Charging pile real-time monitoring system based on cloud platform

Abstract

The invention relates to the technical field of machine type communication, in particular to a charging pile real-time monitoring system based on a cloud platform, which comprises an edge state inspection module, an edge safety judgment module, a pile group communication convergence module and a cloud comprehensive monitoring module. According to the method, key state data of insulation, grounding, contactors and the like are aggregated and calculated on the local side of the pile group, a local safety baseline is set for real-time comparison and judgment, all original parameters are replaced to be directly reported to the cloud, and generated safety state codes and necessary power utilization parameters are aggregated; transmission pressure and data redundancy of a communication network are remarkably reduced, operation cost is reduced, local rapid identification and response of security risks are realized, cloud processing delay is avoided, converged state messages are constructed into pile group feature vectors, the pile group feature vectors are submitted to a cloud for risk classification of a support vector machine model, and the risk classification efficiency is improved. Accurate evaluation of the overall load risk of the pile group is realized, and the safety of a power grid is ensured.

Description

Technical Field

[0001] This invention relates to the field of machine-type communication technology, and in particular to a real-time monitoring system for charging piles based on a cloud platform. Background Technology

[0002] The field of machine-type communication technology involves technologies that allow devices or terminals to automatically exchange data with communication networks. Core aspects of this field include data acquisition by terminal devices, access to wireless communication modules, data transmission protocols, and network architectures for managing and handling massive numbers of device connections. Among these, a traditional cloud-based real-time monitoring system for charging piles refers to an architecture that utilizes cloud servers to collect and process charging pile data. Traditional implementations typically involve deploying current and voltage sensors within each charging pile to collect operating parameters, and then using a built-in GPRS or 4G module to periodically report the collected status data and fault information to a remote cloud platform server for centralized storage via a cellular mobile network.

[0003] In existing technologies, charging pile monitoring systems rely on the cellular modules built into each terminal to periodically report the collected operating parameters and fault information to the cloud platform. When charging piles are deployed on a large scale, this operating mode generates massive amounts of redundant status data and instantaneous communication requests, which can easily cause regional wireless communication network congestion, leading to data upload delays or even data loss, and increasing communication costs. At the same time, all raw data is collected and centrally stored and processed on the cloud server, which places a heavy computational burden on the cloud platform, making it difficult to respond quickly to local faults such as insulation or grounding faults, and lacking the ability to comprehensively assess the risk of the entire charging pile group, thus affecting the stable operation of the power grid. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a real-time monitoring system for charging piles based on a cloud platform.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a real-time monitoring system for charging piles based on a cloud platform includes: The edge status inspection module aggregates and calculates the local status dataset based on the insulation status resistance value, grounding Boolean logic value, power consumption parameters, and contactor adhesion Boolean logic value obtained by the controller, and then transmits it to the edge safety discrimination module. The edge safety discrimination module compares the insulation state resistance value with the minimum insulation resistance threshold based on the local state dataset, checks the grounding Boolean logic value and the contactor adhesion Boolean logic value, determines whether the local state dataset triggers the local safety baseline, generates a safety status code, extracts the power consumption parameters, and transmits the safety status code and power consumption parameters to the pile group communication aggregation module. The pile group communication aggregation module collects the safety status code and the power consumption parameters, constructs a pile group status message consisting of the pile identifier, the safety status code, and the power consumption parameters, and transmits it to the cloud-based integrated monitoring module. The cloud-based integrated monitoring module, based on the pile group status message, combines the safety status code and the power consumption parameters in the pile group status message into a pile group feature vector, calls the support vector machine model to perform risk classification on the pile group feature vector, and calculates the pile group monitoring results representing the power grid load risk level.

[0006] As a further aspect of the present invention, the local status dataset includes electrical integrity features, grounding safety features, load operation features, and contactor reliability features; the safety status code includes a safety level identifier, a safety trigger flag, a safety judgment timestamp, and a fault category code; the pile group status message includes a unique pile identifier, energy consumption statistics, a communication time identifier, and a pile group operation category; and the pile group monitoring results include an operation safety level, energy consumption status assessment, fault trend prediction, and a system health index.

[0007] As a further aspect of the present invention, the specific function of the edge state inspection module is as follows: The electrical data acquisition submodule monitors the output port of the controller in real time, obtains the dynamically changing insulation state resistance value, and performs a preliminary comparison between the insulation state resistance value and the preset electrical safety reference value, marks the fluctuation frequency and amplitude of the insulation state resistance value, and establishes an insulation resistance time sequence snapshot. The grounding status analysis submodule obtains the grounding Boolean logic value fed back by the controller, verifies the stable state of the grounding Boolean logic value, analyzes whether the logic transition of the grounding Boolean logic value in the charging process start and stop phases conforms to the standard grounding protection protocol, and determines the instantaneous effectiveness of the grounding loop. The load parameter extraction submodule obtains the power consumption parameters from the power metering unit of the controller. The power consumption parameters include real-time current, real-time voltage, active power, and reactive power. It calculates the root mean square value and peak factor of the power consumption parameters in the current sampling period and evaluates the current load level and power quality. The contactor diagnostic submodule obtains the contactor adhesion Boolean logic value fed back by the controller, and combines it with the real-time current status in the power parameters to determine whether the contactor adhesion Boolean logic value reflects the on / off state of the physical contacts, and identifies potential adhesion risks or drive failures. The state aggregation encapsulation submodule gathers the insulation resistance timing snapshot, the instantaneous effectiveness of the grounding loop, the current load level and power quality, and the potential sticking risk or drive failure. Based on the data structure definitions of electrical integrity characteristics, grounding safety characteristics, load operation characteristics, and contactor reliability characteristics, it generates a local state dataset.

[0008] As a further aspect of the present invention, the specific function of the edge security discrimination module is as follows: The insulation safety verification submodule acquires the electrical integrity features included in the local state dataset, extracts the insulation state resistance value, and strictly compares the insulation state resistance value with the minimum insulation resistance threshold. If the insulation state resistance value is lower than the minimum insulation resistance threshold, it is determined to be an insulation failure. The grounding logic verification submodule acquires the grounding safety features included in the local status dataset, extracts the grounding Boolean logic value, and checks whether the grounding Boolean logic value is "true". If the grounding Boolean logic value is "false", it is determined to be a grounding fault. The contactor status judgment submodule obtains the contactor reliability characteristics included in the local status dataset, extracts the contactor adhesion Boolean logic value, checks whether the contactor adhesion Boolean logic value is "true", and if the contactor adhesion Boolean logic value is "true", it is determined that the contactor is adhesion. The local baseline triggering submodule integrates the determination results of the insulation failure, the grounding fault, and the contactor adhesion to determine whether the local status dataset triggers the local safety baseline. The local safety baseline defines a logical rule that triggers when any key electrical safety feature fails, and generates a local safety trigger signal. The status code generation submodule, based on the local safety trigger signal and according to the structural definition of safety level identifier, safety trigger flag, safety judgment timestamp, and fault category code, fills in the safety status information, generates a safety status code, and extracts the power consumption parameters from the load operation characteristics of the local status dataset.

