Charging pile fault early warning management method and system

Abstract

The invention relates to the technical field of charging pile fault early warning, and discloses a charging pile fault early warning management method and system, which realize the real-time performance and continuity of data acquisition through robot process automation, thoroughly get rid of the limitation of traditional fixed-period inspection, and can capture the sudden abnormal data of a charging pile in real time. Meanwhile, a large amount of manual inspection work is replaced by automatic data acquisition and intelligent model prediction, the human input cost is greatly reduced, the problems of low efficiency, judgment deviation and the like possibly occurring in manual operation are avoided, accurate quantification of the fault probability is achieved through feature extraction and model operation, and the fault detection accuracy is improved. The operation and maintenance personnel can quickly position the high-risk equipment based on the clear probability threshold value and intervene in treatment in advance, so that the timeliness and accuracy of fault early warning are improved, optimal configuration of operation and maintenance resources is realized, and the problems of equipment damage, potential safety hazards, service interruption and the like caused by fault omission are effectively avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of charging pile fault early warning, in particular to a charging pile fault early warning management method and system. BACKGROUND

[0002] As the core infrastructure of electric vehicle energy supply, the operation stability of charging piles not only relates to the charging experience of users, but also directly affects the charging safety and regional traffic efficiency. On this basis, the traditional device management mode relying on passive response has been difficult to meet the operation and maintenance needs in high-frequency use scenarios. Building an efficient and intelligent charging station fault early warning function that can identify potential risks in advance and actively avoid faults has become a key requirement to ensure the continuous and stable operation of charging piles and improve the service capability of infrastructure, which is of great significance to support cross-regional traffic coordination and new energy industry development.

[0003] However, the existing charging pile fault repair mainly relies on manual inspection. The operation and maintenance personnel need to go to the site according to the fixed cycle and standard, check the charging gun, cable, pile body appearance, display lamp, internal components and other hardware of the charging pile one by one, and test the charging function, emergency stop function, card swiping function and background data link state to check for safety hazards and ensure normal operation of the equipment. However, the above method is limited by the fixed cycle and is difficult to capture sudden abnormalities of the equipment in real time. Moreover, it has high labor cost and low efficiency, and is easy to miss fault risks due to inspection gaps. SUMMARY

[0004] The present application provides a charging pile fault early warning management method and system, which solves the technical problem that the existing technology is limited by the fixed cycle, is difficult to capture sudden abnormalities of the equipment in real time, has high labor cost and low efficiency, and is easy to miss fault risks due to inspection gaps.

[0005] The first aspect of the present application provides a charging pile fault early warning management method, comprising:

[0006] Obtaining real-time data of a to-be-tested charging pile through robotic process automation, and preprocessing the real-time data;

[0007] Extracting a feature vector of the preprocessed real-time data and inputting a preset target fault prediction model;

[0008] Performing fault prediction on the feature vector through the target fault prediction model to generate a fault probability score;

[0009] Comparing the fault probability score with a preset probability threshold, determining the fault information of the to-be-tested charging pile according to the comparison result, and sending the fault information and corresponding early warning information to an operation and maintenance personnel.

[0010] Optionally, the real-time data of the to-be-tested charging pile is acquired through the robotic process automation, and the real-time data is preprocessed, including:

[0011] The real-time data of the to-be-tested charging pile collected by the sensor is collected in a timing manner through the robotic process automation;

[0012] The real-time data is sequentially subjected to data cleaning processing and normalization processing.

[0013] Optionally, the feature vector of the preprocessed real-time data is extracted and input into a preset target fault prediction model, including:

[0014] The historical data of the to-be-tested charging pile is acquired through the robotic process automation;

[0015] The historical data is divided into a training set and a test set according to a preset ratio;

[0016] The historical data of the training set is input into a preset initial fault prediction model for training, to generate an updated fault prediction model;

[0017] The historical data of the test set is input into the updated fault prediction model for testing, to generate a target fault prediction model;

[0018] The feature vector of the preprocessed real-time data is extracted;

[0019] The feature vector is input into the target fault prediction model.

[0020] Optionally, the fault probability score is compared with a preset probability threshold value, the fault information of the to-be-tested charging pile is determined according to a comparison result, and the fault information and corresponding early warning information are sent to an operation and maintenance personnel, including:

[0021] It is determined whether the fault probability score is less than a preset first probability threshold value;

[0022] If yes, it is determined that the current running state of the to-be-tested charging pile is a normal state;

[0023] If no, it is determined whether the fault probability score is greater than or equal to the first probability threshold value and less than a preset second probability threshold value;

[0024] If yes, it is determined that the current running state of the to-be-tested charging pile is a state that needs attention;

[0025] If no, it is determined whether the fault probability score is greater than or equal to the second probability threshold value and less than a preset third probability threshold value;

[0026] If so, it is determined that the current operating status of the charging pile under test has a potential fault risk, and the warning information corresponding to the potential fault risk is obtained and sent to the operation and maintenance personnel.

