Intelligent charging gun adaptive overheat protection method and system based on multi-source data

Abstract

The application provides a method and system for intelligent charging gun adaptive overheat protection based on multi-source data, wherein the method comprises the following steps: collecting charging gun thermal management data in real time, processing the charging gun thermal management data to obtain a smooth sequence of the charging gun thermal management data; sliding window intercepts the smooth sequence, calculates a thermal cumulative anomaly score of a current monitoring period; and corrects and reasons a risk level; calculates a similarity vector; determines an adaptive protection threshold; compares the corrected thermal cumulative anomaly score with the adaptive protection threshold, and determines a charging gun protection strategy according to the comparison result. It can be seen that the application can realize timely and appropriate adaptive protection in the dynamic process of charging gun operation by collecting multi-dimensional data in real time, fusing a fuzzy reasoning model to reason a risk level, adjusting an adaptive protection threshold, and finally determining a charging gun protection strategy through the comparison result of the thermal cumulative anomaly score and the adaptive protection threshold.

Description

Technical Field

[0001] This invention relates to the field of overheat protection technology for charging guns, and in particular to an intelligent overheat protection method and system for charging guns based on multi-source data. Background Technology

[0002] The rapid development of electric vehicle charging infrastructure has made the charging gun, as a key component directly connecting the vehicle and the power source, crucial. Its safety and reliability directly impact the stable operation of the entire charging process and the user experience. With the increasing prevalence of high-power fast charging, the heat generated by the charging gun during operation has significantly increased. If the temperature cannot be effectively controlled in a timely manner, overheating faults can easily occur, even leading to safety accidents. Therefore, building a reliable overheat protection mechanism has become an urgent industry need.

[0003] Currently, most charging gun overheat protection relies on a single temperature threshold or a simple fixed protection strategy. This approach has revealed significant shortcomings in practical use. The real-world environment in which charging guns operate is complex and variable. Multiple factors, such as interface temperature, cable temperature, ambient temperature, charging current magnitude, and voltage fluctuations, act simultaneously and interact with each other. Traditional methods struggle to capture the dynamic relationships between these factors, leading to contradictory protection actions that either trigger prematurely, affecting charging efficiency, or react too slowly to prevent the temperature from continuing to rise.

[0004] Overheating is not caused by a single parameter, but rather by strong coupling and interaction between multiple data sources. For example, when the charging current suddenly increases, the cable and interface temperatures rise rapidly. However, if the ambient temperature is low and heat dissipation is good, the actual overheating risk may not be severe. Conversely, in a high-temperature environment, even with a moderate current, the rate of heat accumulation will significantly accelerate. This characteristic of multiple parameters jointly determining the overheating state makes it difficult to accurately reflect the true thermal evolution process by relying solely on a fixed temperature point or a single variable, resulting in a serious disconnect between protection strategies and actual thermal risks.

[0005] Therefore, how to achieve timely and appropriate adaptive protection during the dynamic operation of the charging gun has become a key issue that urgently needs to be addressed in the field of intelligent charging gun overheat protection. Summary of the Invention

[0006] In view of this, the technical problem to be solved by the present invention is to provide an intelligent charging gun adaptive overheat protection method and system based on multi-source data, which can realize timely and appropriate adaptive protection during the dynamic process of charging gun operation.

[0007] The technical solution of this invention is implemented as follows: This invention proposes an adaptive overheat protection method for intelligent charging guns based on multi-source data, comprising the following steps: S1. Real-time acquisition of charging gun thermal management data, including charging gun interface temperature sequence, cable temperature sequence, ambient temperature sequence, charging current sequence and voltage sequence, and processing of the charging gun thermal management data to obtain a smoothed sequence of the charging gun thermal management data. S2. Extract the smooth sequence according to the preset length sliding window, and calculate the current change slope, interface temperature rise rate, cable temperature rise rate, voltage instantaneous drop number, ambient temperature change slope, dynamic temperature gradient and temperature coupling strength according to the extracted smooth sequence. S3. Calculate the thermal accumulation anomaly score for the current monitoring period based on the current change slope, interface temperature rise rate, cable temperature rise rate, and number of instantaneous voltage drops. S4. Correct the heat accumulation anomaly score according to the slope of the ambient temperature change and the preset heat dissipation conditions to obtain the corrected heat accumulation anomaly score. S5. Input the corrected thermal accumulation anomaly score, dynamic temperature gradient and temperature coupling strength into the fuzzy inference model to infer the risk level; S6. Combine the risk levels corresponding to the current monitoring cycle and the historical monitoring cycles within the preset time window into a risk evolution sequence, and calculate its similarity with each mode in the preset overheating mode library to obtain a similarity vector. S7. Determine the overheating mode corresponding to the maximum value in the similarity vector; and determine the adaptive protection threshold based on the overheating mode; S8. Compare the corrected thermal accumulation anomaly score with the adaptive protection threshold, and determine the charging gun protection strategy based on the comparison result.