[0009] As a further aspect of the present invention, the specific function of the pile group communication aggregation module is as follows: The local message receiving submodule establishes a secure communication link with the edge security discrimination module, receives the security status code in real time, synchronously acquires the power consumption parameters, and performs data verification on the security status code and the power consumption parameters to ensure that the integrity of the data is not compromised during transmission. The charging pile identifier loading submodule calls the locally stored charging pile identifier. The charging pile identifier is a globally unique device code used to uniquely identify an individual charging pile on the cloud platform, ensuring that the format of the charging pile identifier conforms to the cloud access specification. The status message construction submodule obtains the verified security status code, the power consumption parameters, and the loaded pile identifier. Based on the unique pile identifier, energy consumption statistics, communication time identifier, and predetermined data format of the pile group operation category, it maps the security status code to the pile group operation category and the energy consumption statistics. Combines the current system timestamp as the communication time identifier, and constructs the pile group status message. The message queue management submodule stores the completed pile group status messages into a local message queue to be uploaded, and sorts the pile group status messages in the message queue to be uploaded by priority, giving priority to messages including high-risk security status codes. The data encryption upload submodule extracts the pile group status message from the message queue to be uploaded, encrypts the pile group status message using an asymmetric encryption algorithm, and transmits the encrypted pile group status message to the cloud integrated monitoring module through the secure socket layer channel.

[0010] As a further aspect of the present invention, the process by which the cloud-based integrated monitoring module generates the feature vector of the pile group includes: The message parsing and extraction submodule receives and decrypts the pile group status message, parses the data structure of the pile group status message, extracts the unique identifier of the pile, energy consumption statistics, communication time identifier and the pile group operation category, and backtracks to extract the original safety status code and power consumption parameters from the pile group status message. The feature quantization encoding submodule obtains the safety level identifier and fault category code from the safety status code, performs sequential encoding on the safety level identifier, performs one-thermal encoding on the fault category code, performs Z-score normalization on the continuous values ​​in the power consumption parameters, and generates standard energy features and quantized safety features. The time-series feature construction submodule, based on the communication time identifier, aggregates the historical status messages of the pile group within the past preset time window of the unique identifier of the pile body, and counts the slope of the change trend of the power consumption parameters and the trigger frequency of the safety status code to construct time-series dynamic features; The feature vector normalization submodule integrates the standard energy feature, the quantized security feature, and the temporal dynamic feature, and concatenates and aligns all features according to the fixed dimensions required by the input layer of the support vector machine model, filling missing values ​​with zero values ​​to generate the pile group feature vector.

[0011] As a further aspect of the present invention, the process by which the cloud-based integrated monitoring module calculates the monitoring results of the pile group includes: The model risk classification submodule inputs the generated pile group feature vector into the pre-trained support vector machine model. The support vector machine model uses the radial basis kernel function to calculate the distance from the pile group feature vector to the decision hyperplane and outputs the preliminary classification label of the power grid load risk level, which includes three levels: high risk, medium risk, and low risk. The energy consumption status assessment submodule calculates the deviation between the current charging pile's instantaneous energy consumption and the historical average energy consumption based on the standard energy characteristics in the pile group feature vector. Combined with the power grid load risk level, it assesses the instantaneous contribution or impact of the current charging pile on the power grid stability and generates the energy consumption status assessment. The fault trend prediction submodule uses a hidden Markov model to analyze the evolution of the temporal dynamic features in the pile group feature vector, including the transition probability of the fault category encoding, to predict the probability of a critical fault occurring in the next cycle and generate the fault trend prediction. The system health calculation submodule integrates the preliminary classification label of the power grid load risk level, the energy consumption status assessment, and the fault trend prediction, and calculates a comprehensive score through a weighted fusion algorithm. The comprehensive score is the system health index. The power grid load risk level, the energy consumption status assessment, the fault trend prediction, and the system health index are combined to generate the pile group monitoring results.

[0012] As a further aspect of the present invention, when establishing the insulation resistance time sequence snapshot, the electrical data acquisition submodule sets a fixed-length sliding time window, continuously acquires the insulation state resistance value sequence within the sliding time window, analyzes the spectral characteristics of the insulation state resistance value sequence using fast Fourier transform, identifies target frequency fluctuations caused by moisture or cable aging, and calculates the mean, variance, and minimum value of the insulation state resistance values ​​within the sliding time window. The spectral characteristics, the mean, the variance, and the minimum value are then encapsulated as the electrical integrity feature. When evaluating the current load level and power quality, the load parameter extraction submodule not only acquires the real-time current and real-time voltage from the power consumption parameters, but also synchronously monitors the grid frequency offset, calculates the phase difference between the real-time current and the real-time voltage to obtain the real-time power factor, iteratively analyzes the harmonic components of the real-time current, calculates the total harmonic distortion value, and aggregates the real-time power factor, the total harmonic distortion value, the grid frequency offset, the active power, and the reactive power together to form the load operating characteristics.

[0013] As a further aspect of the present invention, when the local baseline triggering submodule determines whether the local state dataset triggers the local safety baseline, the local safety baseline is defined as a multi-dimensional logic matrix. The multi-dimensional logic matrix includes not only the triggering condition that the insulation state resistance value is lower than the minimum insulation resistance threshold, but also the composite triggering condition that the number of consecutive losses of the grounding Boolean logic value exceeds a preset threshold, and the logic conflict between the contactor adhesion Boolean logic value and the current feedback value in the power consumption parameters. The satisfaction of any one of the conditions generates the local safety trigger signal. When generating the security status code, the status code generation submodule acquires the local security trigger signal. If the local security trigger signal is true, it sets the security level identifier to "high risk", sets the security trigger flag to "local trigger", and fills in the corresponding fault category code according to the specific conditions that trigger the local security baseline. At the same time, it records the current system time as the security determination timestamp.

