Power-off management method, device, equipment and storage medium for charging cabinet

Abstract

The present invention relates to the field of power-off control of charging cabinets, and in particular to a power-off management method, device, equipment and storage medium for charging cabinets. The method comprises the following steps: collecting real-time charging status information of the charging cabinet based on multiple sensors; performing time-series digital processing on the real-time charging status information, and filtering and noise reduction to obtain a noise-reduced optimized charging status data sequence; performing charging port status matching analysis on the noise-reduced optimized charging status data sequence one by one, and performing charging load calculation to obtain load data for each charging port; obtaining the charging cabinet historical operation log; performing user demand analysis on the charging cabinet historical operation log, and performing deep demand trend learning to generate user demand trend change rules. The present invention accurately locates the faulty charging port of the charging cabinet, realizes single-point power-off processing of the faulty charging port, and improves the operating efficiency of the charging cabinet.

Description

Technical Field

[0001] The present invention relates to the field of power-off control of charging cabinets, and in particular to a power-off management method, device, equipment and storage medium for a charging cabinet. Background Art

[0002] In today's society, shared power banks have become a popular and convenient charging service. People can easily obtain charging sources in various public places, which greatly improves the quality of life. However, shared charging cabinets also face some difficult technical problems in the long-term operation process.

[0003] As an electric device, shared charging cabinets are susceptible to various internal and external factors, and unexpected power outages or charging failures are unavoidable. Once such failures occur, it will not only affect the user's charging experience, but may also cause property losses. However, most shared charging cabinets on the market are still relatively backward in terms of fault prediction and power outage management technology.

[0004] When a traditional shared charging cabinet detects an abnormal situation in a charging port, it cannot accurately locate which charging port has a fault and often cuts off the power to the entire charging cabinet, causing other normal charging ports to be unable to operate. This will lead to waste of resources and create additional failure risks. In order to solve the above problems, there is an urgent need to develop an intelligent power-off management method for shared power bank charging cabinets. Summary of the invention

[0005] In order to solve the above-mentioned technical problems, the present invention proposes a power-off management method, device, equipment and storage medium for a charging cabinet to solve at least one of the above-mentioned technical problems.

[0006] To achieve the above object, the present invention provides a power-off management method for a charging cabinet, comprising the following steps:

[0007] Step S1: collecting real-time charging status information of the charging cabinet based on multiple sensors; performing time-series digital processing on the real-time charging status information, and performing filtering and noise reduction to obtain a noise reduction optimized charging status data sequence;

[0008] Step S2: performing charging port status matching analysis on the noise reduction optimized charging status data sequence one by one, and performing charging load calculation to obtain load data for each charging port;

[0009] Step S3: Obtain the historical operation log of the charging cabinet; perform user demand analysis on the historical operation log of the charging cabinet, and conduct in-depth demand trend learning to generate the changing rules of user demand trends;

[0010] Step S4: performing multiple rounds of load iteration simulation on the load data of each charging port according to the changing rules of user demand trends, and performing short-term charging load prediction to construct a charging load prediction time series curve;

[0011] Step S5: locating the abnormal charging port on the charging load prediction time series curve based on the preset theoretical charging load curve, and performing fault probability prediction, thereby obtaining fault probability prediction data of the abnormal charging port;

[0012] Step S6: Perform single-point power outage processing based on the fault probability prediction data of the abnormal charging port, perform intelligent decision optimization, and build an intelligent prediction power outage management model.

[0013] The present invention also provides a power failure management device for a charging cabinet, comprising:

[0014] A data optimization module is used to collect real-time charging status information of the charging cabinet based on multiple sensors; perform time-series digital processing on the real-time charging status information, and perform filtering and noise reduction to obtain a noise-reduced and optimized charging status data sequence;

[0015] A load calculation module is used to perform a charging port state matching analysis on the noise reduction optimized charging state data sequence one by one, and perform charging load calculation to obtain the load data of each charging port;

[0016] The demand trend module is used to obtain the historical operation log of the charging cabinet; perform user demand analysis on the historical operation log of the charging cabinet, and conduct in-depth demand trend learning to generate the changing rules of user demand trends;

[0017] The load prediction module is used to perform multiple rounds of load iteration simulation on the load data of each charging port according to the changing rules of user demand trends, and to perform short-term charging load prediction and construct a charging load prediction time series curve;

[0018] A fault prediction module is used to locate abnormal charging ports based on a preset theoretical charging load curve and a charging load prediction time series curve, and to perform fault probability prediction, thereby obtaining fault probability prediction data of the abnormal charging port;

[0019] The power outage management module is used to handle single-point power outages based on the fault probability prediction data of abnormal charging ports, perform intelligent decision-making optimization, and build an intelligent predictive power outage management model.

[0020] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the power off management method of the charging cabinet described in any one of the above items are implemented.

[0021] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the power off management method for a charging cabinet described in any one of the above items are implemented.

