph includes: S4001. Obtain the causal and outcome characteristics of the causal hierarchical path, and calculate the effect coefficient between the causal and outcome characteristics; In this embodiment of the invention, for each causal hierarchical path, the first step is to obtain its causal characteristics and result characteristics, analyze the change in the result characteristics when the causal characteristics change, for example, a quantitative indicator of how much the result characteristics change when the causal characteristics change by one unit, and obtain the effect coefficient between the causal characteristics and result characteristics of this causal hierarchical path by quantifying the change.

[0043] In this embodiment, the covariance and variance of the causal and outcome features of different causal hierarchical paths are obtained to quantify the changes between the causal and outcome features, i.e., the corresponding effect coefficients are: ,in, Represents the effect coefficient. This represents the covariance between causal and outcome characteristics. Variance representing causal characteristics, Indicates causal characteristics. Indicates the characteristics of the result; S4002: Obtain the occurrence time of cause features and result features from the historical equipment failure dataset, construct a response curve, and obtain the time verification score based on the response curve; In this embodiment of the invention, the occurrence times of causal features and result features are obtained from a historical equipment failure dataset. A response curve is constructed with the occurrence time as the x-axis and the occurrence probability of the result feature as the y-axis. The probability of the causal feature occurring before the result feature is compared based on the response curve, and a first score is obtained by quantification. If the causal feature occurs before the result feature, the first score is 0.2; otherwise, it is 0. The morphological characteristics of the response curve are analyzed. Since there is a true causal relationship between the causal feature and the result feature, the probability of the result feature should increase after the occurrence of the causal feature. peak The descending single peak, in this embodiment, has a morphological feature including a single peak, which is quantified to obtain a second score. If the response curve conforms to this morphological feature, the second score is 0.2; otherwise, it is 0. The time verification score is obtained by weighted fusion based on the first score and the second score. S4003: The spatial verification score is obtained based on the equipment location data in the cause characteristics and the occurrence probability in the result characteristics; In this embodiment of the invention, different distance thresholds are preset to obtain the device location data of the corresponding device for the cause feature. The probability of occurrence of the result feature when the distance between the device location data and the distance threshold is within the distance threshold is obtained. By fitting different distance thresholds and their corresponding occurrence probabilities, a fitting result is obtained, and it is determined whether the probability of occurrence of the result feature decreases as the distance threshold increases, thus obtaining the decay trend of the fitting result. If a decay trend exists, the quantization score is 0.2. Furthermore, based on the fitting result, the slope of the fitting result is obtained through linear regression. The decay distance is calculated based on the slope, and corresponding quantization scores are set for different decay thresholds. If the decay distance meets a certain decay distance, a corresponding quantization score is obtained. For example, if the decay distance is less than 200 meters, the quantization score is 0.2. The goodness of fit of the fitting result is obtained through the fitting result and real data. The goodness of fit is... ,in, Indicates the goodness of fit. This indicates the probability of the occurrence of the resulting feature. This indicates the fitting result. The mean of the probability of the fitted result is represented. If the goodness of fit is greater than the preset goodness of fit threshold, the quantization score is 0.2. The decay trend, decay distance and goodness of fit of the fitted result are quantified and fused to obtain the spatial verification score. S4004: Based on historical maintenance data, verify the cause characteristics and result characteristics to obtain the intervention verification score; In this embodiment of the invention, based on the equipment's historical maintenance data, causal characteristics and result characteristics are verified. If there is a genuine causal relationship between the causal characteristics and result characteristics, then after maintenance based on the causal characteristics, the probability of the result characteristics occurring should significantly decrease. An intervention verification score is obtained after verification based on this characteristic. The occurrence probability of the result characteristics before and after correcting the causal characteristics is obtained, and the failure rate change data is obtained based on the occurrence probabilities before and after correction, thereby obtaining the intervention verification score. ,in, Indicates the intervention validation score. This indicates the probability of the occurrence of the characteristic in the result before correction. This indicates the probability of the occurrence of the corrected result feature.

[0044] S4005. Based on the time verification score, spatial verification score, intervention verification score, and effect coefficient, the path integral is obtained.