[0014] As a further embodiment of the present invention, after outputting the preliminary classification label of the power grid load risk level, the model risk classification submodule also needs to calculate the Euclidean distance from the pile group feature vector to the decision hyperplane, and convert the Euclidean distance into a classification confidence score. The classification confidence score and the preliminary classification label together constitute the operation safety level. When calculating the system health index, the system health calculation submodule uses the weighted fusion algorithm, which is specifically as follows: ; in, Represents the system's health index. It is a risk score quantified based on the stated operational safety level. It is an energy consumption score quantified based on the aforementioned energy consumption status assessment. It is a trend score obtained by quantifying the aforementioned fault trend prediction. , , These represent the fusion weights corresponding to the risk score, the energy consumption score, and the trend score, respectively, and the fusion weights are dynamically adjusted based on historical data.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by aggregating and calculating key status data such as insulation, grounding, and contactors on the local side of the pile group, and setting a local safety baseline for real-time comparison and judgment, this method replaces the direct reporting of all original parameters to the cloud. Only the generated safety status codes and necessary power consumption parameters are aggregated. This approach significantly reduces the transmission pressure and data redundancy of the communication network, reduces operating costs, and, more importantly, enables rapid local identification and response to safety risks, avoiding delays in cloud processing. Finally, the aggregated status messages are constructed into a pile group feature vector and submitted to the cloud for risk classification using a support vector machine model. This achieves an accurate assessment of the overall load risk of the pile group and ensures power grid safety. Attached Figure Description

[0016] Figure 1 This is the overall architecture diagram of the real-time monitoring system for charging piles of the present invention; Figure 2 This is a flowchart of the edge state inspection module of the present invention; Figure 3 This is a flowchart of the edge security discrimination module of the present invention; Figure 4 This is a flowchart of the pile group communication aggregation module of the present invention; Figure 5 This is a flowchart of the cloud-based integrated monitoring module of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the software-based technical solution is described in detail below with reference to system architecture diagrams and embodiments. It should be understood that the specific embodiments described herein are only for explaining the technical solutions of this invention and do not constitute a limitation on the scope of protection.

[0018] In the description of this invention, the system architecture relationships or data processing flows indicated by terms such as "layer," "module," "interface," "data flow," "client," and "server" are all defined based on the architecture diagram or flowchart corresponding to the embodiments. This way of describing is only used to clearly illustrate the logical relationships between the elements in the technical solution, and not to limit the physical deployment form. The term "multiple" includes two or more technical units, including but not limited to multiple data nodes, processing threads, service instances, or functional components and other scalable elements. The specific number is determined according to the actual business scenario and needs to be specifically specified.

[0019] Please see Figure 1 and Figure 2 This invention provides a technical solution: a real-time monitoring system for charging piles based on a cloud platform, comprising: The edge status inspection module aggregates and calculates the local status dataset based on the insulation status resistance value, grounding Boolean logic value, power consumption parameters, and contactor adhesion Boolean logic value obtained by the controller, and then transmits it to the edge safety discrimination module. The local state dataset includes electrical integrity characteristics, grounding safety characteristics, load operation characteristics, and contactor reliability characteristics; The specific functions of the edge status inspection module are as follows: The electrical data acquisition submodule monitors the output port of the controller in real time, obtains the dynamically changing insulation resistance value, and makes a preliminary comparison between the insulation resistance value and the preset electrical safety reference value, marks the fluctuation frequency and amplitude of the insulation resistance value, and establishes a time-series snapshot of the insulation resistance. When establishing an insulation resistance time sequence snapshot, the electrical data acquisition submodule sets a fixed-length sliding time window, continuously acquires the insulation state resistance value sequence within the sliding time window, uses fast Fourier transform to analyze the spectral characteristics of the insulation state resistance value sequence, identifies target frequency fluctuations caused by moisture or cable aging, and calculates the mean, variance, and minimum value of the insulation state resistance values ​​within the sliding time window. The spectral characteristics, mean, variance, and minimum value are then encapsulated as electrical integrity features. The grounding status analysis submodule obtains the grounding Boolean logic value fed back by the controller, verifies the stable state of the grounding Boolean logic value, analyzes whether the logic transition of the grounding Boolean logic value during the start and stop phases of the charging process conforms to the standard grounding protection protocol, and determines the instantaneous effectiveness of the grounding loop. The load parameter extraction submodule obtains power consumption parameters from the power metering unit of the controller. The power consumption parameters include real-time current, real-time voltage, active power, and reactive power. It calculates the root mean square value and peak factor of the power consumption parameters in the current sampling period and evaluates the current load level and power quality. When assessing the current load level and power quality, the load parameter extraction submodule not only obtains the real-time current and real-time voltage from the power consumption parameters, but also needs to simultaneously monitor the offset of the grid frequency, calculate the phase difference between the real-time current and real-time voltage to obtain the real-time power factor, iteratively analyze the harmonic components of the real-time current, calculate the total harmonic distortion value, and aggregate the real-time power factor, total harmonic distortion value, grid frequency offset, active power, and reactive power together into the load operating characteristics. The contactor diagnostic submodule obtains the contactor sticking Boolean logic value fed back by the controller, and combines it with the real-time current status in the power parameters to determine whether the contactor sticking Boolean logic value reflects the on / off state of the physical contacts, and identifies potential sticking risks or drive failures. The state aggregation encapsulation submodule gathers insulation resistance timing snapshots, instantaneous effectiveness of grounding loops, current load levels and power quality, as well as potential sticking risks or drive failures. Based on the data structure definitions of electrical integrity characteristics, grounding safety characteristics, load operation characteristics, and contactor reliability characteristics, it generates a local state dataset.

[0020] After the edge status inspection module is activated, the electrical data acquisition submodule first continuously monitors the controller output port. The sampling frequency for this monitoring action is set to 1000Hz. In one embodiment, the insulation state resistance value output by the controller is a dynamically changing analog voltage signal, which is converted into a digital value by the electrical data acquisition submodule through a 16-bit analog-to-digital converter (ADC). A sliding time window is set with a fixed length of 1 second. Within this 1-second sliding time window, 1000 insulation state resistance values ​​are continuously acquired, forming a numerical sequence. For example, the acquired sequence is... .

[0021] Then, the numerical sequence of these 1000 points was processed using the Fast Fourier Transform (FFT). Specifically, the time-domain sequence was... The signal is then converted to the frequency domain. By analyzing the spectral characteristics, fluctuations at specific frequencies are identified. For example, experimental verification shows that cable aging often leads to abnormal low-frequency noise peaks in the 10Hz-20Hz frequency band, while humid environments may introduce specific high-frequency interference in the 30Hz-50Hz frequency band. In this embodiment, a peak with an amplitude exceeding the preset interference baseline (0.05MΩ) was found at 15Hz, thereby identifying the target frequency fluctuation.

[0022] Simultaneously, statistical parameters were calculated for 1000 absolute state resistance values ​​within this sliding time window. The mean was calculated as follows: One example is a window with a mean of 0.8 MΩ. Calculate the variance: One example has a variance of 0.04. Determine the minimum value: One example shows a minimum value of 0.75 MΩ. Finally, the identified spectral characteristics (15 Hz peak), mean (0.8 MΩ), and variance (0.04) are calculated. The minimum (0.75MΩ) package is defined as an electrical integrity feature. This electrical integrity feature is marked as... .

[0023] Next, the grounding status analysis submodule obtains the grounding Boolean logic value from the controller. During the charging process initiation phase, after the controller issues a grounding command, the grounding status analysis submodule continuously reads the grounding Boolean logic value 10 times within the following 500 milliseconds. If all 10 reads are "true," the stable state of the grounding Boolean logic value is verified as valid. Subsequently, during the charging process stop phase, the controller issues a disconnect command, and the submodule again continuously reads the value 10 times within the following 500 milliseconds. If all 10 reads are "false," the logic transition is determined to conform to the standard grounding protection protocol. In this embodiment, the reading is "true" during the initiation phase and "false" during the stop phase, thus determining the instantaneous validity of the grounding loop as "valid." This result is encapsulated as a grounding safety feature. .