[0022] The beneficial effects of the present invention are specifically as follows: by collecting the real-time charging status information of the charging cabinet through multiple sensors, performing time-series digital processing and filtering noise reduction, the charging status changes can be accurately captured, the data quality and accuracy can be improved, and the noise reduction and optimized charging status data sequence can be obtained, which is helpful to eliminate interference and noise, and provide a clear data basis for subsequent analysis; the state matching analysis and charging load calculation of each charging port can help to understand the working status and load condition of each charging port, and provide specific data support for power outage management; obtaining the load data of each charging port can help monitor the operating status of the charging cabinet and discover abnormal situations in time; analyzing the historical operation log of the charging cabinet can understand the usage and potential problems of the charging cabinet, and provide historical reference for power outage management; generating the trend change law of user demand can help predict the change of user demand, and provide more support for power outage management. Add intelligent decision-making basis, multiple rounds of load iteration simulation and short-term charging load prediction can help predict the changing trend of charging load, prepare and adjust charging management strategy in advance, build charging load prediction time series curve to help visualize load changes, and provide a basis for abnormal detection and prevention. Perform anomaly detection on charging load prediction time series curve according to preset theoretical charging load curve, which helps to quickly locate abnormal charging ports and improve safety and efficiency. Predicting the failure probability of abnormal charging ports can help to provide early warning and take corresponding maintenance measures to reduce the risk of failure. Single-point power outage processing and intelligent decision optimization based on failure probability prediction data of abnormal charging ports can improve the safety and reliability of charging cabinets. Building an intelligent predictive power outage management model helps to optimize the power outage management process and improve the overall efficiency and operation quality of charging equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic flow chart of the steps of a power failure management method for a charging cabinet according to the present invention;

[0024] Figure 2 Detailed implementation flow chart of step S1;

[0025] Figure 3 Detailed implementation flow chart of step S2;

[0026] Figure 4 Detailed implementation flow chart of step S3. DETAILED DESCRIPTION

[0027] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0028] The present application example provides a power-off management method, device, equipment and storage medium for a charging cabinet. The execution subjects of the power-off management method, device, equipment and storage medium for the charging cabinet include but are not limited to: mechanical equipment, data processing platform, cloud server node, network upload equipment, etc. equipped with the system can be regarded as the general computing node of the present application, and the data processing platform includes but is not limited to: at least one of an audio and image management system, an information management system, and a cloud data management system.

[0029] See also Figures 1 to 4 The present invention provides a power-off management method for a charging cabinet, and the power-off management method for a charging cabinet comprises the following steps:

[0030] Step S1: collecting real-time charging status information of the charging cabinet based on multiple sensors; performing time-series digital processing on the real-time charging status information, and performing filtering and noise reduction to obtain a noise reduction optimized charging status data sequence;

[0031] Step S2: performing charging port status matching analysis on the noise reduction optimized charging status data sequence one by one, and performing charging load calculation to obtain load data for each charging port;

[0032] Step S3: Obtain the historical operation log of the charging cabinet; perform user demand analysis on the historical operation log of the charging cabinet, and conduct in-depth demand trend learning to generate the changing rules of user demand trends;

[0033] Step S4: performing multiple rounds of load iteration simulation on the load data of each charging port according to the changing rules of user demand trends, and performing short-term charging load prediction to construct a charging load prediction time series curve;

[0034] Step S5: locating the abnormal charging port on the charging load prediction time series curve based on the preset theoretical charging load curve, and performing fault probability prediction, thereby obtaining fault probability prediction data of the abnormal charging port;

[0035] Step S6: Perform single-point power outage processing based on the fault probability prediction data of the abnormal charging port, perform intelligent decision optimization, and build an intelligent prediction power outage management model.

[0036] By collecting the real-time charging status information of the charging cabinet through multiple sensors and performing time-series digital processing and filtering noise reduction, the charging status changes can be accurately captured, the data quality and accuracy can be improved, and the noise reduction and optimized charging status data sequence can be obtained, which helps to eliminate interference and noise, and provide a clear data basis for subsequent analysis. The status matching analysis and charging load calculation of each charging port can help to understand the working status and load conditions of each charging port, and provide specific data support for power outage management. Obtaining the load data of each charging port can help monitor the operating status of the charging cabinet and detect abnormal conditions in time. Analyzing the historical operation log of the charging cabinet can understand the usage and potential problems of the charging cabinet, provide historical references for power outage management, and generate user demand trend change rules to help predict user demand changes and provide more intelligent decision-making for power outage management. Based on this, multiple rounds of load iteration simulation and short-term charging load prediction can help predict the changing trend of the charging load, prepare and adjust the charging management strategy in advance, and construct a charging load prediction time series curve to help visualize load changes and provide a basis for abnormal detection and prevention. Anomaly detection of the charging load prediction time series curve based on the preset theoretical charging load curve can help quickly locate abnormal charging ports and improve safety and efficiency. Predicting the failure probability of abnormal charging ports can help provide early warning and take corresponding maintenance measures to reduce the risk of failure. Single-point power outage processing and intelligent decision-making optimization based on the failure probability prediction data of abnormal charging ports can improve the safety and reliability of charging cabinets. Constructing an intelligent predictive power outage management model can help optimize the power outage management process and improve the overall efficiency and operation quality of charging equipment.

[0037] In the embodiment of the present invention, refer to Figure 1 , is a schematic flow chart of the steps of a power outage management method for a charging cabinet of the present invention. In this example, the steps of the power outage management method for the charging cabinet include:

[0038] Step S1: collecting real-time charging status information of the charging cabinet based on multiple sensors; performing time-series digital processing on the real-time charging status information, and performing filtering and noise reduction to obtain a noise reduction optimized charging status data sequence;

[0039] In this embodiment, a microcontroller (such as Arduino, Raspberry Pi or PLC) is used to connect all sensors to collect data in real time. The collected data is transmitted to the central processing unit in real time via wireless (such as Wi-Fi, Bluetooth) or wired (such as RS-485, Ethernet). A data acquisition program is written to read the sensor data regularly (for example, once every second), and store the data in the memory or directly transmit it to the database. The collected real-time charging status information is uniformly formatted to prepare for subsequent processing. The data format usually includes timestamp, sensor type, reading, etc. The real-time data from different sensors are integrated together to form a time-series data stream for subsequent processing. A timestamp is added to each data point to ensure the time series of the data point so that it can be processed later. Continue to analyze and process, sample and quantize the analog signal, and convert it into a digital signal. Use ADC (analog-to-digital converter) to convert the analog output of the sensor into a digital value, and store the digitized data in a database or real-time data stream for subsequent analysis. Select a suitable filter based on the data characteristics, input the real-time charging status data into the selected filter for filtering. When using a low-pass filter, you can set the cutoff frequency to remove noise exceeding this frequency, and save the filtered and denoised charging status data sequence into a database to form a noise-reduced and optimized charging status data sequence. Verify the denoised data sequence to ensure the smoothness and authenticity of the data and check whether it still reflects the actual charging status.