[0045] The verification score is obtained by fusing the time verification score, the space verification score, and the intervention verification score. ,in, Indicates the verification score. Indicates the time verification score. Indicates the spatial verification score. Indicates the intervention validation score. , , The corresponding parameters are used to obtain the effect coefficients between different causal features and outcome features in the entire causal hierarchical path, and then weighted and fused with the validation score to obtain the path integral. In this embodiment of the invention, path integrals are mapped to a spacetime tensor to obtain a spacetime causal tensor. The specific mapping method is as follows: calculate the path integrals of different causal levels in the spacetime tensor, and multiply the path integrals by the corresponding failure rates of the causal level paths in the spacetime tensor. By traversing all causal level paths, the spacetime causal tensor is obtained. This method amplifies the failure rates of strong causal paths, increasing their importance in the tensor decomposition process.

[0046] In this embodiment of the invention, the hierarchical tensor decomposition of the spatiotemporal causal tensor based on the hierarchical allocation result to obtain the factor matrix and factor weights includes: S4006. Based on the result of the hierarchical allocation, constrain the direction of the spatiotemporal causal tensor decomposition. Through the direction of tensor decomposition, first perform tensor decomposition on environmental parameters and spatial dimension features to obtain environmental factor matrix, spatial factor matrix and corresponding factor weights. Based on the hierarchical allocation results of the spatiotemporal tensor, since environmental parameters and spatial dimensional features belong to the root cause layer, the spatiotemporal causal tensor is first decomposed hierarchically. Then, with the other dimensions of the spatiotemporal causal tensor fixed, singular value decomposition is used to decompose the environmental parameters and spatial dimensional features of the spatiotemporal tensor, yielding an environmental factor matrix, a spatial factor matrix, and corresponding factor weights. S4007: Based on the environmental factor matrix, spatial factor matrix and corresponding factor weights, decompose the charging user information and charging equipment attributes in the spatiotemporal causal tensor to obtain the customer factor matrix, attribute factor matrix and corresponding factor weights. The attribute layer initialization conditions are constructed based on the environmental factor matrix, spatial factor matrix, and corresponding factor weights to form the tensor decomposition attribute layer initialization conditions. , ,in, , Indicates the initialization condition. Represents the spatial factor matrix, Represents the environmental factor matrix. , , , The attribute layer initialization matrix is ​​represented by the alternating least squares method based on the initialization conditions. In the attribute layer of the spatiotemporal causal tensor, namely the dimensions of charging user information and charging equipment attributes, matrix decomposition is performed to obtain the customer factor matrix, attribute factor matrix and corresponding factor weights. S4008: Based on the environmental factor matrix, spatial factor matrix, customer factor matrix, attribute factor matrix and corresponding factor weights, the time dimension features in the spatiotemporal causal tensor are decomposed to obtain the time factor matrix and corresponding factor weights. The state layer initialization conditions for tensor decomposition are constructed based on the environmental factor matrix, spatial factor matrix, customer factor matrix, and attribute factor matrix. The state layer initialization conditions are as follows: ,in, This indicates the initialization conditions of the state layer. Represents the spatial factor matrix, Represents the environmental factor matrix. Represents the customer factor matrix. Represents the attribute factor matrix, , , , The state layer initialization matrix is ​​represented by the alternating least squares method, which decomposes the time dimension features of the spatiotemporal causal tensor based on the state layer initialization conditions to obtain the time factor matrix and the corresponding factor weights. S4009: Based on the environmental factor matrix, time factor matrix, spatial factor matrix, customer factor matrix, attribute factor matrix and corresponding factor weights, the fault description data in the spatiotemporal causal tensor is decomposed to obtain the fault factor matrix and corresponding factor weights. The fault layer initialization conditions are constructed based on the environmental factor matrix, time factor matrix, spatial factor matrix, customer factor matrix, attribute factor matrix, and corresponding factor weights, using tensor decomposition. The fault layer initialization conditions are as follows: The Indicates the fault layer initialization conditions. Represents the spatial factor matrix, Represents the environmental factor matrix. Represents the customer factor matrix. Represents the attribute factor matrix, Represents the time factor matrix, , , , , The fault layer initialization matrix is ​​decomposed using alternating least squares in the dimension of the fault description data of the spatiotemporal causal tensor to obtain the fault factor matrix and the corresponding factor weights.

[0047] S500: Based on the factor matrix, factor weights, and spatiotemporal causal tensor, anomaly scores are obtained. Based on the anomaly scores, fault description data, and fault cause-effect graph, the fault is located, and the location path is obtained. The cause of the equipment fault is determined by combining the factor matrix and factor weights, including the following steps: S5001: Based on the factor matrix and factor weights, tensor reconstruction is performed to obtain the reconstructed causal tensor; In this embodiment of the invention, the environmental factor matrix, time factor matrix, customer factor matrix, attribute factor matrix, fault factor matrix and corresponding factor weights obtained by decomposition are multiplied together to obtain the reconstructed causal tensor.