[0024] Subsequently, the load parameter extraction submodule obtains the power consumption parameters from the controller's power metering unit. The obtained real-time current is 16.1A, real-time voltage is 221V, active power is 3.5kW, and reactive power is 0.3kVar. The grid frequency offset of +0.02Hz is simultaneously monitored. The phase difference between the real-time current and real-time voltage is calculated, and the real-time power factor is obtained using the inverse cosine function. In this example, the apparent power... kVA, power factor Simultaneously, the harmonic components of the real-time current are iteratively analyzed, and the total harmonic distortion (THD) value is calculated. Specifically, this calculation involves measuring the harmonic current at each harmonic order. (h=2,3,…,n), calculate In this example, the measured THD value was 4.5%. The real-time power factor (0.997), total harmonic distortion (4.5%), grid frequency offset (+0.02Hz), active power (3.5kW), and reactive power (0.3kVar) were aggregated to form the load operating characteristic. This characteristic was tagged as... .

[0025] The contactor diagnostic submodule obtains the contactor sticking Boolean logic value fed back from the controller; the current value is "false" (indicating no sticking). Simultaneously, it considers the real-time current status (16.1A, non-zero) in the load operating characteristics. The judgment logic is as follows: when the contactor sticking Boolean logic value is "false" and the real-time current is non-zero (greater than 1A), or when the contactor sticking Boolean logic value is "true" and the real-time current is zero (less than 0.1A), both are considered consistent. In this embodiment, the value is "false" and the current is 16.1A, indicating a consistent status. The potential risk of sticking or drive failure is determined to be "no risk." This result is encapsulated as a contactor reliability feature. .

[0026] Finally, the state aggregation encapsulation submodule gathers all the above results and generates a local state dataset according to a preset data structure definition. The specific content of this dataset is as follows: .

[0027] Please see Figure 1 and Figure 3 The edge safety discrimination module compares the insulation resistance value with the minimum insulation resistance threshold based on the local status dataset, checks the grounding Boolean logic value and the contactor adhesion Boolean logic value, determines whether the local status dataset triggers the local safety baseline, generates a safety status code, extracts the power consumption parameters, and transmits the safety status code and power consumption parameters to the pile group communication aggregation module. The security status code includes a security level identifier, a security trigger flag, a security determination timestamp, and a fault category code; The specific functions of the edge security discrimination module are as follows: The insulation safety verification submodule acquires the electrical integrity features included in the local status dataset, extracts the insulation status resistance value, and strictly compares the insulation status resistance value with the minimum insulation resistance threshold. If the insulation status resistance value is lower than the minimum insulation resistance threshold, it is determined to be an insulation failure. The grounding logic check submodule obtains the grounding safety features included in the local status dataset, extracts the grounding Boolean logic value, and checks whether the grounding Boolean logic value is "true". If the grounding Boolean logic value is "false", it is determined to be a grounding fault. The contactor status assessment submodule obtains the contactor reliability characteristics included in the local status dataset, extracts the contactor adhesion Boolean logic value, and checks whether the contactor adhesion Boolean logic value is "true". If the contactor adhesion Boolean logic value is "true", it is determined that the contactor is adhesion. The local baseline triggering submodule integrates the judgment results of insulation failure, grounding fault, and contactor adhesion to determine whether the local status dataset triggers the local safety baseline. The local safety baseline defines the logical rule that triggers when any key electrical safety feature fails, and generates a local safety trigger signal. When the local baseline triggering submodule determines whether the local status dataset triggers the local safety baseline, the local safety baseline is defined as a multi-dimensional logic matrix. The multi-dimensional logic matrix includes not only the triggering condition that the insulation state resistance value is lower than the minimum insulation resistance threshold, but also the triggering condition that the number of consecutive loss of ground Boolean logic value exceeds the preset threshold, and the composite triggering condition that the contactor adhesion Boolean logic value conflicts with the current feedback value in the power parameters. The satisfaction of any one of the conditions will generate a local safety trigger signal. The status code generation submodule, based on the local safety trigger signal and according to the structural definition of safety level identifier, safety trigger flag, safety judgment timestamp, and fault category code, fills in the safety status information, generates a safety status code, and extracts power consumption parameters from the load operation characteristics of the local status dataset. When generating a security status code, the status code generation submodule obtains the local security trigger signal. If the local security trigger signal is true, it sets the security level identifier to "high risk", sets the security trigger flag to "local trigger", and fills in the corresponding fault category code according to the specific conditions for triggering the local security baseline. At the same time, it records the current system time as a security judgment timestamp.

[0028] After receiving the local state dataset transmitted by the edge state inspection module, the edge security discrimination module starts its specific function implementation.

[0029] First, the insulation safety verification submodule starts. This submodule obtains electrical integrity features from the local state dataset. The insulation resistance values ​​are extracted, specifically the average (0.8 MΩ) and the minimum (0.75 MΩ). These two values ​​are then rigorously compared with the minimum insulation resistance threshold.

[0030] Minimum insulation resistance threshold ( The settings were determined based on destructive testing of 500 charging piles with varying degrees of aging. The testing process included 1000 hours of operation in a temperature cycle (-20°C to +60°C) and an 85% humidity environment, simulating different degrees of cable insulation damage. Experimental data (as shown in Table 1) indicate that when the insulation resistance is below 1.0 MΩ, the probability of the leakage current device (RCD) malfunctioning or failing increases to over 5.2%.

[0031] Table 1 Experimental Data on Insulation Resistance and Fault Probability ; As shown in Table 1, to set a safety margin, the minimum insulation resistance threshold is set. The resistance is set to 1.0 MΩ. In this embodiment, the extracted insulation resistance value (average 0.8 MΩ) is lower than... (1.0MΩ). Therefore, the insulation safety verification submodule determined that the insulation had failed.

[0032] Next, the grounding logic verification submodule obtains the grounding safety characteristics. Extract the grounding Boolean logic value, which is "valid" (equivalent to the Boolean value "true"). Check if this value is "true". In this embodiment, the value is "true", so it is not determined to be a grounding fault.

[0033] The contactor condition assessment submodule obtains contactor reliability characteristics. Extract the Boolean logic value for contactor adhesion (equivalent to "no risk", Boolean value is "false"). Check if this value is "true". In this embodiment, the value is "false", so it is not determined that the contactor is stuck.

[0034] Subsequently, the local baseline triggering submodule is activated. The local safety baseline upon which this submodule is based is defined as a multi-dimensional logic matrix. The triggering conditions of this matrix include not only the insulation state resistance value (…). ) below the minimum insulation resistance threshold ( ), also includes grounded Boolean logic values ​​( Number of consecutive losses ( Exceeding the preset threshold ( ), and contactor sticking Boolean logic values ​​( ) and the current feedback value in the power consumption parameters ( () A compound triggering condition for logical conflict.