[0027] If not, then determine whether the fault probability score is greater than or equal to the third probability threshold;

[0028] If so, it is determined that the current operating status of the charging pile under test is in the event of an impending fault, and the alarm information corresponding to the impending fault is obtained and sent to the maintenance personnel.

[0029] Optionally, it also includes:

[0030] After the maintenance personnel handle the potential fault risk or the early warning case corresponding to the impending fault, they record the current maintenance data.

[0031] Based on the feature vector of the real-time data, the fault probability score, and the operation and maintenance data, the accuracy of the target fault prediction model is determined.

[0032] The accuracy rate is compared with a preset warning accuracy threshold, and the target fault prediction model is updated based on the comparison result to generate a new target fault prediction model.

[0033] A second aspect of the present invention provides a charging pile fault early warning management system, comprising:

[0034] The preprocessing module is used to automatically acquire real-time data of the charging pile under test through robotic process and to preprocess the real-time data.

[0035] The extraction module is used to extract the feature vectors of the preprocessed real-time data and input them into the preset target fault prediction model;

[0036] The prediction module is used to predict the faults in the feature vector using the target fault prediction model and generate a fault probability score.

[0037] The comparison module is used to compare the fault probability score with a preset probability threshold, determine the fault information of the charging pile under test based on the comparison result, and send the fault information and the corresponding early warning information to the operation and maintenance personnel.

[0038] Optionally, the preprocessing module includes:

[0039] The data acquisition submodule is used to automatically collect real-time data of the charging pile under test from the sensors at regular intervals through a robotic process.

[0040] The processing submodule is used to perform data cleaning and normalization processing on the real-time data in sequence.

[0041] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the charging pile fault early warning management method as described in any of the preceding claims.

[0042] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, it implements the charging pile fault early warning management method as described in any of the preceding claims.

[0043] The fifth aspect of the present invention provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein, when the program instructions are executed by a computer, the computer performs the charging pile fault early warning management method as described in any of the preceding claims.

[0044] As can be seen from the above technical solutions, the present invention has the following advantages:

[0045] This invention employs robotic process automation (RPA) technology to collect real-time operational data from charging piles under test. After preprocessing, feature vectors are extracted and input into a preset target fault prediction model. The model generates a fault probability score, which is then compared with a preset probability threshold to determine fault information. Fault and early warning information are simultaneously pushed to maintenance personnel. This invention achieves real-time and continuous data collection through RPA, completely eliminating the limitations of traditional fixed-cycle inspections. It can instantly capture sudden abnormal data from charging piles, solving the core problem of missed faults due to inspection intervals in existing technologies from the data source. Simultaneously, automated data collection and intelligent model prediction replace a large amount of manual inspection work, significantly reducing labor costs and avoiding inefficiencies and judgment errors that may occur with manual operations. Through feature extraction and model calculation, the fault probability is accurately quantified, allowing maintenance personnel to quickly locate high-risk equipment based on clear probability thresholds and intervene early. This improves the timeliness and accuracy of fault warnings, optimizes the allocation of maintenance resources, and effectively avoids equipment damage, safety hazards, and service interruptions caused by missed faults. Attached Figure Description

[0046] 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.

[0047] Figure 1 This is a flowchart of the steps of a charging pile fault early warning management method provided in Embodiment 1 of the present invention;

[0048] Figure 2 This is a structural block diagram of a charging pile fault early warning management system provided in Embodiment 2 of the present invention;

[0049] Figure 3 This is a structural block diagram of a computer device provided in Embodiment 3 of the present invention. Detailed Implementation

[0050] This invention provides a charging pile fault early warning management method and system to solve the technical problems of existing technologies that are limited by fixed cycles, making it difficult to capture sudden equipment anomalies in real time, and are characterized by high labor costs, low efficiency, and easy to miss fault risks due to gaps in inspection.

[0051] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0052] Please see Figure 1 , Figure 1 The flowchart illustrates the steps of a charging pile fault early warning management method provided in Embodiment 1 of the present invention.

[0053] The present invention provides a charging pile fault early warning management method, comprising the following steps:

[0054] Step 101: Obtain real-time data of the charging pile under test through robotic process automation, and preprocess the real-time data.

[0055] In this embodiment of the invention, Robotic Process Automation (RPA) refers to a technology that automatically executes repetitive and rule-based business processes by simulating human operational behavior, enabling data capture, processing, and transmission without modifying existing systems.

[0056] The charging pile under test refers to the charging equipment that is in operation and included in the monitoring scope of this fault early warning management system.

[0057] Real-time data refers to the immediate data that reflects the current operating status of the charging pile under test and its associated equipment, which is dynamically acquired through data acquisition methods. This includes multi-source heterogeneous data such as operation logs, CAN messages, BMS information, temperature and humidity, and current and voltage curves.

[0058] Preprocessing refers to a series of standardized processes performed on the collected real-time data.