[0008] This invention also proposes an intelligent charging gun adaptive overheat protection system based on multi-source data, comprising: The acquisition module is used to acquire charging gun thermal management data in real time. The charging gun thermal management data includes charging gun interface temperature sequence, cable temperature sequence, ambient temperature sequence, charging current sequence and voltage sequence. The module also processes the charging gun thermal management data to obtain a smoothed sequence of the charging gun thermal management data. The first calculation module is used to extract the smoothed sequence according to a preset length sliding window, and calculate the current change slope, interface temperature rise rate, cable temperature rise rate, voltage instantaneous drop number, ambient temperature change slope, dynamic temperature gradient and temperature coupling strength according to the extracted smoothed sequence. The second calculation module is used to calculate the thermal cumulative anomaly score for the current monitoring period based on the current change slope, interface temperature rise rate, cable temperature rise rate, and number of instantaneous voltage drops. The correction module is used to correct the heat accumulation anomaly score according to the slope of the ambient temperature change and the preset heat dissipation conditions, so as to obtain the corrected heat accumulation anomaly score. The inference module is used to input the corrected thermal accumulation anomaly score, dynamic temperature gradient, and temperature coupling strength into the fuzzy inference model to infer the risk level. The third calculation module is used to form a risk evolution sequence by combining the risk levels corresponding to the current monitoring cycle and the historical monitoring cycles within the preset time window, and to calculate the similarity between the sequence and each mode in the preset overheating mode library to obtain a similarity vector. A determination module is used to determine the overheating mode corresponding to the maximum value in the similarity vector; and to determine an adaptive protection threshold based on the overheating mode. The output module is used to compare the corrected thermal accumulation anomaly score with the adaptive protection threshold, and determine the charging gun protection strategy based on the comparison result.

[0009] As can be seen, the intelligent charging gun adaptive overheat protection method and system based on multi-source data proposed in this invention collects multi-dimensional data such as temperature, current, and voltage in real time, combines low-pass filtering and sliding window to calculate the thermal accumulation anomaly score, and corrects it according to the ambient temperature and heat dissipation conditions. It also integrates a fuzzy inference model to infer the risk level, adjusts the adaptive protection threshold, and finally determines the charging gun protection strategy by comparing the thermal accumulation anomaly score with the adaptive protection threshold. Thus, timely and appropriate adaptive protection can be achieved during the dynamic operation of the charging gun. Attached Figure Description

[0010] Figure 1 This is a flowchart of the intelligent charging gun adaptive overheat protection method based on multi-source data proposed in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the intelligent charging gun adaptive overheat protection system based on multi-source data proposed in an embodiment of the present invention. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] like Figure 1As shown in the figure, this invention proposes an adaptive overheat protection method for intelligent charging guns based on multi-source data. This method, characterized by the following steps: S1. Real-time acquisition of charging gun thermal management data, including charging gun interface temperature sequence, cable temperature sequence, ambient temperature sequence, charging current sequence and voltage sequence, and processing of the charging gun thermal management data to obtain a smoothed sequence of charging gun thermal management data. S2. Extract a smooth sequence from a sliding window of a preset length, and calculate the current change slope, interface temperature rise rate, cable temperature rise rate, number of instantaneous voltage drops, ambient temperature change slope, dynamic temperature gradient, and temperature coupling strength based on the extracted smooth sequence. S3. Calculate the cumulative thermal anomaly score for the current monitoring period based on the current change slope, interface temperature rise rate, cable temperature rise rate, and number of instantaneous voltage drops. S4. Correct the heat accumulation anomaly score based on the slope of the ambient temperature change and the preset heat dissipation conditions to obtain the corrected heat accumulation anomaly score. S5. Input the corrected thermal accumulation anomaly score, dynamic temperature gradient and temperature coupling strength into the fuzzy inference model to infer the risk level; S6. Combine the risk levels corresponding to the current monitoring cycle and the historical monitoring cycles within the preset time window into a risk evolution sequence, and calculate its similarity with each mode in the preset overheating mode library to obtain a similarity vector. S7. Determine the overheating mode corresponding to the maximum value in the similarity vector; determine the adaptive protection threshold. S8. Compare the corrected thermal accumulation anomaly score with the adaptive protection threshold, and determine the charging gun protection strategy based on the comparison results.