[0035] The specific rules for the multidimensional logic matrix are as follows: 1. Triggering condition 1 (insulation): 2. Triggering condition 2 (grounding): . Set to 3 times, this value is based on the retry mechanism of the communication protocol. If the grounding status cannot be obtained three consecutive times (e.g., within 300 milliseconds), a fault is considered. 3. Trigger condition 3 (contactor): ( ==“True” AND OR ( == "Fake" AND AND charging status == "stopped").

[0036] In this embodiment, the judgment result of the comprehensive insulation safety verification submodule (insulation failure, 0.8MΩ < 1.0MΩ) satisfies trigger condition 1. Therefore, the local baseline trigger submodule generates a local safety trigger signal with a value of "true".

[0037] Finally, the status code generation submodule starts. This submodule receives a "true" local safety trigger signal. Based on preset rules, it sets the safety level identifier to "high risk" and the safety trigger flag to "local trigger." According to the specific conditions for triggering the local safety baseline (trigger condition 1), it fills in the corresponding fault category code, such as "0x01A," representing "insulation resistance too low." Simultaneously, it records the current system time (e.g., 2025-11-07 10:50:01.100) as a safety determination timestamp. A complete safety status code is then generated. And load running features from the local state dataset. Extracting power consumption parameters .

[0038] Please see Figure 1 and Figure 4 The pile group communication aggregation module collects safety status codes and power consumption parameters, constructs a pile group status message consisting of pile identifiers, safety status codes, and power consumption parameters, and transmits it to the cloud-based integrated monitoring module. The status message of the pile group includes the unique identifier of the pile, energy consumption statistics, communication time identifier, and pile group operation category; The specific functions of the pile group communication aggregation module are as follows: The local message receiving submodule establishes a secure communication link with the edge security discrimination module, receives security status codes in real time, synchronously acquires power consumption parameters, and performs data verification on the security status codes and power consumption parameters to ensure that the integrity of the data is not compromised during transmission. The charging pile identifier loading submodule calls the locally stored charging pile identifier. The charging pile identifier is a globally unique device code used to uniquely identify individual charging piles on the cloud platform, ensuring that the format of the charging pile identifier conforms to the cloud access specifications. The status message construction submodule obtains the verified security status code, power consumption parameters, and loaded pile identifier. Based on the unique pile identifier, energy consumption statistics, communication time identifier, and predetermined data format of pile group operation category, it maps the security status code to the pile group operation category and energy consumption statistics. Combined with the current system timestamp as the communication time identifier, it constructs the pile group status message. The message queue management submodule stores the completed stub group status messages into the local message queue to be uploaded, sorts the stub group status messages in the message queue to be uploaded by priority, and processes messages including high-risk security status codes first. The data encryption upload submodule extracts the pile group status messages from the message queue to be uploaded, encrypts the pile group status messages using an asymmetric encryption algorithm, and transmits the encrypted pile group status messages to the cloud-based integrated monitoring module through the secure socket layer channel.

[0039] After receiving the security status code and power consumption parameters generated by the edge security discrimination module, the pile group communication aggregation module begins to execute its specific functions.

[0040] The local message receiving submodule receives the security status code via the internal bus. and power consumption parameters Upon receipt, the data is immediately checked using a checksum. For example, the edge security discrimination module calculates the CRC32 checksum of the data as 0xAB12CD34 before sending it. The local message receiving submodule performs the same CRC32 calculation on the received data, obtaining the same result of 0xAB12CD34. Since the two are consistent, the data integrity is verified.

[0041] Simultaneously, the pile identifier loading submodule retrieves the pile identifier from local non-volatile memory. This identifier is a globally unique device code, such as "ZC-001". This code "ZC-001" conforms to cloud access specifications (e.g., the format is "region-serial number").

[0042] The status message construction submodule obtains the verified security status code. Power consumption parameters The system also includes the pile identifier "ZC-001". Based on a predetermined data format (JSON), a pile group status message is constructed. This process involves mapping safety status codes to pile group operating categories and energy consumption statistics. For example, the safety level identifier "High Risk" and fault category code "0x01A" are mapped to pile group operating category "03" (representing "Severe Fault"). Active power (3.5kW) and reactive power (0.3kVar) from the power consumption parameters are directly encapsulated as energy consumption statistics. The current system timestamp (e.g., 2025-11-07 10:50:01.200) is used as the communication time identifier. The constructed pile group status message (JSON text) is as follows: {"pileId":"ZC-001","timestamp":"2025-11-07T10:50:01.200Z","runCategory":"03","energyInfo":{"activePower":3.5,"reactivePower":0.3,"voltage":221,"current":16.1,"thd":4.5},"securityCode":{"level":"High Risk","trigger":"Local Trigger","faultCode":"0x01A"}}; The message queue management submodule stores the completed stub group status messages into a local message queue awaiting upload. This queue is a first-in, first-out (FIFO) queue, but it has a priority handling mechanism. The queue manager detects the "level":"high risk" field in the message and sets its priority to 1 (highest priority). If there are already 5 messages with a priority of 3 (normal reporting) in the queue, this new message will be inserted at the head of the queue.

[0043] The data encryption upload submodule extracts the high-priority stub status message from the message queue to be uploaded. It negotiates a symmetric key (e.g., AES-256) with the cloud using an asymmetric encryption algorithm (e.g., RSA-2048). Specifically, a temporary AES-256 session key is encrypted using the cloud's public key; the cloud then decrypts this session key using its private key. Subsequently, the AES-256 session key is used to encrypt the JSON text of the stub status message. Finally, the encrypted data packet is transmitted to the cloud-based integrated monitoring module through an established Secure Sockets Layer (TLS 1.3) channel.