[0040] Step S2: performing charging port status matching analysis on the noise reduction optimized charging status data sequence one by one, and performing charging load calculation to obtain load data for each charging port;

[0041] In this embodiment, the charging state data sequence after noise reduction optimization is extracted from the database to ensure data integrity and timestamp consistency, and the basic information of the charging port, including the number, location, rated power, etc. of the charging port, is integrated to form a charging port information table to facilitate subsequent matching, and a suitable state matching algorithm is selected. Commonly used methods include dynamic time warping (DTW) or timestamp-based matching methods. State matching analysis is performed on each charging port, and the noise reduction data is aligned with the state data of each charging port according to the timestamp to ensure the timing of the data. Real-time data related to each charging port, including current, voltage, charging time, etc., is extracted, and the extracted charging port state data is organized into a structured format, such as a data frame (DataFrame), to facilitate subsequent processing and calculation, and the current and voltage of each charging port are matched. The pressure data are calculated one by one to generate charging load data, the real-time current and voltage values ​​of each charging port are extracted from the matched data, the extracted current and voltage values ​​are substituted into the load calculation formula to obtain the real-time load value of each charging port, and the calculated load data of each charging port is saved in a database or data table with a timestamp and charging port identification to ensure the traceability of the data. The calculated load data is verified to check its rationality and consistency. The verification can be carried out in the following ways: comparing the calculation results with the theoretical load values ​​to check deviations, using statistical methods to detect abnormal values ​​in the load data (such as Z-Score detection), and performing preliminary analysis on the load data of each charging port, such as calculating the average load, maximum load, load fluctuation, etc., to provide a basis for subsequent decision-making.

[0042] Step S3: Obtain the historical operation log of the charging cabinet; perform user demand analysis on the historical operation log of the charging cabinet, and conduct in-depth demand trend learning to generate the changing rules of user demand trends;

[0043] In this embodiment, the storage location of the historical operation log of the charging cabinet is determined, which is usually a database, file system or cloud storage. The characteristics of user needs are determined, including: user active time period (such as peak time period), average charging time, charging frequency, user behavior pattern (such as repeated charging users), data cleaning of the extracted historical operation logs, processing of missing values, outliers and duplicate records, ensuring the accuracy of the analyzed data, statistical analysis of the cleaned data, and calculation of the statistical values ​​of various demand characteristics, such as: charging frequency of each user, total charging amount in each time period, user distribution in peak time period, Pandas, NumPy can be used Use data analysis libraries such as github to process and count data, select a suitable deep learning model for demand trend learning, commonly used models include: LSTM (Long Short-Term Memory Network): suitable for time series data, GRU (Gated Recurrent Unit): another effective model for processing time series, Transformer: suitable for processing complex sequence data, organize demand features into time series format, ensure that data at each time point can be used for model training, data normalization may be required to improve the stability of model training, input the organized data into the selected deep learning model for training, and set appropriate hyperparameters during training, such as learning rate, batch size, and number of training rounds. Use cross-validation or validation sets to monitor model performance and prevent overfitting.

[0044] Step S4: performing multiple rounds of load iteration simulation on the load data of each charging port according to the changing rules of user demand trends, and performing short-term charging load prediction to construct a charging load prediction time series curve;

[0045] In this embodiment, the real-time load data of each charging port is extracted from the previous steps, including the current, voltage and the calculated charging load, and the user demand trend change law is extracted, including information such as peak hours, user charging habits and expected load demand. A load iteration simulation model is designed, usually using Monte Carlo simulation or a random simulation method based on historical data. The model should consider the volatility of user demand and the maximum carrying capacity of the charging port, and set simulation parameters, including: simulation rounds (such as 1000 times), the time period of each round of simulation (such as every hour), the user demand fluctuation range (peak and trough). In each round of simulation, the load data of each charging port is generated according to the user demand trend and historical load data: the amplitude and direction of the load change are randomly selected to simulate the actual workload of the charging port, and the simulation results of each round are recorded, including the load data of each charging port. A suitable short-term load prediction model is selected, such as: linear regression: simple and effective, suitable for linear trends, ARIMA: suitable for time series data, considering autocorrelation, LSTM: suitable for processing complex time series data, able to capture long-term and short-term dependencies, and the simulated load data is organized into a time series grid. The model is constructed by using the time series format, including timestamp, charging port identification and load value, and performing data normalization to improve the stability of model training. The selected prediction model is trained using historical load data and simulation data. Cross-validation or holdout method can be used for model verification to ensure the effectiveness and accuracy of the model. The trained model is used to make short-term predictions of the load in future time periods, and short-term load prediction data for each charging port is generated. The short-term load prediction data is organized into a time series format to ensure that the data at each time point can be used to construct the timing curve. The charging load prediction timing curve is drawn using visualization tools (such as Matplotlib and Seaborn): the X-axis is time and the Y-axis is load value. The predicted value is compared with the actual load data to facilitate observation of the accuracy of the prediction. The results are saved and displayed: the generated timing curve graph is saved as a report or dashboard for relevant personnel to view and analyze, and the prediction results are recorded in the database for subsequent use and analysis.