[0059] Using RPA (Robotic Process Automation) as the edge data entry point, the system continuously and non-intrusively captures multi-source heterogeneous data, such as operation logs, CAN (Controller Area Network) messages, BMS (Battery Management System) information, temperature and humidity, and current and voltage curves, on the local charging pile or in the cloud via robotic process automation scripts. Subsequently, tools in the Python ecosystem, such as Pandas, NumPy, and PySpark, are used for data cleaning, noise reduction, missing value imputation, outlier removal, normalization / standardization, feature encoding, and feature derivation to ensure that the data entering the model has high consistency, high signal-to-noise ratio, and high interpretability. The cleaned data is then transmitted to the model computing service in real time via message queues (Kafka / RabbitMQ) or RESTful APIs.

[0060] Further, step 101 includes the following sub-steps:

[0061] S11. Real-time data of the charging pile under test is collected by sensors at regular intervals through robotic processes.

[0062] In this embodiment of the invention, a sensor refers to a detection device installed on the charging pile body, core components and surrounding environment under test, including current sensors, voltage sensors, temperature sensors, vibration sensors, etc.

[0063] Deploy RPA robots on edge gateways or lightweight servers and configure their execution cycles or event triggering conditions. RPA scripts simulate human operations, accessing the charging pile controller to read real-time parameters via standard protocols (such as MODBUS-TCP, OPC-UA); capturing CAN messages of vehicle-charging pile interaction via a CAN bus analyzer; and pulling historical operation logs from the cloud platform via HTTP / HTTPS requests.

[0064] The RPA script parses key fields (such as voltage, current, temperature, SOC (State of Charge), and error codes) from the acquired data in raw binary or JSON format. These structured data records are then temporarily written to a local solid-state drive's circular buffer or a Redis cache to prevent data loss during network interruptions.

[0065] The RPA robot packages the cached data into a single data file by time window, and attaches a data source identifier and timestamp.

[0066] Data files can be transferred to a central cloud storage (such as AWS S3) or delivered directly to a message queue access point via a secure TLS (Transport Layer Security) connection.

[0067] S12. Perform data cleaning and normalization processing on the real-time data in sequence.

[0068] In this embodiment of the invention, data cleaning refers to the process of preprocessing the collected raw data.

[0069] Normalization refers to a data standardization technique that maps feature data of different dimensions and ranges to a unified numerical range.

[0070] Real-time data undergoes data cleaning processes, including invalid values, missing values, and noise reduction. This involves identifying and removing data that clearly violates physical rules, such as negative current values ​​or temperatures exceeding 200°C. For time-series data with missing values, forward padding, linear interpolation, or padding based on the average of similar devices are used. For high-frequency sampling signals such as current and voltage, low-pass filters are applied to remove high-frequency noise above 50Hz, retaining effective low-frequency signals reflecting device trends. The cleaned real-time data is then mapped to a unified numerical range [0,1].

[0071] Step 102: Extract the feature vectors of the preprocessed real-time data and input them into the preset target fault prediction model.

[0072] In this embodiment of the invention, a feature vector refers to a vector composed of multiple features extracted from the dataset, where each feature represents an attribute or characteristic of the data.

[0073] A target fault prediction model refers to a pre-built and trained model used to predict equipment faults.

[0074] The sliding window technique is used to extract time series features from the preprocessed real-time data, including statistical features such as the mean, standard deviation, and slope of voltage and current, as well as trend features reflecting changes in equipment status. Then, these features are filtered and dimensionality reduced to remove redundant information and retain key features related to faults, forming a fixed-dimensional feature vector. Finally, this feature vector is input into the preset target fault prediction model.

[0075] Further, step 102 includes the following sub-steps:

[0076] S21. Obtain historical data of the charging pile under test through robotic process automation.

[0077] In this embodiment of the invention, historical data refers to various types of data accumulated by the charging pile under test during its past operation, including charging records, fault logs, sensor time-series data, etc.

[0078] When acquiring historical data of a charging pile under test using an RPA robot, the RPA script deployed on the edge gateway first accesses the cloud historical database via HTTPS protocol. Based on the unique device ID of the charging pile under test, it filters and pulls the operating data of the device over the past 12 months, including historical charging records, fault logs, and time-series data of electrical parameters collected by sensors. During the pulling process, the RPA script downloads the data in blocks on a daily basis, and each block is appended with an MD5 (Message-DigestAlgorithm 5) checksum to ensure integrity. After the download is completed, it automatically compares and deduplicates the historical data with the locally cached data, and finally stores the integrated historical data in the local time-series database.

[0079] S22. Divide the historical data into training set and test set according to the preset ratio.

[0080] In this embodiment of the invention, the preset ratio refers to the pre-set ratio of the number of samples used to divide the training set and the test set, such as 7:3 or 8:2.

[0081] The training set refers to the set of samples that are partitioned from historical data and used to train the fault prediction model.

[0082] The test set refers to a sample set extracted from historical data for testing and evaluating the performance of fault prediction models.

[0083] Samples were drawn from the historical data corresponding to each fault type according to a preset ratio of 7:3. 70% of the samples were used to form the training set for model parameter learning and training, and 30% of the samples were used to form the test set.

[0084] S23. Input the historical data of the training set into the preset initial fault prediction model for training, and generate an updated fault prediction model.