[0044] Please see Figure 1 and Figure 5 The cloud-based integrated monitoring module, based on the pile group status messages, combines the safety status code and power consumption parameters in the pile group status messages into a pile group feature vector, calls the support vector machine model to perform risk classification on the pile group feature vector, and calculates the pile group monitoring results representing the power grid load risk level. The monitoring results of the pile group include the operational safety level, energy consumption status assessment, fault trend prediction, and system health index; The cloud-based integrated monitoring module includes the following steps in generating the feature vector of the pile group: The message parsing and extraction submodule receives and decrypts the pile group status message, parses the data structure of the pile group status message, extracts the unique identifier of the pile, energy consumption statistics, communication time identifier and pile group operation category, and backtracks to extract the original safety status code and power consumption parameters from the pile group status message. The feature quantization coding submodule obtains the safety level identifier and fault category code from the safety status code, performs sequential coding on the safety level identifier, performs one-thermal coding on the fault category code, performs Z-score standardization on the continuous values ​​in the power parameters, and generates standard energy characteristics and quantified safety characteristics. The time-series feature construction submodule, based on communication time identifiers, aggregates historical pile group status messages with unique identifiers of pile bodies within the past preset time window, and statistically analyzes the slope of the change trend of power consumption parameters and the trigger frequency of safety status codes to construct time-series dynamic features; The feature vector normalization submodule integrates standard energy features, quantized security features, and temporal dynamic features. It concatenates and aligns all features according to the fixed dimensions required by the input layer of the support vector machine model, fills missing values ​​with zero values, and generates pile group feature vectors. The cloud-based integrated monitoring module's process for calculating the pile group monitoring results includes: The model risk classification submodule inputs the generated pile group feature vector into the pre-trained support vector machine model. The support vector machine model uses the radial basis kernel function to calculate the distance from the pile group feature vector to the decision hyperplane and outputs the preliminary classification label of the power grid load risk level. The power grid load risk level includes three levels: high risk, medium risk, and low risk. After outputting the preliminary classification label of the power grid load risk level, the model risk classification submodule also needs to calculate the Euclidean distance from the pile group feature vector to the decision hyperplane, and convert the Euclidean distance into a classification confidence score. The classification confidence score and the preliminary classification label together constitute the operation safety level. The energy consumption status assessment submodule calculates the deviation between the current charging pile's instantaneous energy consumption and the historical average energy consumption based on the standard energy characteristics in the charging pile group feature vector. Combined with the grid load risk level, it assesses the current charging pile's instantaneous contribution or impact on grid stability and generates an energy consumption status assessment. The fault trend prediction submodule uses a hidden Markov model to analyze the evolution of time-series dynamic features in the pile group feature vector, including the transition probability of fault category encoding, predicts the probability of a critical fault occurring in the next cycle, and generates a fault trend prediction. The system health calculation submodule integrates the preliminary classification labels of the grid load risk level, energy consumption status assessment, and fault trend prediction. It calculates a comprehensive score through a weighted fusion algorithm, which is the system health index. The grid load risk level, energy consumption status assessment, fault trend prediction, and system health index are combined to generate the pile group monitoring results. The system health calculation submodule uses a weighted fusion algorithm when calculating the system health index. The weighted fusion algorithm is as follows: ; in, Represents the system's health index. It is a risk score quantified based on the operational safety level. It is an energy consumption score quantified based on energy status assessment. It is a trend score obtained by quantifying fault trend prediction. , , These represent the fusion weights corresponding to the risk score, energy consumption score, and trend score, respectively, and the fusion weights are dynamically adjusted based on historical data.

[0045] After receiving the encrypted pile group status message, the cloud-based integrated monitoring module decrypts it and initiates the process of generating the pile group feature vector.

[0046] The message parsing and extraction submodule first decrypts the data to obtain a JSON-formatted pile group status message. Parsing this data structure extracts the unique identifier for each pile, "ZC-001", the communication time identifier, "2025-11-07 10:50:01.200Z", the pile group operation category, "03", and energy consumption statistics. Simultaneously, it backtracks to extract the original safety status code. With power consumption parameters .

[0047] The feature quantization encoding submodule obtains the security level identifier "High Risk" and the fault category code "0x01A" from the security status code. The security level identifier is sequentially encoded, with the encoding rule set as follows: "Safe" = 0, "Low Risk" = 1, "Medium Risk" = 2, "High Risk" = 3. Therefore, "High Risk" is encoded as 3. The fault category code "0x01A" is then one-hot encoded. Assuming the system defines 10 fault categories, and code "0x01A" is the second category, its one-hot encoding is a vector. These two factors constitute the quantitative safety characteristics. Simultaneously, continuous values ​​in the power consumption parameters are standardized using Z-score. The standardization is based on the average value in the historical operation database of this pile group (…). ) and standard deviation ( ).For example: Active power : kW, kW. .

[0048] Total Harmonic Distortion : , . Other electrical parameters are processed in the same way to generate standard energy characteristics.

[0049] The time-series feature construction submodule aggregates historical pile group status messages for the pile within the past hour (preset time window) from the time-series database based on the communication time identifier and the unique identifier of the pile "ZC-001". As shown in Table 2.

[0050] Table 2. Summary of Historical Status of ZC-001 Pile ; Linear regression analysis was performed on the data in Table 2 to statistically analyze the slope of the changing trend of power consumption parameters (active power), which was calculated to be 0.2 kW / 5 minutes. The trigger frequency of safety status codes was also analyzed, with "high-risk" triggers occurring once per hour. These constitute the time-series dynamic characteristics.

[0051] The feature vector normalization submodule will normalize the standard energy features ( ), quantified security features ( ) and time-series dynamic characteristics ( The features are then concatenated. The input layer of the Support Vector Machine (SVM) model requires a fixed dimension of 128. After concatenating all features, if the dimension is less than 128, missing values ​​are filled with zeros to generate the final pile group feature vector. .

[0052] Subsequently, the cloud-based integrated monitoring module initiates the calculation process for the pile group monitoring results.

[0053] The model risk classification submodule will The input is fed into a pre-trained Support Vector Machine (SVM) model. This model utilizes the Radial Basis Function (RBF) to compute... The distance to the decision hyperplane. The model outputs a preliminary classification label. In this embodiment, due to the high weights of the "high-risk" code 3 and the Z-score 1.5 of THD in the quantized security features, the model outputs "high risk". Simultaneously, the distance to the decision hyperplane is calculated. The Euclidean distance to the decision hyperplane for the "high-risk" category is calculated, and this distance (e.g., 0.15) is converted into a classification confidence score (e.g., 0.95). Together, they constitute the operational safety level: "high risk (95% confidence)".

[0054] The energy consumption status assessment submodule is based on Standard energy characteristics in The deviation between the current instantaneous energy consumption (3.5kW) and the historical average energy consumption (3.0kW) is calculated to be +16.7%. Considering the grid load risk level is "high risk," the current charging piles are assessed as posing a "high impact" to grid stability. An energy consumption status assessment is generated: "High Impact."

[0055] The fault trend prediction submodule utilizes Hidden Markov Models (HMMs) for analysis. The evolution of time-series dynamic characteristics. The Hidden Markov Model (HMM) defines three states: "Safe," "Insulation Warning," and "Insulation Failure" (High Risk). By analyzing historical data, the model calculates the probability of transitioning from the "Safe" state to the "Insulation Failure" state (e.g., ...). Based on the current "high-risk" state, the probability of maintaining the "insulation fault" state for the next period (e.g., 1 hour) is predicted to be 0.8. Fault trend prediction: "80% probability of maintaining the fault".

[0056] The system health calculation submodule synthesizes the above results and calculates the system health index using a weighted fusion algorithm. The algorithm formula is as follows: ; in, This represents the system's health index, with a value range of 0-100, where 100 represents the optimal health status. The fusion weight representing the risk score is used to incorporate the operational safety level score into the calculation of the system health index. This represents a risk score quantified based on the operational safety level, with a value range of 0-100. The fusion weight representing the energy consumption score is used to incorporate the energy consumption status assessment score into the calculation of the system health index. This represents the energy consumption score quantified based on the energy consumption status assessment, with a value range of 0-100. The fusion weight representing the trend score is used to incorporate the fault trend prediction score into the calculation of the system health index. This represents the trend score obtained by quantifying the fault trend prediction, with a value range of 0-100.