[0046] Step S5: locating the abnormal charging port on the charging load prediction time series curve based on the preset theoretical charging load curve, and performing fault probability prediction, thereby obtaining fault probability prediction data of the abnormal charging port;

[0047] In this embodiment, a preset theoretical charging load curve is determined and obtained, which is usually based on the normal load mode obtained by historical data, industry standards or user demand analysis. The theoretical load curve is standardized to ensure that it has the same time unit and data structure as the charging load prediction timing curve, which is convenient for subsequent comparison and selection of a suitable anomaly detection method, for example: Threshold-based method: Set a load deviation threshold. When the deviation between the predicted load and the theoretical load exceeds the threshold, it is marked as abnormal. Z-Score Method: Calculate the Z-Score of the predicted load and the theoretical load. When the Z-Score exceeds the set value, mark it as abnormal. Dynamic Time Warping (DTW): Used to measure the similarity between two time series and find the part that deviates from the normal mode. Compare the predicted load timing curve of each charging port with the theoretical load curve to identify the charging port with abnormal load: Calculate the load deviation at each time point, apply the selected anomaly detection method, mark the abnormal charging port and its abnormal time period, and record the detected abnormal charging port and its abnormal state in the database with a timestamp and the nature of the anomaly for subsequent analysis. Select a suitable fault probability prediction model, such as: Logistic regression: Suitable for processing binary classification problems, and can output the probability of an event. Support vector machine (SVM): suitable for handling classification problems with complex decision boundaries, random forest: improves prediction performance by integrating multiple decision trees, neural network: suitable for handling complex nonlinear relationships, especially when the amount of data is large, collect relevant data of abnormal charging ports, including: historical fault records, current load data, charging status information (such as current, voltage, temperature, etc.), other available features (such as time, date), use labeled historical data to train the selected model to learn the pattern of fault occurrence, use cross-validation method to evaluate the performance of the model, adjust hyperparameters to optimize the prediction results, use the trained model to predict the fault probability of the detected abnormal charging port, and output the fault probability value of each abnormal charging port.

[0048] Step S6: Perform single-point power outage processing based on the fault probability prediction data of the abnormal charging port, perform intelligent decision optimization, and build an intelligent prediction power outage management model.

[0049] In this embodiment, the fault probability prediction results are extracted from the previous steps, and abnormal charging ports with a fault probability higher than a preset threshold are identified. These charging ports need to be powered off immediately to prevent potential safety hazards. Ensure that the charging cabinet is equipped with a single-point power-off control system, which usually includes an intelligent switch or relay that can control the power supply of the charging port in real time according to instructions. For each identified abnormal charging port, a power-off instruction is issued. This can be achieved by writing a control program. The program logic is as follows: Check the fault probability data. If the fault probability is ≥ the preset threshold, perform the power-off operation. After the power is off, collect the power-off response data in real time, including: the current, voltage, and power before the power is off, the response time after the power is off, and the state change of the charging port (such as from "charging" to "power off"). The collected power-off response data is stored in the database for subsequent analysis and model training. According to the response data and the fault prediction results, an intelligent decision-making optimization framework is designed. The framework can be optimized using reinforcement learning (such as Q-Learning or deep Q network) or a rule-based decision system.

[0050] In this embodiment, refer to Figure 2 , is a flowchart of detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include:

[0051] Step S11: Collecting real-time charging status information of the charging cabinet based on multiple sensors;

[0052] Step S12: performing time-series digital processing on the real-time charging status information to generate a charging status data sequence;

[0053] Step S13: Calculate the parameter mean of the charging state data sequence and extract the state parameter average value;

[0054] Step S14: performing abnormal outlier detection based on the average value of the state parameters and marking abnormal outlier data points;

[0055] Step S15: performing outlier filtering on abnormal outlier data points to obtain an outlier optimized charging state data sequence;

[0056] Step S16: filtering and denoising the outlier optimized charging state data sequence to obtain a denoised optimized charging state data sequence.

[0057] In this embodiment, suitable sensors (such as current sensors, voltage sensors, temperature sensors, etc.) are selected to monitor different states of the charging cabinet, and the sensors are installed at key positions of the charging cabinet to ensure that the charging state information can be accurately collected. A data acquisition system is designed, and a microcontroller (such as Arduino or Raspberry Pi) is used to connect all sensors to obtain charging state data in real time. The collected real-time data is transmitted to a data processing server through a wireless network or a wired network to ensure the reliability and timeliness of data transmission. The real-time collected charging state information is formatted to ensure a unified data structure (such as timestamp, sensor type, reading, etc.), and a timestamp is added to each data point for subsequent time series analysis. The analog signal is digitized (such as sampling and quantization) to generate a digitized charging state data sequence to ensure that the data can be used for subsequent analysis. The charging state data sequence is grouped according to time periods (such as every minute, every hour), and the parameter mean is calculated for the data in each time period. The average value of state parameters such as current, voltage, and temperature is extracted, and the calculated average value of the state parameters is saved in a database or a data table for subsequent processing and analysis. A suitable anomaly detection algorithm is selected, such as Z-Score. Method, IQR (interquartile range) method or machine learning method (such as isolation forest), perform anomaly detection on the extracted state parameter average value, mark outlier data points, for Z-Score method, usually set a threshold (such as Z value greater than 3) to judge the anomaly, record the detected abnormal outlier data points for subsequent processing, delete these abnormal data from the charging state data sequence according to the previously marked abnormal outlier data points, reorganize the filtered data to ensure data continuity and integrity, save the charging state data sequence after outlier optimization to a new data table or database for subsequent use, select a suitable filter (such as low-pass filter, Kalman filter, mean filter, etc.) for data denoising, apply the selected filter to the charging state data sequence after outlier optimization, filter the data according to the set parameters, verify the denoised data sequence, check the smoothness and authenticity of the data, and ensure that the data still reflects the true charging state.