[0013] As can be seen, the embodiments of the present invention collect multi-dimensional data such as temperature, current, and voltage in real time, combine low-pass filtering and sliding window to calculate the thermal accumulation anomaly score, and make corrections based on ambient temperature and heat dissipation conditions. The risk level is inferred by fuzzy inference model, and the adaptive protection threshold is adjusted. Finally, the charging gun protection strategy is determined by comparing the thermal accumulation anomaly score with the adaptive protection threshold, so as to achieve timely and appropriate adaptive protection during the dynamic operation of the charging gun.

[0014] In a preferred embodiment of the present invention, S1 specifically includes: S101. Real-time acquisition of charging gun thermal management data, including charging gun interface temperature sequence, cable temperature sequence, ambient temperature sequence, charging current sequence, and voltage sequence. S102. Perform low-pass filtering on the charging gun thermal management data to obtain a smoothed sequence of the charging gun thermal management data.

[0015] Specifically, the purpose of this embodiment is to obtain raw time-series data directly related to the thermal and electrical states during the operation of the charging gun, ensuring that the monitoring covers both thermal and electrical parameters and comprehensively reflects the operating status of the equipment.

[0016] Among them, the charging gun interface temperature sequence is the core target of overheat monitoring. As the part with the most dense current contact and the weakest heat dissipation, the interface is a high-incidence point of overheating faults. Its temperature change directly reflects the heat accumulation state of the core components and is the core basis for overheating judgment.

[0017] Cable temperature sequences are used to supplement heat distribution monitoring. Cables are responsible for current transmission, and their temperature changes reflect line loss and heat dissipation. At the same time, they need to be coordinated with interface temperatures to provide a basis for subsequent calculations of dynamic temperature gradients and temperature coupling strength, and to avoid overlooking potential cable overheating hazards.

[0018] Ambient temperature sequence is key to operating condition calibration. Ambient temperature directly affects the heat dissipation efficiency of the charging gun. Its data is used for subsequent calculation of the slope of ambient temperature change and correction of heat accumulation anomaly score, eliminating misjudgments and omissions caused by environmental interference, and ensuring that the monitoring results are consistent with the actual operating scenario.

[0019] The charging current sequence is a heat source monitoring item. The current is the core source of heat generation in the charging gun. The heat generation is proportional to the square of the current. Its changing trend reflects the fluctuation of the heat generation rate and is a key trigger for abnormal heat accumulation. At the same time, it provides core input features for the subsequent isolated forest model.

[0020] Voltage sequence is a supplementary item for abnormal causes. Instantaneous voltage drops can reflect abnormalities in the charging circuit, such as poor contact or circuit failure. These abnormalities are often accompanied by increased local resistance and intensified heat generation. The data is used to count the number of instantaneous voltage drops, providing supplementary basis for the calculation of abnormality scores and improving the dimensions of abnormality monitoring.

[0021] This embodiment, by collecting the above data sequence in real time, can achieve a complete monitoring closed loop from heat source to heat distribution status to environmental impact to loop anomaly, providing comprehensive and necessary data support for subsequent feature calculation, model reasoning, and protection strategy formulation.

[0022] After data acquisition, to eliminate high-frequency noise caused by sensor jitter, on-site electromagnetic interference, sampling spikes, etc., and to filter out invalid interference signals to obtain a smooth data sequence that closely reflects the actual operating state of the equipment, low-pass filtering can be performed to obtain a smooth sequence. The five types of acquired data sequences are each subjected to low-pass filtering. Using low-pass filtering algorithms adapted to industrial scenarios, such as Butterworth low-pass filtering, and with a pre-set reasonable cutoff frequency based on data characteristics, each type of sequence is processed separately. After filtering out high-frequency noise, the corresponding five types of smooth sequences are output, ensuring that the timestamps of the smooth sequences are consistent with those of the original sequences.

[0023] As can be seen, this embodiment, by collecting multi-source data sequences, comprehensively covers the core parameters related to charging gun overheating, effectively avoiding monitoring blind spots caused by missing parameters. By eliminating noise interference, the data curves are smoother, clearly presenting the true trend of parameter changes and preventing false fluctuations caused by noise from affecting subsequent data processing.

[0024] In a preferred embodiment of the present invention, S2 specifically includes: S201. Extract a smooth sequence from a sliding window of a preset length, and calculate the current change slope, interface temperature rise rate, cable temperature rise rate, number of instantaneous voltage drops, and ambient temperature change slope based on the extracted smooth sequence. S202. Calculate the dynamic temperature gradient based on the average value of the charging gun interface temperature sequence and the cable temperature sequence, and calculate the temperature coupling strength through the Pearson correlation coefficient of the interface temperature and cable temperature sequences.