[0048] S5002: Compare the reconstructed causal tensor and the spatiotemporal causal tensor to obtain statistical residual results, and calculate the outlier score based on the statistical residual results; The reconstructed causal tensor obtained in this embodiment of the invention can reflect the general pattern of device operating data under normal operating conditions. Therefore, by obtaining the difference between the reconstructed causal tensor and the spatiotemporal causal tensor, abnormal data can be located. To obtain the difference between the reconstructed causal tensor and the spatiotemporal causal tensor, the statistical residual results between the reconstructed causal tensor and the spatiotemporal causal tensor are obtained. The statistical residual results include residuals, residual mean, and residual standard deviation. The residual is... The mean of the residuals is The standard deviation of the residual is ,in Represents the residual. Represents the spacetime causality tensor. This represents the reconstruction of the causal tensor. This represents the mean of the residuals. This indicates the number of values ​​in the tensor. Indicates the standard deviation of the residuals; This invention provides an outlier score based on the residuals, the residual mean, and the residual standard deviation. The outlier score is... ,in, Indicates abnormal scores. This represents the mean of the residuals. Represents the residual. This represents the standard deviation of the residuals.

[0049] S5003: Based on the anomaly score and fault description data, perform fault location in the fault cause-effect graph and determine the location path; In this embodiment, a preset anomaly threshold is established. If the anomaly score is greater than the threshold, the fault is located in the fault causal graph based on the anomaly score and the corresponding fault description data. During the fault location process, this embodiment first obtains the starting point of the fault location in the fault causal graph based on the fault description data, and then traces back along the fault causal graph to obtain the location path. For example, several causal hierarchical paths with anomaly scores greater than the preset anomaly threshold are first selected in the fault causal graph. Based on the causal hierarchical paths, the corresponding fault description data in the spatiotemporal tensor is obtained, such as "charging gun fault," and the location path is obtained by tracing back level by level along the causal hierarchical path, such as "Pudong District, July, ride-hailing, Model A, current fluctuation," which causes the charging gun of the charging equipment to malfunction.

[0050] S5004: Calculate the comprehensive weight of the positioning path, combine it with the anomaly score to obtain a comprehensive score, and determine the cause of the equipment failure based on the comprehensive score ranking result.

[0051] Since multiple location paths may be obtained at this time, for further filtering, this embodiment obtains a comprehensive weight based on the anomaly score, factor matrix, and factor weight of the corresponding location path. The comprehensive weight is... ; Furthermore, in this embodiment of the invention, based on the comprehensive weight, for several causal hierarchical paths obtained from the abnormal scores, the location path with the highest comprehensive weight is selected as the cause of the device failure.

[0052] in, This represents the overall weight of the candidate fault causes. Indicate the cause of the candidate fault The corresponding factor weights, The value at the corresponding position in the factor matrix represents the candidate fault cause. Indicates the total number of factors. Indicates the factor number, Dimensions representing candidate fault causes.

[0053] Example 2: A fault location system for charging equipment, such as Figure 2 As shown, it includes: The feature extraction module 100 acquires fault data of the charging equipment, extracts time dimension data and spatial dimension data based on the fault data of the charging equipment, and performs preprocessing and feature extraction to obtain time dimension features and spatial dimension features. The causal verification module 200 acquires basic information related to the charging equipment and fault description data in the fault data, constructs a spatiotemporal tensor by combining the time dimension features and spatial dimension features, and determines the causal relationship and corresponding causal probability by combining the historical equipment fault dataset. The causal graph construction module 300 allocates the spatiotemporal tensor hierarchically through the causal relationship and constructs a fault causal graph by combining the causal relationship and causal probability. Tensor decomposition module 400 calculates the path integral in the fault causal graph, obtains a spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and performs hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain the factor matrix and factor weights. The fault location module 500 obtains anomaly scores based on the factor matrix, factor weights, and spatiotemporal causal tensor. It locates the fault based on the anomaly scores, fault description data, and fault causal graph, obtains the location path, and determines the cause of the equipment fault by combining the factor matrix and factor weights.

[0054] Various changes and modifications made without departing from the spirit and scope of this invention, and all equivalent technical solutions, also fall within the scope of this invention.

[0055] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0056] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention 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.

[0057] This invention is described with reference to flowchart illustrations and / or block diagrams of the method, terminal device (system), and computer program product according to 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 terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, 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.

[0058] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate 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.

[0059] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal 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.