[0058] In this embodiment, refer to Figure 3 , is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include:

[0059] Step S21: Identify the charging port distribution of the charging cabinet to obtain charging port distribution data;

[0060] Step S22: performing charging port status matching analysis on the noise reduction optimized charging status data sequence one by one based on the charging port distribution data, so as to obtain real-time charging status data of each charging port;

[0061] Step S23: Calculate the current and voltage of each charging port in real time based on the real-time charging status data to generate the real-time current and voltage parameters of each charging port;

[0062] Step S24: Calculate the charging load based on the real-time current and voltage parameters of each charging port to obtain the load data of each charging port.

[0063] In this embodiment, computer vision technology (such as image processing and deep learning) is used to analyze the image of the charging cabinet to identify the specific location and number of the charging ports. The image data of the charging cabinet is collected by a high-resolution camera or a laser scanner. These images should include multiple angles to ensure comprehensive recognition of the charging ports. The collected images are preprocessed, including denoising, contrast enhancement, etc., to improve the accuracy of subsequent recognition. The location of the charging port is identified by a feature detection algorithm (such as SIFT, SURF, or CNN) and marked on the image. The identified charging port information (including location coordinates, quantity, etc.) is recorded as charging port distribution data and stored in a database for subsequent analysis. The charging port distribution data is associated with a noise reduction optimized charging status data sequence to ensure that the status data of each charging port can correspond to a specific charging port. Status matching analysis is performed on each charging port one by one. The charging status data is matched with the corresponding charging port through a timestamp or other identifier, and each charging port is extracted from the matched data. The real-time charging status information of each charging port (such as charging current, charging voltage, etc.) is collected and organized into structured data. Based on the real-time charging status data of each charging port, the current and voltage parameters are calculated in real time. The data read by the sensor can be used directly, or the calculation can be performed using a formula. For possible abnormal data (such as current and voltage beyond the normal range), abnormal value judgment and processing are performed to ensure the accuracy of the calculation results. The calculated real-time current and voltage parameters of each charging port are recorded in a data table with a timestamp and charging port identification information. The charging load of each charging port is calculated using the current and voltage parameters: load (W) = current (A) × voltage (V). The above formula is applied to the real-time current and voltage parameters of each charging port to calculate the corresponding charging load. The calculated load data of each charging port is saved in the database to ensure real-time update and traceability of the data. The charging load data is monitored regularly, historical data analysis is performed, charging performance trends are identified, and a basis is provided for subsequent optimization decisions.

[0064] In this embodiment, refer to Figure 4 , is a flowchart of detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include:

[0065] Step S31: Obtain the historical operation log of the charging cabinet;

[0066] Step S32: Divide the historical operation log of the charging cabinet into sliding windows, so as to obtain operation logs of multiple windows;

[0067] Step S33: performing statistics on the power usage time distribution of users in each window on the operation logs of multiple windows, so as to obtain the power usage time distribution characteristics of users in each window;

[0068] Step S34: Performing user demand analysis on the user power consumption time distribution characteristics of each window, and extracting user demand data of each window;

[0069] Step S35: mining the demand fluctuation evolution of the user demand data in each window to generate charging cabinet user demand evolution data;

[0070] Step S36: Conduct deep demand trend learning on the charging cabinet user demand evolution data to generate user demand trend change rules.

[0071] In this embodiment, it is ensured that the charging cabinet has the function of recording historical operation logs, including the user's charging time, charging power, charging status, abnormal events and other information. The operation logs are stored in the database to ensure the integrity and queryability of the data. A relational database (such as MySQL) or a non-relational database (such as MongoDB) can be used. According to the needs, the historical operation logs of the charging cabinet are exported regularly to ensure that the latest user charging data is obtained. The size and step size of the sliding window are determined. For example, the window size can be set to one hour and the step size can be set to 30 minutes. The historical operation logs are divided according to the defined sliding windows. Each window contains all the operation log records within a certain time period. The data of the divided multiple windows are saved as structured data for subsequent analysis. Lists or data frames (such as Pandas) can be used to store these data. The user's power usage time information is extracted from the operation log of each window, which usually includes the time when charging starts. The time between the start and end of charging is counted, and the power consumption time in each window is statistically analyzed to calculate the total power consumption time, the average power consumption time, and the distribution of the power consumption time (such as frequency distribution). The user power consumption time distribution characteristics of each window are recorded in the data table for subsequent analysis to determine the definition of user demand characteristics, which may include charging frequency, average charging time, user active time period, etc. Based on the power consumption time distribution characteristics of each window, the user's charging demand is analyzed, for example, the peak time period and the user's preferred charging time are identified, and the analysis results are organized into structured data to form user demand data for each window. The user demand data of multiple windows are compared and analyzed to identify the fluctuation trend of demand, for example, the peak and trough of demand change are detected through time series analysis, and the demand change of each window is recorded to generate user demand evolution data, which usually includes time series data and fluctuation trend analysis results, and a suitable deep learning model (such as LSTM, GRU) is selected. Or Transformer) for demand trend learning. These models are particularly suitable for processing time series data. They organize the user demand evolution data into a format suitable for model input, including feature selection and data normalization. The prepared data is used to train the deep learning model and optimize the model parameters to identify the changing patterns of user demand. After the model training is completed, the trained model is used to predict future user demand and generate the changing patterns of user demand trends. The changing patterns of user demand trends are applied to the charging cabinet management system to optimize the allocation and scheduling of charging resources and improve user experience.