[0025] Specifically, using a fixed-length sliding window that slides along the time axis in real time allows for focusing on the most recent preset number of monitoring data points, avoiding interference from irrelevant historical data in the current state analysis, and meeting the timeliness requirements of real-time monitoring. For example, using a fixed sliding window of a preset length, such as 10 sampling points, the five smoothed sequences obtained in S102 are truncated in real time to obtain the latest data within the window; based on the data within the window, the following five types of features are calculated sequentially: Current change slope: The slope value is obtained by fitting the relationship between current data and time within a window using linear regression; Interface temperature rise rate: The difference between the final and initial values ​​of the charging gun interface temperature within the calculation window is divided by the window duration to obtain the interface temperature rise rate per unit time. Cable temperature rise rate: The same calculation method as the interface temperature rise rate is used, and it is calculated based on the cable temperature data within the window; Voltage instantaneous drop count: A preset voltage drop judgment threshold, which counts the number of times the instantaneous voltage drop exceeds the threshold within the window; Slope of ambient temperature change: The slope value is obtained by fitting the relationship between ambient temperature data and time within the window through linear regression.

[0026] The core function of the current change slope is to reflect the dynamic trend of the charging current and capture sudden current changes. Since current is the primary source of heat generation in the charging gun, its changes are directly related to fluctuations in the heat generation rate, providing key input features for the subsequent isolated forest model and assisting in quantifying the degree of abnormal heat accumulation. The interface temperature rise rate directly reflects the speed of heat accumulation at the charging gun interface; a higher rate indicates faster heat accumulation and a higher risk of overheating. It is a key input for the isolated forest model, accurately capturing early signs of interface overheating. The cable temperature rise rate supplements the reflection of the cable's heat accumulation status, avoiding the overheating risks of cables that are missed by only monitoring the interface temperature. Working in conjunction with the interface temperature rise rate, it provides comprehensive coverage. The thermal change trend of the core components of the charging gun is analyzed, which also provides input for the isolated forest model, improving the comprehensiveness of anomaly identification. The number of instantaneous voltage drops can quantify the degree of anomaly in the charging circuit. Instantaneous voltage drops usually correspond to problems such as poor circuit contact and line faults. These anomalies are accompanied by increased local resistance and intensified heat generation. As a supplementary input to the isolated forest model, it can improve the identification of the causes of thermal accumulation anomalies and reduce missed detections. The core function of the slope of the ambient temperature change is to reflect the dynamic changes of the current environmental heat dissipation conditions. It provides the sole core basis for determining the correction coefficient and is used to eliminate the interference of environmental heat dissipation conditions on the thermal accumulation anomaly score, ensuring that the subsequent anomaly score is consistent with the actual working conditions.

[0027] Simultaneously, this embodiment also calculates the dynamic temperature gradient and temperature coupling strength. The dynamic temperature gradient is obtained by extracting the average value of the interface temperature sequence and the average value of the cable temperature sequence within the current sliding window, and calculating the absolute value of the difference between the two average values. This reflects the non-uniformity of heat conduction within the charging gun. Under normal circumstances, the interface and cable are connected by a metal conductor, resulting in good heat conduction. Their temperatures should be close and change synchronously, with a small gradient. If the interface is much higher than the cable, it indicates heat accumulation at the interface, abnormal contact resistance, or impaired heat dissipation. If the cable is much higher than the interface, it suggests potential internal cable damage, such as broken strands or insulation aging leading to eddy current heating. The temperature coupling strength is calculated based on the interface temperature sequence and cable temperature sequence within the current sliding window using the Pearson correlation coefficient formula. This correlation coefficient is the temperature coupling strength, with a value range of [-1, 1]. It measures the synchronicity and temporal consistency of the temperature changes between the interface and cable. Under normal operating conditions: Current changes cause both to heat up simultaneously, and the temperature curves are highly synchronized, with the temperature coupling strength approaching 1; When there are abnormal operating conditions, such as poor contact, the interface temperature responds quickly while the cable temperature lags behind, and the temperature coupling strength will decrease; When a partial short circuit in the cable causes a sudden rise in cable temperature while the interface remains unchanged, the temperature coupling strength will decrease significantly or even become negatively correlated.

[0028] As can be seen, this embodiment captures recent data through a sliding window, which is suitable for real-time monitoring scenarios and avoids historical data interfering with the current status judgment; the dynamic temperature gradient can directly reflect the temperature difference between the interface and the cable, and promptly capture potential local overheating risks; the temperature coupling strength can identify abnormal heat distribution, providing additional evidence for overheating risk assessment.