[0085] In this embodiment of the invention, the initial fault prediction model refers to a pre-built, untrained model framework.

[0086] Updating the fault prediction model refers to the model optimized using training data.

[0087] The feature vectors in the training set data are associated with the corresponding fault labels, which are determined based on historical fault logs and include types such as normal operation, abnormal voltage, excessive temperature, and communication failure. Then, the associated training data is input into the preset initial fault prediction model for training.

[0088] It is worth mentioning that: 1) Model selection and initialization:

[0089] LightGBM gradient boosting decision tree was chosen as the base model because it performs well on tabular data and has a fast training speed.

[0090] The model is initialized as a binary classifier, and its core is to build a series of decision trees, each of which learns to correct the prediction error of the preceding tree.

[0091] 2) Iterative training and parameter search:

[0092] A Bayesian optimization strategy is employed to perform intelligent search in the hyperparameter space (including num_leaves, learning_rate, feature_fraction, etc.).

[0093] In each round of parameter testing, temporal cross-validation is used: the training set and validation set are divided in chronological order to prevent future information leakage and to evaluate the generalization ability of the parameters.

[0094] The model learns by minimizing the binary cross-entropy loss function:

[0095]

[0096] In the formula, For true labels (1 or 0). This represents the failure probability predicted by the model.

[0097] 3) Model solidification:

[0098] The model corresponding to the parameter set with the highest AUC (Area Under the ROC Curve) and the smallest log loss on the test set is selected as the updated fault prediction model, and it is saved as a binary file.

[0099] S24. Input the historical data of the test set into the updated fault prediction model for testing, and generate the target fault prediction model.

[0100] In this embodiment of the invention, the target fault prediction model refers to the final model whose performance indicators have met the preset requirements after being verified by the test set.

[0101] The feature vectors from the test set data are input into the updated fault prediction model to obtain the fault prediction results output by the model, including the probability of fault occurrence and the predicted fault type. Then, the prediction results are compared with the actual fault labels in the test set, and the model's prediction accuracy, recall, F1 score and other performance indicators are calculated. If all indicators reach the preset threshold (e.g., accuracy ≥ 85%, recall ≥ 80%), the updated fault prediction model is determined as the target fault prediction model.

[0102] Specifically, 1) the solidified optimal model (updated fault prediction model) is used to make a one-time prediction on the test set.

[0103] Calculate AUC: Plot the ROC (Receiver Operating Characteristic Curve), calculate the area under the curve, and evaluate the overall ranking ability of the model.

[0104] Calculate the F1-Score: This evaluates the classification accuracy at a set threshold by comprehensively considering precision and recall.

[0105] 2) Calculate PSI (Population Stability Index): Compare the feature distribution differences between the training set and the test set (or recent online data). If PSI < 0.1, the data distribution is considered stable and the model performance is reliable, and the current updated fault prediction model is identified as the target fault prediction model; if PSI > 0.25, it indicates that the data has drifted significantly and the model may need to be retrained.

[0106] S25. Extract the feature vector of the preprocessed real-time data.

[0107] In this embodiment of the invention, a sliding time window technique is used to process the preprocessed time-series data such as voltage, current, and temperature in blocks. The window size is set to 5 minutes according to the operating characteristics of the charging pile, and the step size is 1 minute. Then, statistical features are calculated in each window, including indicators reflecting the distribution and fluctuation of the data such as mean, standard deviation, peak value, valley value, and kurtosis. At the same time, the time domain signal is converted into a frequency domain signal through fast Fourier transform, and frequency domain features such as spectral peak and main frequency are extracted. Then, feature engineering is constructed in combination with the physical characteristics of the equipment operation to generate an original feature set containing multiple dimensions such as electrical features, temperature features, and frequency features. Finally, redundant features are removed by screening with Pearson correlation coefficient, and core features strongly correlated with faults are retained to form a fixed-dimensional feature vector.

[0108] S26. Input the feature vector into the target fault prediction model.

[0109] In this embodiment of the invention, the feature vector is dimension-verified to ensure that it is consistent with the input dimension during model training, and then the feature vector is passed into the target fault prediction model through the input layer of the model.

[0110] Step 103: Perform fault prediction on the feature vector using the target fault prediction model to generate a fault probability score.

[0111] In this embodiment of the invention, fault prediction refers to the process of using a target fault prediction model to analyze the input feature vector and predict the types of faults that the device may experience in the future and the probability of their occurrence.

[0112] Fault probability score refers to the probability distribution vector output by the model, which represents the likelihood of various types of faults occurring. Each element corresponds to the probability of a fault type occurring.

[0113] After the feature vector is input into the target fault prediction model, the model performs nonlinear transformations and depth mining on the features through multiple hidden layers. Combining this with fault patterns learned during training, the model calculates the probability of occurrence for each type of fault corresponding to the feature vector. Finally, it outputs a value between 0 and 1, which is the fault probability score. This score represents the degree of similarity between the current state and historical fault states, based on patterns learned from historical data. The higher the score, the greater the risk of a fault occurring.