[0072] In this embodiment, step S4 includes the following steps:

[0073] Step S41: performing multiple rounds of load iteration simulation on the load data of each charging port according to the changing rule of the user demand trend, and generating multiple charging port load simulation data;

[0074] Step S42: Calculating the load average of multiple charging port load simulation data to obtain the charging port load simulation average;

[0075] Step S43: performing short-term charging load prediction according to the charging port load simulation mean value to generate short-term charging load prediction data;

[0076] Step S44: Perform time series distribution fitting on the short-term charging load prediction data to construct a charging load prediction time series curve.

[0077] In this embodiment, key parameters such as peak hours, valley hours, and variation range are extracted from the previously generated user demand trend variation law to guide load simulation and design a load simulation model. A random simulation method based on historical load data can usually be used. The model needs to consider the volatility of user demand and the actual capacity of the charging port. Multiple rounds of load iteration simulation are performed. Each round of simulation generates load data for each charging port based on changes in user demand trends. The results of each round of simulation are recorded, including the load value of each charging port in different time periods. The generated multiple charging port load simulation data are saved in a database or data frame for subsequent calculation and analysis. The load simulation data of multiple charging ports are organized into structured data to ensure that the load data of each charging port in the same time period can correspond. The load data of each charging port in all simulation rounds are averaged, and the calculated charging port load simulation average is recorded in a new data table for subsequent use. A suitable short-term load forecasting model is selected, such as linear regression, ARIMA, SARIMA Or deep learning model (such as LSTM), select according to the nature of the data and prediction needs, organize the simulated mean of the charging port load into time series data, ensure that the data structure is suitable for input into the prediction model, use historical load data to train the model, verify the prediction performance of the model, and use cross-validation or holdout method to evaluate the accuracy of the model. Based on the trained model, perform short-term load prediction on the simulated mean of the charging port load to generate short-term charging load prediction data, and save the generated short-term charging load prediction data to the database for subsequent analysis and decision support. Select appropriate time series analysis tools and methods, such as time series analysis. The short-term charging load forecast data can be fitted with a time series distribution using a solution, smoothing method (such as moving average) or a fitting method (such as polynomial fitting) to identify trends, seasonality and residuals in the data. The least squares method or other optimization methods can be used for fitting. A charging load forecast time series curve can be constructed based on the fitting results to graphically display the changing trend of the charging load. This will help to intuitively understand the load changes and verify the consistency between the fitting curve and the actual load forecast data. Adjustments can be made when necessary to improve the accuracy and reliability of the forecast. The constructed charging load forecast time series curve can be displayed on a dashboard or report to facilitate decision-making and optimization by relevant personnel.

[0078] In this embodiment, step S5 includes the following steps:

[0079] Step S51: performing abnormal comparison and identification on the charging load prediction time series curve based on the preset theoretical charging load curve, and extracting the abnormal change phase curve;

[0080] Step S52: extract multiple points from the abnormal change phase curve to obtain multiple abnormal change curve points;

[0081] Step S53: performing abnormal position feature analysis on multiple abnormal change curve points, thereby generating abnormal stage position features;

[0082] Step S54: locating the abnormal charging port of the charging cabinet according to the position characteristics of the abnormal stage, and marking the abnormal charging port;

[0083] Step S55: performing a fault probability prediction on the abnormal charging port, thereby obtaining fault probability prediction data of the abnormal charging port.

[0084] In this embodiment, a preset theoretical charging load curve is determined, which is usually based on historical data or industry standards and reflects the charging load mode under normal working conditions. A suitable anomaly detection method is selected, such as the Z-Score method, dynamic time warping (DTW) or a threshold-based comparison method. The charging load prediction timing curve is compared with the theoretical charging load curve to identify the abnormal change stage that deviates from the normal range. The time period of all abnormal changes is recorded, and the corresponding change stage curve is extracted. The multi-point extraction method is determined, and a peak detection algorithm, an inflection point detection or a threshold-based segmentation method can be used to perform multi-point extraction on each abnormal change stage curve to find key change points, such as the maximum value, the minimum value and the point with rapid change. Multiple key points of each abnormal change curve and their corresponding timestamps are recorded. A suitable feature analysis method is selected, such as principal component analysis (PCA), cluster analysis or statistical feature extraction method, and multiple abnormal change curve points are analyzed to extract position features, such as change amplitude, change rate, duration, etc. The generated abnormal stage position features are recorded in a data table to provide a basis for subsequent charging port positioning. A suitable positioning algorithm is selected, and a feature matching-based method or machine learning-based positioning algorithm can be used. A deep learning model (such as decision tree and random forest) is used to identify abnormal charging ports, and the location information of the charging ports is integrated with the location features of the abnormal stage to ensure that the status and features of each charging port correspond. According to the extracted location features of the abnormal stage, each charging port of the charging cabinet is analyzed, and the abnormal charging ports are identified and marked. A suitable fault probability prediction model is selected, such as logistic regression, support vector machine (SVM) or deep learning model (such as neural network). The status data and features of the abnormal charging ports are organized into a model input format, including historical fault records, load data and abnormal detection results. The selected fault probability prediction model is trained and verified using historical data to ensure the accuracy and robustness of the model. Based on the trained model, the fault probability of the marked abnormal charging ports is predicted to generate fault probability prediction data, and the prediction results are recorded in the database for subsequent maintenance and decision support. At the same time, priority maintenance and resource allocation of the charging ports can be performed according to the failure probability.