[0057] Formula parameter description: 1. Fusion weight ( , , The weights are determined based on regression analysis of 10,000 historical failure samples to ascertain the contribution of each factor to the overall health of the system. Weights are determined using AHP (Analytic Hierarchy Process) combined with expert scoring. For example, safety risk is considered the primary factor.

[0058] Experimental verification process: Data from 1000 charging piles that have experienced malfunctions were selected, a logistic regression model was constructed, and analysis was performed. , , The impact on whether it ultimately leads to a shutdown (Y=1 or 0). The model coefficients indicate... Its contribution is approximately 50%. Approximately 30%, Approximately 20%. Example settings: , , .and .

[0059] 2. Quantitative scoring ( , , ): (Risk Score): Quantified based on the operational safety level of "High Risk (95% confidence level)". Quantification rules: "Safe" = 100, "Low Risk" = 70, "Medium Risk" = 40, "High Risk" = 10. Therefore, .

[0060] (Energy Consumption Score): Quantified based on the energy consumption status assessment of "High Impact". Quantification rules: "Optimization Contribution" = 100, "Normal" = 80, "Slight Impact" = 60, "High Impact" = 20. Therefore, .

[0061] (Trend Score): Quantified based on the failure trend prediction of "80% probability of maintenance failure". Quantification rules: ,in This is the predicted failure probability. Therefore, .

[0062] Example calculation: Substitute the above parameter values ​​into the formula: .

[0063] The result It is a comprehensive score. The criteria for judging the health index are set as follows: [80-100] is "healthy", [50-79] is "sub-healthy", and [0-49] is "dangerous". The current calculation result 15 is in the "dangerous" range. This result indicates that the current system health index of charging pile "ZC-001" is in the "dangerous" range. The system health calculation submodule combines the grid load risk level (high risk), energy consumption status assessment (high impact), fault trend prediction (80% probability of maintaining fault), and system health index (15) to generate the final pile group monitoring result.

[0064] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the scope of protection defined by the claims of the present invention.

Claims

1. A real-time monitoring system for charging piles based on a cloud platform, characterized in that, The system includes: The edge status inspection module aggregates and calculates the local status dataset based on the insulation status resistance value, grounding Boolean logic value, power consumption parameters, and contactor adhesion Boolean logic value obtained by the controller, and then transmits it to the edge safety discrimination module. The edge safety discrimination module compares the insulation state resistance value with the minimum insulation resistance threshold based on the local state dataset, checks the grounding Boolean logic value and the contactor adhesion Boolean logic value, determines whether the local state dataset triggers the local safety baseline, generates a safety status code, extracts the power consumption parameters, and transmits the safety status code and power consumption parameters to the pile group communication aggregation module. The pile group communication aggregation module collects the safety status code and the power consumption parameters, constructs a pile group status message consisting of the pile identifier, the safety status code, and the power consumption parameters, and transmits it to the cloud-based integrated monitoring module. The cloud-based integrated monitoring module, based on the pile group status message, combines the safety status code and the power consumption parameters in the pile group status message into a pile group feature vector, calls the support vector machine model to perform risk classification on the pile group feature vector, and calculates the pile group monitoring results representing the power grid load risk level.

2. The real-time monitoring system for charging piles based on a cloud platform according to claim 1, characterized in that, The local status dataset includes electrical integrity features, grounding safety features, load operation features, and contactor reliability features. The safety status code includes a safety level identifier, a safety trigger flag, a safety judgment timestamp, and a fault category code. The pile group status message includes a unique pile identifier, energy consumption statistics, a communication time identifier, and a pile group operation category. The pile group monitoring results include an operation safety level, energy consumption status assessment, fault trend prediction, and a system health index.

3. The real-time monitoring system for charging piles based on a cloud platform according to claim 1, characterized in that, The specific functions of the edge state inspection module are as follows: The electrical data acquisition submodule monitors the output port of the controller in real time, obtains the dynamically changing insulation state resistance value, and performs a preliminary comparison between the insulation state resistance value and the preset electrical safety reference value, marks the fluctuation frequency and amplitude of the insulation state resistance value, and establishes an insulation resistance time sequence snapshot. The grounding status analysis submodule obtains the grounding Boolean logic value fed back by the controller, verifies the stable state of the grounding Boolean logic value, analyzes whether the logic transition of the grounding Boolean logic value in the charging process start and stop phases conforms to the standard grounding protection protocol, and determines the instantaneous effectiveness of the grounding loop. The load parameter extraction submodule obtains the power consumption parameters from the power metering unit of the controller. The power consumption parameters include real-time current, real-time voltage, active power, and reactive power. It calculates the root mean square value and peak factor of the power consumption parameters in the current sampling period and evaluates the current load level and power quality. The contactor diagnostic submodule obtains the contactor adhesion Boolean logic value fed back by the controller, and combines it with the real-time current status in the power parameters to determine whether the contactor adhesion Boolean logic value reflects the on / off state of the physical contacts, and identifies potential adhesion risks or drive failures. The state aggregation encapsulation submodule gathers the insulation resistance timing snapshot, the instantaneous effectiveness of the grounding loop, the current load level and power quality, and the potential sticking risk or drive failure. Based on the data structure definitions of electrical integrity characteristics, grounding safety characteristics, load operation characteristics, and contactor reliability characteristics, it generates a local state dataset.

4. The real-time monitoring system for charging piles based on a cloud platform according to claim 3, characterized in that, The specific functions of the edge security discrimination module are as follows: The insulation safety verification submodule acquires the electrical integrity features included in the local state dataset, extracts the insulation state resistance value, and strictly compares the insulation state resistance value with the minimum insulation resistance threshold. If the insulation state resistance value is lower than the minimum insulation resistance threshold, it is determined to be an insulation failure. The grounding logic verification submodule acquires the grounding safety features included in the local status dataset, extracts the grounding Boolean logic value, and checks whether the grounding Boolean logic value is "true". If the grounding Boolean logic value is "false", it is determined to be a grounding fault. The contactor status judgment submodule obtains the contactor reliability characteristics included in the local status dataset, extracts the contactor adhesion Boolean logic value, checks whether the contactor adhesion Boolean logic value is "true", and if the contactor adhesion Boolean logic value is "true", it is determined that the contactor is adhesion. The local baseline triggering submodule integrates the determination results of the insulation failure, the grounding fault, and the contactor adhesion to determine whether the local status dataset triggers the local safety baseline. The local safety baseline defines a logical rule that triggers when any key electrical safety feature fails, and generates a local safety trigger signal. The status code generation submodule, based on the local safety trigger signal and according to the structural definition of safety level identifier, safety trigger flag, safety judgment timestamp, and fault category code, fills in the safety status information, generates a safety status code, and extracts the power consumption parameters from the load operation characteristics of the local status dataset.