[0114] It is worth mentioning that the target fault prediction model learns complex patterns from a large amount of historical data. For example, when the core temperature is > 85℃ and the current fluctuation rate suddenly increases, there is an 80% probability that a fault will occur within 1 hour. Based on the above example, it makes predictions on real-time data and outputs a fault probability score between 0 and 1.

[0115] Step 104: Compare the fault probability score with the preset probability threshold, determine the fault information of the charging pile under test based on the comparison result, and send the fault information and corresponding warning information to the operation and maintenance personnel.

[0116] In this embodiment of the invention, the preset probability threshold refers to a pre-set critical probability value for judging the risk of a fault. The threshold is calibrated by the system based on historical fault data and operation and maintenance experience. If the threshold is exceeded, a fault warning is triggered.

[0117] The comparison result refers to the clear judgment conclusion obtained by comparing the probability value of each type of fault in the fault probability classification with the preset probability threshold one by one.

[0118] Fault information refers to a collection of key information including fault type (such as abnormal voltage or excessive temperature), fault probability value, and fault occurrence timestamp.

[0119] Warning information refers to the notification content generated based on fault information, including fault details, warning level (high / medium / low), and handling suggestions.

[0120] Maintenance personnel refer to professionals responsible for the daily maintenance, troubleshooting, and repair of charging piles.

[0121] The probability values ​​of each type of fault in the fault probability classification are compared one by one with the preset probability threshold. If the probability value of a certain type of fault exceeds the threshold, it is determined that the charging pile under test has a fault risk of that type. The fault type, the corresponding probability value, and the fault occurrence timestamp are integrated into fault information. Then, based on the fault type, the preset warning level and handling suggestions are matched to generate warning information containing fault details, warning level, and handling guidance. Finally, the fault information and warning information are pushed to the mobile terminal of the operation and maintenance personnel in real time through the API or SMS gateway by calling the RPA script.

[0122] Furthermore, step 104 includes the following sub-steps:

[0123] S31. Determine whether the fault probability score is less than the preset first probability threshold.

[0124] In this embodiment of the invention, the first probability threshold refers to a pre-set low-risk judgment threshold, which is set to 0.1 in this invention.

[0125] The probability values ​​corresponding to each type of fault in the fault probability classification are compared with 0.1 (i.e., the first probability threshold) to determine whether the probability values ​​corresponding to each type of fault in the fault probability classification are all less than 0.1.

[0126] S32. If yes, then the current operating status of the charging pile under test is determined to be normal.

[0127] In this embodiment of the invention, the normal state refers to the fact that the various operating parameters (such as voltage, current, temperature, etc.) of the charging pile under test are within a preset safe range.

[0128] If the fault probability score is less than 0.1, then the current operating status of the charging pile under test is determined to be normal.

[0129] Specifically, Normal: Fault probability score <0.1, meaning no operation is required, only log recording.

[0130] S33. If not, determine whether the fault probability score is greater than or equal to the first probability threshold and less than the preset second probability threshold.

[0131] In this embodiment of the invention, the second probability threshold refers to a pre-set medium-risk threshold, which is set to 0.5 in this invention.

[0132] If the failure probability score is greater than or equal to 0.1, then determine whether the failure probability score is less than 0.5.

[0133] S34. If yes, then determine that the current operating status of the charging pile under test is a status that needs attention.

[0134] In this embodiment of the invention, the state that needs to be paid attention to refers to the device operating state between the normal state and the fault state.

[0135] If the fault probability score is greater than or equal to 0.1 and less than 0.5, the data collection cycle will be automatically shortened, for example, from 3 minutes / time to 1 minute / time, and the feature monitoring dimensions of the device will be enhanced. At the same time, a notification will be generated that includes the fault type, the current probability value and the recommended monitoring frequency.

[0136] Specifically, Watch: 0.1 ≤ Fault probability score < 0.5, meaning the system marks the device in the background and increases its data collection frequency to closely observe trend changes.

[0137] S35. If not, determine whether the fault probability score is greater than or equal to the second probability threshold and less than the preset third probability threshold.

[0138] In this embodiment of the invention, the third probability threshold refers to a pre-set high-risk threshold, which is set to 0.85 in this invention.

[0139] If the failure probability score is greater than or equal to 0.5, then determine whether the failure probability score is less than 0.85.

[0140] S36. If so, determine that the current operating status of the charging pile under test has potential fault risks, obtain the warning information corresponding to the potential fault risks, and send the potential fault risks and warning information to the operation and maintenance personnel.

[0141] In this embodiment of the invention, potential failure risk refers to a state in which the equipment has not experienced a substantial failure at present, but the failure probability score is in the high-risk range, indicating that a failure may occur in the future, and the maintenance personnel need to be alerted to intervene in advance through early warning information.

[0142] If the fault probability score is greater than or equal to 0.5 and less than 0.85, it is determined that the current operating status of the charging pile under test has potential fault risks. The system will automatically associate the historical handling cases and maintenance plans corresponding to the fault type to generate early warning information. This information includes the fault type, risk level (e.g., high risk), expected fault occurrence time window, and targeted handling suggestions (e.g., a 24-hour shutdown for maintenance). At the same time, the system will push the details of the potential fault risk (including the fault probability score and snapshots of key operating parameters) and the early warning information to the mobile terminal of the maintenance personnel in real time through RPA script to call the API or SMS gateway.