[0085] In this embodiment, step S6 includes the following steps:

[0086] Step S61: Calculating the failure risk prediction of the failure probability prediction data of the abnormal charging port to obtain a failure risk prediction value of the charging port;

[0087] Step S62: comparing the fault risk prediction value of the charging port with the preset charging port risk threshold, and generating a charging port fault warning signal when the fault risk prediction value of the charging port is greater than or equal to the preset charging port risk threshold;

[0088] Step S63: performing single-point power-off processing on the abnormal charging port based on the charging port fault warning signal, and collecting charging port power-off response data;

[0089] Step S64: Perform intelligent decision optimization on the charging port power outage response data and build an intelligent prediction power outage management model.

[0090] In this embodiment, a suitable fault risk prediction model is selected. The models that may be used include logistic regression, decision tree or other machine learning models. The historical fault data is used for training, and the fault probability prediction data of the abnormal charging port is collected, including historical charging data, load data and related environmental parameters, and organized into a model input format. The prepared data is input into the fault risk prediction model, and the fault risk prediction value of the charging port is calculated. The output probability of the model can be used to represent the possibility of a fault, and a preset threshold value of the charging port risk is determined. The threshold value can be set based on historical data analysis and expert experience, usually between 0 and 1. The calculated fault risk prediction value is compared with the preset risk threshold. If the prediction value is greater than or equal to the threshold, it means that the charging port has a high fault risk. When the fault risk prediction value exceeds the threshold, a charging port fault warning signal is generated, and the relevant maintenance personnel can be notified through system notification, alarm light or SMS. Design or use the existing single-point power-off control system, and power off the abnormal charging port through the power controller or intelligent switch. After the fault warning signal is generated, a power-off command is immediately issued to cut off the power supply of the charging port. After the power-off processing, the power-off response data of the charging port is collected in real time, including the power-off time, the current and voltage changes before the power-off, and other information. Select the appropriate data Analytical tools and algorithms, such as machine learning algorithms (such as random forests, support vector machines) or deep learning models (such as neural networks), organize the power outage response data and extract relevant features, such as power outage reaction time, load change rate, and environmental factors. The extracted features are used to build an intelligent predictive power outage management model, and the model is trained to optimize power outage decisions. Cross-validation can be used to evaluate the performance of the model. According to the prediction results of the model, the power outage decision process of the charging port is optimized to ensure that power outages can be handled quickly and effectively when there is a risk of failure. The results of the constructed intelligent predictive power outage management model are recorded and integrated into the charging cabinet management system to improve the reliability and safety of the system.

[0091] The present invention also provides a power failure management device for a charging cabinet, comprising:

[0092] A data optimization module is used to collect real-time charging status information of the charging cabinet based on multiple sensors; perform time-series digital processing on the real-time charging status information, and perform filtering and noise reduction to obtain a noise-reduced and optimized charging status data sequence;

[0093] A load calculation module is used to perform a charging port state matching analysis on the noise reduction optimized charging state data sequence one by one, and perform charging load calculation to obtain the load data of each charging port;

[0094] The demand trend module is used to obtain the historical operation log of the charging cabinet; perform user demand analysis on the historical operation log of the charging cabinet, and conduct in-depth demand trend learning to generate the changing rules of user demand trends;

[0095] The load prediction module is used to perform multiple rounds of load iteration simulation on the load data of each charging port according to the changing rules of user demand trends, and to perform short-term charging load prediction and construct a charging load prediction time series curve;

[0096] A fault prediction module is used to locate abnormal charging ports based on a preset theoretical charging load curve and a charging load prediction time series curve, and to perform fault probability prediction, thereby obtaining fault probability prediction data of the abnormal charging port;

[0097] The power outage management module is used to handle single-point power outages based on the fault probability prediction data of abnormal charging ports, perform intelligent decision-making optimization, and build an intelligent predictive power outage management model.

[0098] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the power off management method of the charging cabinet described in any one of the above items are implemented.

[0099] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the power off management method for a charging cabinet described in any one of the above items are implemented.

[0100] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the transformer abnormality monitoring method described in any one of the above items are implemented.

[0101] Those skilled in the art clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0102] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it is stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or partly or all or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media for storing program codes such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks or optical disks.

[0103] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0104] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A power-off management method for a charging cabinet, characterized in that: The following steps are involved: Step S1: collecting real-time charging status information of the charging cabinet based on multiple sensors; performing time-series digital processing on the real-time charging status information, and performing filtering and noise reduction to obtain a noise reduction optimized charging status data sequence; Step S2: performing charging port status matching analysis on the noise reduction optimized charging status data sequence one by one, and performing charging load calculation to obtain load data for each charging port; Step S3: Obtain the historical operation log of the charging cabinet; perform user demand analysis on the historical operation log of the charging cabinet, and conduct in-depth demand trend learning to generate the changing rules of user demand trends; Step S4: performing multiple rounds of load iteration simulation on the load data of each charging port according to the changing rules of user demand trends, and performing short-term charging load prediction to construct a charging load prediction time series curve; Step S5: locating the abnormal charging port on the charging load prediction time series curve based on the preset theoretical charging load curve, and performing fault probability prediction, thereby obtaining fault probability prediction data of the abnormal charging port; Step S6: Perform single-point power outage processing based on the fault probability prediction data of the abnormal charging port, perform intelligent decision optimization, and build an intelligent prediction power outage management model.