[0060] It should be noted that: The phrase "an embodiment" or "an embodiment" used in this specification means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. Therefore, the phrase "an embodiment" or "an embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment.

[0061] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0062] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of each module is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0063] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The modules in the system of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional modules in the various embodiments of this invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0064] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0065] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for locating faults in charging equipment, characterized in that, Includes the following steps: Obtain fault data of charging equipment, extract time dimension data and spatial dimension data from the fault data of charging equipment, and perform preprocessing and feature extraction to obtain time dimension features and spatial dimension features. Acquire basic information related to the charging equipment and fault description data from the fault data, construct a spatiotemporal tensor by combining the time dimension features and spatial dimension features, and determine the causal relationship and corresponding causal probability by combining the historical equipment fault dataset. The spatiotemporal tensor is hierarchically allocated using the causal relationships, and a fault causal graph is constructed by combining the causal relationships and causal probabilities. Calculate the path integral in the fault causal graph, obtain the spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and perform hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain the factor matrix and factor weights. Anomaly scores are obtained based on the factor matrix, factor weights, and spatiotemporal causal tensor. Faults are located based on the anomaly scores, fault description data, and fault causal graph, and the location path is obtained. The cause of equipment failure is determined by combining the factor matrix and factor weights.

2. The charging equipment fault location method according to claim 1, characterized in that, The step of extracting time-dimensional and spatial-dimensional data from the fault data of the charging device, and performing preprocessing and feature extraction to obtain time-dimensional and spatial-dimensional features includes the following steps: Based on the fault data of the charging equipment, the fault occurrence time is obtained, a preset time window is established, and the equipment operation parameters during the operation process of the equipment within the event window are extracted according to the fault occurrence time. Combined with the time series of equipment operation, time dimension data is formed, which includes electrical parameters and equipment temperature parameters. Acquire spatial dimension data of the charging device, the spatial dimension data including device location data and time series; Preprocessing is performed on time-dimensional data and spatial-dimensional data to obtain preprocessed time-dimensional data and spatial-dimensional data. The preprocessing includes one or more of the following: deduplication, outlier removal, and missing value processing. By using the time series of device operation, the preprocessed time dimension data and spatial dimension data are aligned and correlated to obtain aligned time dimension data and spatial dimension data. Feature extraction is performed on the aligned time dimension data and spatial dimension data respectively to obtain time dimension features and spatial dimension data. The time dimension features include device statistical features, change trend features and data form features, and the spatial dimension features include device coding features and device density features.

3. The charging equipment fault location method according to claim 1, characterized in that, The process of acquiring basic information related to the charging equipment and fault description data from the fault data, constructing a spatiotemporal tensor by combining the time and spatial dimension features, and determining causal relationships and corresponding causal probabilities by combining historical equipment fault datasets includes the following steps: Acquire basic information about the charging device, including charging user information, charging device attributes, and environmental parameters, and construct a spatiotemporal tensor by combining the time dimension features and spatial dimension features. Based on prior knowledge of the charging device, causal relationships between elements of each dimension of the spatiotemporal tensor are extracted. The prior knowledge includes thermal-related knowledge, electrical-related knowledge, usage-related knowledge, and device attribute-related knowledge. By obtaining the corresponding causal and outcome features through causal relationships, and then filtering based on the causal relationships between the causal and outcome features, mixed features are obtained. The sample number of result features under confounding features and causal features is obtained from historical equipment failure data to obtain the first sample number. The total sample number of confounding features and causal features is obtained to obtain the second sample number. The conditional probability is obtained based on the first sample number and the second sample number. The marginal probability is obtained by using the current number of promiscuous feature samples and the total number of promiscuous feature samples. The causal probability between the current causal feature and the result feature is obtained based on conditional probability and marginal probability. The causal probability is: in, Represents causal probability. Represents conditional probability. Represents marginal probability. Indicates causal characteristics. Indicates the characteristics of the result. Indicates mixed characteristics, Indicates the first A hybrid feature, Indicates the number of hybrid features.

4. The charging equipment fault location method according to claim 1, characterized in that, The step of hierarchically allocating the spatiotemporal tensor through the causal relationship and constructing a fault causal graph by combining the causal relationship and causal probability includes the following steps: The causal relationships are used to assign each dimension of the spatiotemporal tensor to a causal hierarchy, resulting in a hierarchy assignment result that includes a root cause layer, an attribute layer, a state layer, and a fault layer. Based on the hierarchical allocation results and causal relationships, a causal hierarchical path is formed, and a fault causal graph is formed based on the causal hierarchical path and the corresponding causal probability.