[0143] Specifically, an alert is defined as a fault probability score of 0.5 ≤ Fault Probability Score < 0.85, indicating that a potential fault risk has been identified. The system automatically generates an alert work order on the operations and maintenance platform and notifies the operations and maintenance personnel via SMS or office software that "Charging pile No. XX has a potential fault risk and it is recommended to check it."

[0144] S37. If not, determine whether the fault probability score is greater than or equal to the third probability threshold.

[0145] In this embodiment of the invention, if the fault probability score is not greater than or equal to 0.5 and not less than 0.85, then it is determined whether the fault probability score is greater than or equal to 0.85.

[0146] S38. If so, determine that the current operating status of the charging pile under test is in the event of an impending fault, obtain the alarm information corresponding to the impending fault, and send the impending fault and alarm information to the maintenance personnel.

[0147] In this embodiment of the invention, "imminent failure" refers to a state in which the equipment failure probability score reaches or exceeds 0.85, indicating that a substantial failure is very likely to occur in the short term and requires emergency intervention.

[0148] Alarm information refers to emergency notifications generated in response to impending faults, which include key information such as fault level, handling plan, troubleshooting checklist, and operation instructions.

[0149] If the fault probability score is ≥0.85, it is determined that the current operating status of the charging pile under test is indicative of an impending fault. The system will automatically retrieve the emergency response plan, key component checklist, and emergency shutdown operation guidelines corresponding to the fault type, and integrate them to generate alarm information. This information clearly indicates the fault level as "urgent", the fault occurrence probability score, the core abnormal operating parameters, and the mandatory requirement of "immediately rushing to the site for handling". Subsequently, the details of the impending fault and alarm information will be simultaneously pushed to the maintenance personnel through three channels: WeChat pop-up, SMS, and voice call.

[0150] Specifically, a fault probability score ≥0.85 indicates a highly probable impending fault. The system immediately triggers the highest-level alarm, strongly notifying maintenance personnel through multiple channels (telephone, SMS, app push), and may execute automatic safety policies, such as suggesting the scheduling system stop assigning new orders to the affected pile, or initiating a gradual power reduction process.

[0151] Furthermore, this method also includes:

[0152] S41. After the operation and maintenance personnel handle the early warning cases corresponding to potential fault risks or impending faults, they record the current operation and maintenance data.

[0153] In this embodiment of the invention, operation and maintenance data refers to various types of information generated and recorded by operation and maintenance personnel during the handling of early warning cases or the performance of routine maintenance work, including processing time, measures, replacement of parts, causes of failure, results, etc.

[0154] After maintenance personnel complete the handling of the early warning cases, such as performing equipment inspections, component replacements, or system debugging, the system will automatically prompt the maintenance personnel to record the maintenance data on the designated maintenance management interface or mobile application. This data includes, but is not limited to, the handling time, the personnel ID, the specific maintenance measures taken, the model and serial number of the replaced parts, the fault cause analysis, the handling result (such as fault resolution, temporary relief, or inability to handle), and the initial status parameters of the equipment after it resumes operation. After the maintenance personnel submit the record, the system will associate and store this maintenance data with the corresponding early warning case information and update the equipment's health record.

[0155] S42. Based on the feature vectors, fault probability scores, and operation and maintenance data of real-time data, determine the accuracy of the target fault prediction model.

[0156] In this embodiment of the invention, accuracy refers to the proportion of samples that the model predicts to be faulty, but which actually actually experience a fault.

[0157] The model evaluation process is initiated periodically. First, all alert cases processed and recorded by operations and maintenance personnel within a certain period are used as the evaluation sample set. For each alert case in this sample set, the system extracts the feature vector of the real-time data that triggered the alert, the fault probability score output by the model at that time, and the subsequent operations and maintenance data recorded by the personnel. Next, the system compares the fault probability score output by the model with the actual fault situations recorded in the operations and maintenance data to calculate key performance indicators such as the model's prediction accuracy, recall, and precision.

[0158] S43. Compare the accuracy rate with the preset warning accuracy rate threshold, update the target fault prediction model based on the comparison results, and generate a new target fault prediction model.

[0159] In this embodiment of the invention, the early warning accuracy threshold refers to a numerical standard of accuracy that is pre-set by the system to determine whether the model performance meets the standard.

[0160] The comparison result refers to the conclusion drawn by comparing the current accuracy of the target fault prediction model with the preset early warning accuracy threshold.

[0161] The new target fault prediction model refers to a fault prediction model with better performance and higher prediction accuracy generated after the update process.