2. The power-off management method for a charging cabinet according to claim 1, characterized in that: The specific steps of step S1 are: Step S11: Collecting real-time charging status information of the charging cabinet based on multiple sensors; Step S12: performing time-series digital processing on the real-time charging status information to generate a charging status data sequence; Step S13: Calculate the parameter mean of the charging state data sequence and extract the state parameter average value; Step S14: performing abnormal outlier detection based on the average value of the state parameters and marking abnormal outlier data points; Step S15: performing outlier filtering on abnormal outlier data points to obtain an outlier optimized charging state data sequence; Step S16: filtering and denoising the outlier optimized charging state data sequence to obtain a denoised optimized charging state data sequence.

3. The power failure management method of the charging cabinet according to claim 1, characterized in that: The specific steps of step S2 are: Step S21: Identify the charging port distribution of the charging cabinet to obtain charging port distribution data; Step S22: performing charging port status matching analysis on the noise reduction optimized charging status data sequence one by one based on the charging port distribution data, so as to obtain real-time charging status data of each charging port; Step S23: Calculate the current and voltage of each charging port in real time based on the real-time charging status data to generate the real-time current and voltage parameters of each charging port; Step S24: Calculate the charging load based on the real-time current and voltage parameters of each charging port to obtain the load data of each charging port.

4. The power-off management method for a charging cabinet according to claim 1, characterized in that: The specific steps of step S3 are: Step S31: Obtain the historical operation log of the charging cabinet; Step S32: Divide the historical operation log of the charging cabinet into sliding windows, so as to obtain operation logs of multiple windows; Step S33: performing statistics on the power usage time distribution of users in each window on the operation logs of multiple windows, so as to obtain the power usage time distribution characteristics of users in each window; Step S34: Performing user demand analysis on the user power consumption time distribution characteristics of each window, and extracting user demand data of each window; Step S35: mining the demand fluctuation evolution of the user demand data in each window to generate charging cabinet user demand evolution data; Step S36: Conduct deep demand trend learning on the charging cabinet user demand evolution data to generate user demand trend change rules.

5. The power-off management method of the charging cabinet according to claim 1, characterized in that: The specific steps of step S4 are: Step S41: performing multiple rounds of load iteration simulation on the load data of each charging port according to the changing rule of the user demand trend, and generating multiple charging port load simulation data; Step S42: Calculating the load average of multiple charging port load simulation data to obtain the charging port load simulation average; Step S43: performing short-term charging load prediction according to the charging port load simulation mean value to generate short-term charging load prediction data; Step S44: Perform time series distribution fitting on the short-term charging load prediction data to construct a charging load prediction time series curve.

6. The power-off management method for a charging cabinet according to claim 1, characterized in that: The specific steps of step S5 are: Step S51: performing abnormal comparison and identification on the charging load prediction time series curve based on the preset theoretical charging load curve, and extracting the abnormal change phase curve; Step S52: extract multiple points from the abnormal change phase curve to obtain multiple abnormal change curve points; Step S53: performing abnormal position feature analysis on multiple abnormal change curve points, thereby generating abnormal stage position features; Step S54: locating the abnormal charging port of the charging cabinet according to the position characteristics of the abnormal stage, and marking the abnormal charging port; Step S55: performing a fault probability prediction on the abnormal charging port, thereby obtaining fault probability prediction data of the abnormal charging port.

7. The power-off management method for a charging cabinet according to claim 1, characterized in that: The specific steps of step S6 are: Step S61: Calculating the failure risk prediction of the failure probability prediction data of the abnormal charging port to obtain a failure risk prediction value of the charging port; Step S62: comparing the fault risk prediction value of the charging port with the preset charging port risk threshold, and generating a charging port fault warning signal when the fault risk prediction value of the charging port is greater than or equal to the preset charging port risk threshold; Step S63: performing single-point power-off processing on the abnormal charging port based on the charging port fault warning signal, and collecting charging port power-off response data; Step S64: Perform intelligent decision optimization on the charging port power outage response data and build an intelligent prediction power outage management model.

8. A power-off management device for a charging cabinet, characterized in that: The method for managing power failure of a charging cabinet according to claim 1 comprises: A data optimization module is used to collect real-time charging status information of the charging cabinet based on multiple sensors; perform time-series digital processing on the real-time charging status information, and perform filtering and noise reduction to obtain a noise-reduced and optimized charging status data sequence; A load calculation module is used to perform a charging port state matching analysis on the noise reduction optimized charging state data sequence one by one, and perform charging load calculation to obtain the load data of each charging port; The demand trend module is used to obtain the historical operation log of the charging cabinet; perform user demand analysis on the historical operation log of the charging cabinet, and conduct in-depth demand trend learning to generate the changing rules of user demand trends; The load prediction module is used to perform multiple rounds of load iteration simulation on the load data of each charging port according to the changing rules of user demand trends, and to perform short-term charging load prediction and construct a charging load prediction time series curve; A fault prediction module is used to locate abnormal charging ports based on a preset theoretical charging load curve and a charging load prediction time series curve, and to perform fault probability prediction, thereby obtaining fault probability prediction data of the abnormal charging port; The power outage management module is used to handle single-point power outages based on the fault probability prediction data of abnormal charging ports, perform intelligent decision-making optimization, and build an intelligent predictive power outage management model.

9. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the power off management method of the charging cabinet described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the power off management method of the charging cabinet described in any one of claims 1 to 7 are implemented.

Images