5. The real-time monitoring system for charging piles based on a cloud platform according to claim 4, characterized in that, The specific functions of the pile group communication aggregation module are as follows: The local message receiving submodule establishes a secure communication link with the edge security discrimination module, receives the security status code in real time, synchronously acquires the power consumption parameters, and performs data verification on the security status code and the power consumption parameters to ensure that the integrity of the data is not compromised during transmission. The charging pile identifier loading submodule calls the locally stored charging pile identifier. The charging pile identifier is a globally unique device code used to uniquely identify an individual charging pile on the cloud platform, ensuring that the format of the charging pile identifier conforms to the cloud access specification. The status message construction submodule obtains the verified security status code, the power consumption parameters, and the loaded pile identifier. Based on the unique pile identifier, energy consumption statistics, communication time identifier, and predetermined data format of the pile group operation category, it maps the security status code to the pile group operation category and the energy consumption statistics. Combines the current system timestamp as the communication time identifier, and constructs the pile group status message. The message queue management submodule stores the completed pile group status messages into a local message queue to be uploaded, and sorts the pile group status messages in the message queue to be uploaded by priority, giving priority to messages including high-risk security status codes. The data encryption upload submodule extracts the pile group status message from the message queue to be uploaded, encrypts the pile group status message using an asymmetric encryption algorithm, and transmits the encrypted pile group status message to the cloud integrated monitoring module through the secure socket layer channel.

6. The real-time monitoring system for charging piles based on a cloud platform according to claim 5, characterized in that, The cloud-based integrated monitoring module generates the feature vector of the pile group by including: The message parsing and extraction submodule receives and decrypts the pile group status message, parses the data structure of the pile group status message, extracts the unique identifier of the pile, energy consumption statistics, communication time identifier and the pile group operation category, and backtracks to extract the original safety status code and power consumption parameters from the pile group status message. The feature quantization encoding submodule obtains the safety level identifier and fault category code from the safety status code, performs sequential encoding on the safety level identifier, performs one-thermal encoding on the fault category code, performs Z-score normalization on the continuous values ​​in the power consumption parameters, and generates standard energy features and quantized safety features. The time-series feature construction submodule, based on the communication time identifier, aggregates the historical status messages of the pile group within the past preset time window of the unique identifier of the pile body, and counts the slope of the change trend of the power consumption parameters and the trigger frequency of the safety status code to construct time-series dynamic features; The feature vector normalization submodule integrates the standard energy feature, the quantized security feature, and the temporal dynamic feature, and concatenates and aligns all features according to the fixed dimensions required by the input layer of the support vector machine model, filling missing values ​​with zero values ​​to generate the pile group feature vector.

7. The real-time monitoring system for charging piles based on a cloud platform according to claim 6, characterized in that, The cloud-based integrated monitoring module calculates the monitoring results of the pile group by including: The model risk classification submodule inputs the generated pile group feature vector into the pre-trained support vector machine model. The support vector machine model uses the radial basis kernel function to calculate the distance from the pile group feature vector to the decision hyperplane and outputs the preliminary classification label of the power grid load risk level, which includes three levels: high risk, medium risk, and low risk. The energy consumption status assessment submodule calculates the deviation between the current charging pile's instantaneous energy consumption and the historical average energy consumption based on the standard energy characteristics in the pile group feature vector. Combined with the power grid load risk level, it assesses the instantaneous contribution or impact of the current charging pile on the power grid stability and generates the energy consumption status assessment. The fault trend prediction submodule uses a hidden Markov model to analyze the evolution of the temporal dynamic features in the pile group feature vector, including the transition probability of the fault category encoding, to predict the probability of a critical fault occurring in the next cycle and generate the fault trend prediction. The system health calculation submodule integrates the preliminary classification label of the power grid load risk level, the energy consumption status assessment, and the fault trend prediction, and calculates a comprehensive score through a weighted fusion algorithm. The comprehensive score is the system health index. The power grid load risk level, the energy consumption status assessment, the fault trend prediction, and the system health index are combined to generate the pile group monitoring results.

8. The real-time monitoring system for charging piles based on a cloud platform according to claim 3, characterized in that, When establishing the insulation resistance time-series snapshot, the electrical data acquisition submodule sets a fixed-length sliding time window, continuously acquires the insulation state resistance value sequence within the sliding time window, analyzes the spectral characteristics of the insulation state resistance value sequence using fast Fourier transform, identifies target frequency fluctuations caused by moisture or cable aging, and calculates the mean, variance, and minimum value of the insulation state resistance values ​​within the sliding time window. The spectral characteristics, the mean, the variance, and the minimum value are then encapsulated into the electrical integrity feature. When evaluating the current load level and power quality, the load parameter extraction submodule not only acquires the real-time current and real-time voltage from the power consumption parameters, but also synchronously monitors the grid frequency offset, calculates the phase difference between the real-time current and the real-time voltage to obtain the real-time power factor, iteratively analyzes the harmonic components of the real-time current, calculates the total harmonic distortion value, and aggregates the real-time power factor, the total harmonic distortion value, the grid frequency offset, the active power, and the reactive power together to form the load operating characteristics.

9. The real-time monitoring system for charging piles based on a cloud platform according to claim 4, characterized in that, When the local baseline triggering submodule determines whether the local state dataset triggers the local safety baseline, the local safety baseline is defined as a multi-dimensional logic matrix. The multi-dimensional logic matrix includes not only the triggering condition that the insulation state resistance value is lower than the minimum insulation resistance threshold, but also the composite triggering condition that the number of consecutive losses of the grounding Boolean logic value exceeds a preset threshold, and the logic conflict between the contactor adhesion Boolean logic value and the current feedback value in the power consumption parameters. The satisfaction of any one of the conditions generates the local safety trigger signal. When generating the security status code, the status code generation submodule acquires the local security trigger signal. If the local security trigger signal is true, it sets the security level identifier to "high risk", sets the security trigger flag to "local trigger", and fills in the corresponding fault category code according to the specific conditions that trigger the local security baseline. At the same time, it records the current system time as the security determination timestamp.

10. The real-time monitoring system for charging piles based on a cloud platform according to claim 7, characterized in that, After outputting the preliminary classification label of the power grid load risk level, the model risk classification submodule also needs to calculate the Euclidean distance from the pile group feature vector to the decision hyperplane, and convert the Euclidean distance into a classification confidence score. The classification confidence score and the preliminary classification label together constitute the operation safety level. When calculating the system health index, the system health calculation submodule uses the weighted fusion algorithm, which is specifically as follows: ； in, Represents the system's health index. It is a risk score quantified based on the stated operational safety level. It is an energy consumption score quantified based on the aforementioned energy consumption status assessment. It is a trend score obtained by quantifying the aforementioned fault trend prediction. , , These represent the fusion weights corresponding to the risk score, the energy consumption score, and the trend score, respectively, and the fusion weights are dynamically adjusted based on historical data.

Images