[0162] After obtaining the current accuracy of the target fault prediction model, this accuracy is compared with a preset warning accuracy threshold. If the current accuracy is higher than or equal to the threshold, it indicates that the model performance is good and meets the actual application requirements, so there is no need to update the model, and the current target fault prediction model can continue to be used. If the current accuracy is lower than the preset warning accuracy threshold, it indicates that the model's predictive ability may have declined or is insufficient, and optimization is required. At this time, the model update process is automatically triggered, merging the samples accumulated in the recent period, containing new feature vectors, fault probability scores, and corresponding operation and maintenance data, with the original training dataset to form an updated comprehensive training dataset. Subsequently, the original model is retrained using this new dataset, and the model parameters are adjusted to generate a new, higher-performing target fault prediction model. The specific operation is as follows:

[0163] 1) Warning handling and feedback entry: After handling the warning on-site, the operation and maintenance personnel record the final conclusion in the operation and maintenance system (e.g., "Confirmed as a fan failure, which has been replaced" or "No abnormalities were found during inspection, false alarm").

[0164] 2) Data closed loop: The feature data of this warning, the model prediction probability, and the actual results confirmed by the operation and maintenance personnel are combined as a new labeled sample and stored in the "model retraining database".

[0165] 3) Model Monitoring and Iteration: The system continuously monitors the online performance of the model, such as early warning accuracy, false alarm rate, and PSI. When performance degrades or data distribution shifts, the model update process is automatically triggered. The model is then reinforced and evaluated using a dataset containing new feedback data, and then upgraded to a better target fault prediction model.

[0166] Please see Figure 2 , Figure 2 This is a structural block diagram of a charging pile fault early warning management system provided in Embodiment 2 of the present invention.

[0167] The present invention provides a charging pile fault early warning management system, comprising:

[0168] The preprocessing module 201 is used to acquire real-time data of the charging pile under test through robotic process automation and to preprocess the real-time data.

[0169] Extraction module 202 is used to extract the feature vector of the preprocessed real-time data and input it into the preset target fault prediction model;

[0170] Prediction module 203 is used to predict faults in feature vectors using a target fault prediction model and generate fault probability scores.

[0171] The comparison module 204 is used to compare the fault probability score with the preset probability threshold, determine the fault information of the charging pile under test based on the comparison result, and send the fault information and the corresponding warning information to the operation and maintenance personnel.

[0172] Furthermore, the preprocessing module 201 includes:

[0173] The data acquisition submodule is used to automatically collect real-time data of the charging pile under test from the sensors at regular intervals through a robotic process.

[0174] The processing submodule is used to perform data cleaning and normalization processing on real-time data in sequence.

[0175] Furthermore, the extraction module 202 includes:

[0176] The acquisition submodule is used to automatically acquire historical data of the charging pile under test through a robotic process.

[0177] The partitioning submodule is used to divide historical data into training and testing sets according to a preset ratio;

[0178] The training submodule is used to input historical data from the training set into a preset initial fault prediction model for training, and to generate an updated fault prediction model.

[0179] The test submodule is used to input historical data of the test set into the updated fault prediction model for testing, and generate the target fault prediction model.

[0180] The extraction submodule is used to extract the feature vectors of the preprocessed real-time data;

[0181] The input submodule is used to input feature vectors into the target fault prediction model.

[0182] Furthermore, the comparison module 204 includes:

[0183] The first judgment submodule is used to determine whether the fault probability score is less than a preset first probability threshold.

[0184] The normal state submodule is used to determine the current operating state of the charging pile under test as normal if the condition is met.

[0185] The second judgment submodule is used to determine whether the fault probability score is greater than or equal to the first probability threshold and less than the preset second probability threshold if no.

[0186] The status monitoring submodule is used to determine the current operating status of the charging pile under test as a status that needs to be monitored if the status is as described above.

[0187] The third judgment submodule is used to determine whether the fault probability score is greater than or equal to the second probability threshold and less than the preset third probability threshold if no.

[0188] The sending submodule is used to determine if the current operating status of the charging pile under test has potential fault risks, obtain the warning information corresponding to the potential fault risks, and send the potential fault risks and warning information to the operation and maintenance personnel.

[0189] The fourth judgment submodule is used to determine whether the fault probability score is greater than or equal to the third probability threshold if the condition is not met.

[0190] The information sending submodule is used to determine if the current operating status of the charging pile under test is indicative of an impending fault, obtain the alarm information corresponding to the impending fault, and send the impending fault and alarm information to the maintenance personnel.

[0191] Furthermore, this system also includes:

[0192] The recording submodule is used to record the current operation and maintenance data after the operation and maintenance personnel handle the early warning cases corresponding to potential fault risks or impending faults;

[0193] The accuracy submodule is used to determine the accuracy of the target fault prediction model based on feature vectors, fault probability scores, and operation and maintenance data from real-time data.

[0194] The update submodule is used to compare the accuracy with the preset warning accuracy threshold, update the target fault prediction model based on the comparison results, and generate a new target fault prediction model.

[0195] Please see Figure 3 , Figure 3 This is a structural block diagram of a computer device provided in Embodiment 3 of the present invention.

[0196] An electronic device according to an embodiment of the present invention includes: a memory 301 and a processor 302. The memory 301 stores a computer program. When the computer program is executed by the processor 302, the processor 302 executes the charging pile fault early warning management method as described in any of the above embodiments.