5. The charging equipment fault location method according to claim 1, characterized in that, The path integral in the fault cause-effect graph is obtained through the following steps: Obtain the covariance and variance between causal features and outcome features of different causal paths in the fault cause-effect graph, and then obtain the effect coefficients between causal features and outcome features based on the covariance and variance; Based on the historical equipment failure dataset, the occurrence time of cause features and result features is obtained, response curves are constructed, and the response curves are evaluated to obtain time verification scores. Obtain the device location data corresponding to the cause characteristics, obtain the occurrence probability of the result characteristics within different preset distance threshold ranges based on the device location data, perform fitting based on the distance thresholds and occurrence probabilities, and obtain the spatial verification score through the fitting results; Intervention validation scores are obtained by analyzing historical maintenance data from historical equipment failure sets, identifying serious cause characteristics and outcome characteristics, and then validation scores are obtained by combining time-based and space-based validation scores. Finally, path integrals are derived based on effect coefficients and validation scores.

6. The charging equipment fault location method according to claim 1, characterized in that, The process of obtaining a spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and then performing hierarchical tensor decomposition on the spatiotemporal causal tensor based on the hierarchical allocation result to obtain a factor matrix and factor weights includes the following steps: The path integral is mapped to the spacetime tensor to obtain the spacetime causality tensor; Based on the result of the hierarchical allocation, the decomposition direction of the spatiotemporal causal tensor is constrained, and the environmental parameters and spatial dimension features of the spatiotemporal causal tensor are decomposed to obtain the environmental factor matrix, the spatial factor matrix and the corresponding factor weights. Based on the environmental factor matrix, spatial factor matrix and corresponding factor weights, the dimensions of charging user information and charging equipment attributes are decomposed to obtain the customer factor matrix, attribute factor matrix and corresponding factor weights. The dimensions of the time dimension features are decomposed based on the environmental factor matrix, spatial factor matrix, customer factor matrix, attribute factor matrix and corresponding factor weights to obtain the time factor matrix and corresponding factor weights. The dimensions of the fault description data are decomposed based on the environmental factor matrix, spatial factor matrix, customer factor matrix, attribute factor matrix, and time factor matrix to obtain the fault factor matrix and the corresponding factor weights.

7. The charging equipment fault location method according to claim 1, characterized in that, The process of obtaining anomaly scores based on the factor matrix, factor weights, and spatiotemporal causal tensor, locating faults based on anomaly scores, fault description data, and fault cause-effect graphs, obtaining location paths, and determining the cause of equipment faults by combining the factor matrix and factor weights includes the following steps: Based on the factor matrix and the corresponding factor weights, tensor reconstruction is performed to obtain the reconstructed causal tensor. Statistical analysis is performed by reconstructing the causal tensor and the spatiotemporal causal tensor to obtain statistical residual results. Anomaly scores are obtained based on the statistical residual results. The statistical residual results include residuals, residual mean, and residual standard deviation. Based on the anomaly score, the fault location path is obtained through fault location and path backtracking in the fault cause-effect graph; Calculate the comprehensive weight of the positioning path, combine it with the anomaly score to obtain a comprehensive score, and sort the data based on the comprehensive score to determine the cause of the equipment failure.

8. A fault location system for charging equipment, characterized in that, include: The feature extraction module acquires fault data of the charging equipment, extracts time dimension data and spatial dimension data based on the fault data of the charging equipment, and performs preprocessing and feature extraction to obtain time dimension features and spatial dimension features. The causal verification module acquires basic information related to the charging equipment and fault description data from the fault data, constructs a spatiotemporal tensor by combining the time dimension features and spatial dimension features, and determines the causal relationship and corresponding causal probability by combining the historical equipment fault dataset. The causal graph construction module allocates the spatiotemporal tensor hierarchically through the causal relationships and constructs a fault causal graph by combining the causal relationships and causal probabilities. The tensor decomposition module calculates the path integral in the fault causal graph, obtains the spatiotemporal causal tensor based on the path integral and the spatiotemporal tensor, and performs hierarchical tensor decomposition on the spatiotemporal causal tensor based on the result of the hierarchical allocation to obtain the factor matrix and factor weights. The fault location module obtains anomaly scores based on the factor matrix, factor weights, and spatiotemporal causal tensor. It then locates the fault based on the anomaly scores, fault description data, and fault causal graph, obtains the location path, and determines the cause of the equipment fault by combining the factor matrix and factor weights.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

10. A fault location device for charging equipment, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 7.

Images