[0197] Memory 301 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 301 has storage space 303 for program code 313 for performing any of the method steps described above. For example, storage space 303 for program code may include various program codes 313 for implementing the various steps in the methods described above. These program codes may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When run by a computing processing device, this code causes the computing processing device to perform the various steps in the methods described above. These program codes may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When this code is run by a computing device, it causes the computing device to execute the various steps in the charging pile fault early warning management method described above.

[0198] Embodiment 4 of the present invention provides a computer-readable storage medium storing a computer program thereon, which, when executed, implements the charging pile fault early warning management method as described in any embodiment of the present invention.

[0199] Embodiment 5 of the present invention provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer performs the charging pile fault early warning management method as described in any of the above embodiments.

[0200] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0201] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0202] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0203] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0204] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable 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, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0205] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A charging pile fault early warning management method, characterized in that, include: The robot process automates the acquisition of real-time data of the charging pile under test and preprocesses the real-time data. Extract the feature vectors of the preprocessed real-time data and input them into the preset target fault prediction model; The target fault prediction model is used to predict faults in the feature vector to generate a fault probability score. The fault probability score is compared with a preset probability threshold. Based on the comparison result, the fault information of the charging pile under test is determined, and the fault information and the corresponding warning information are sent to the operation and maintenance personnel.

2. The charging pile fault early warning management method according to claim 1, characterized in that, The process of automatically acquiring real-time data of the charging pile under test through robotic processes and preprocessing the real-time data includes: The robot process automatically collects real-time data from sensors on the charging pile under test at regular intervals. The real-time data is then subjected to data cleaning and normalization processes in sequence.

3. The charging pile fault early warning management method according to claim 1, characterized in that, The step of extracting the feature vectors of the preprocessed real-time data and inputting them into a preset target fault prediction model includes: The historical data of the charging pile under test is automatically acquired through the robotic process. The historical data is divided into a training set and a test set according to a preset ratio; The historical data of the training set is input into the preset initial fault prediction model for training, and an updated fault prediction model is generated. The historical data of the test set is input into the updated fault prediction model for testing, and a target fault prediction model is generated. Extract the feature vector from the preprocessed real-time data; The feature vector is input into the target fault prediction model.

4. The charging pile fault early warning management method according to claim 1, characterized in that, The step of comparing the fault probability score with a preset probability threshold, determining the fault information of the charging pile under test based on the comparison result, and sending the fault information and corresponding early warning information to the maintenance personnel includes: Determine whether the fault probability score is less than a preset first probability threshold; If so, then the current operating status of the charging pile under test is determined to be normal. If not, determine whether the fault probability score is greater than or equal to the first probability threshold and less than the preset second probability threshold; If so, then the current operating status of the charging pile under test is determined to be a status requiring attention; If not, then determine whether the fault probability score is greater than or equal to the second probability threshold and less than the preset third probability threshold; If so, it is determined that the current operating status of the charging pile under test has a potential fault risk, and the warning information corresponding to the potential fault risk is obtained and the potential fault risk and the warning information are sent to the operation and maintenance personnel. If not, then determine whether the fault probability score is greater than or equal to the third probability threshold; If so, it is determined that the current operating status of the charging pile under test is in the event of an impending fault, and the alarm information corresponding to the impending fault is obtained and sent to the maintenance personnel.

5. The charging pile fault early warning management method according to claim 4, characterized in that, Also includes: After the maintenance personnel handle the potential fault risk or the early warning case corresponding to the impending fault, they record the current maintenance data. Based on the feature vector of the real-time data, the fault probability score, and the operation and maintenance data, the accuracy of the target fault prediction model is determined. The accuracy rate is compared with a preset warning accuracy threshold, and the target fault prediction model is updated based on the comparison result to generate a new target fault prediction model.

6. A charging pile fault early warning management system, characterized in that, include: The preprocessing module is used to automatically acquire real-time data of the charging pile under test through robotic process and to preprocess the real-time data. The extraction module is used to extract the feature vectors of the preprocessed real-time data and input them into the preset target fault prediction model; The prediction module is used to predict the faults in the feature vector using the target fault prediction model and generate a fault probability score. The comparison module is used to compare the fault probability score with a preset probability threshold, determine the fault information of the charging pile under test based on the comparison result, and send the fault information and the corresponding early warning information to the operation and maintenance personnel.

7. The charging pile fault early warning management system according to claim 6, characterized in that, The preprocessing module includes: The data acquisition submodule is used to automatically collect real-time data of the charging pile under test from the sensors at regular intervals through a robotic process. The processing submodule is used to perform data cleaning and normalization processing on the real-time data in sequence.

8. An electronic device, characterized in that, The device includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor performs the charging pile fault early warning management method as described in any one of claims 1-5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the charging pile fault early warning management method as described in any one of claims 1-5.

10. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, wherein when the program instructions are executed by a computer, the computer performs the charging pile fault early warning management method as described in any one of claims 1-5.

Images