[0029] In a preferred embodiment of the present invention, S3 specifically includes: S301, Pre-trained isolated forest model; S302. Input the current change slope, interface temperature rise rate, cable temperature rise rate, and number of instantaneous voltage drops into the isolated forest model to calculate the thermal accumulation anomaly score at the current moment.

[0030] In detail, the Isolation Forest is based on the core principle that abnormal samples are more likely to be randomly isolated than normal samples. It constructs multiple random decision trees to randomly divide the feature space and calculates the average isolation path length of the samples. The shorter the path length, the further the sample deviates from the normal distribution and the higher the degree of abnormality. During the model training process, the range and sensitivity of the abnormality score are calibrated by using historical normal operation data and a small amount of abnormal data to adapt to the operating characteristics of the charging gun. When training an isolated forest model, historical data of the charging gun under normal operating conditions and a small amount of known thermal accumulation anomaly data can be collected. The current change slope, interface temperature rise rate, cable temperature rise rate, and number of instantaneous voltage drops corresponding to these data can be extracted to form the model training dataset. For example, historical operating data of a certain model of DC charging gun (rated power 60kW) can be collected, including 1000 sets of normal operating data (covering normal indoor temperature, outdoor light wind, and other conventional operating conditions, with a sampling period of 1 second / time) and 50 sets of known thermal accumulation anomaly data (covering scenarios such as slight interface overheating and cable heat accumulation). For each set of data, four types of features calculated by S201 are extracted. For example, the features of one set of normal data are: current change slope 0.02℃ / s, interface temperature rise rate 0.05℃ / s, cable temperature rise rate 0.04℃ / s, and number of instantaneous voltage drops 0; the features of one set of anomaly data are: current change slope 0.8℃ / s, ... The interface temperature rise rate is 0.3℃ / s, the cable temperature rise rate is 0.25℃ / s, and the number of instantaneous voltage drops is 2. These four features from 1050 sets of data were compiled and summarized to form a complete model training dataset. Then, model construction was performed: an isolated forest model was built, setting reasonable parameters such as the number of decision trees and the sample contamination rate. Considering the real-time requirements of charging gun monitoring, an isolated forest model was built, with 100 decision trees to balance model accuracy and inference speed, avoiding insufficient accuracy due to too few decision trees and slow inference due to too many. The sample contamination rate was set to 0.05, adapting to a training dataset with 5% abnormal data, reflecting the scarcity of charging gun fault samples. Other parameters were set according to industry default standards, completing the initial model construction. Finally, model training was performed: the training dataset was input into the model, and the model was trained by randomly partitioning the feature space and calculating the isolated path length of samples. After training, the model parameters were saved. The 1050 feature datasets prepared above are divided into a training set of 735 groups and a validation set of 315 groups in a 7:3 ratio. The training set is then input into the constructed Isolation Forest model. The model automatically and randomly partitions the feature space composed of the four types of features, and calculates the isolation path length for each sample in each group. Normal samples have longer path lengths, while abnormal samples have shorter path lengths. The path length and abnormality level are calibrated through the validation set to ensure that the model can accurately distinguish between normal and abnormal features. After training, the model's decision tree parameters, path length calibration standards, and other core information are saved. During subsequent monitoring, only the four types of features in real time need to be input to quickly call the model to output the abnormality score.

[0031] Based on the inference principle of the isolated forest model, the feature data at the current moment can be input into the trained model. The model calculates the average isolated path length of the feature samples through the trained random decision tree. Taking the S301 as a benchmark, the average isolated path length of normal samples was calibrated with 1000 sets of normal data. The mean value of the average isolated path length of normal samples was μ=3.2 and the standard deviation was σ=0.8. The preset abnormal critical standard value was μ-σ=2.4. It is assumed that the four real-time features output by the S201 at the current time are normal samples: the slope of the current change is 0. The following parameters were input into the model: 0.03℃ / s, interface temperature rise rate 0.06℃ / s, cable temperature rise rate 0.05℃ / s, and voltage instantaneous drop count 0. After inputting these parameters, 100 decision trees were traversed. The single path length of 80 trees was 3, and the single path length of 20 trees was 4. The sum of all path lengths was 80×3+20×4=320. The average isolated path length was 320÷100=3.2. This value is consistent with the mean μ, indicating that the sample is close to the normal sample and consistent with the normal operating conditions corresponding to the features.

[0032] Assuming the four real-time features output by S201 are abnormal samples: current change slope 0.8℃ / s, interface temperature rise rate 0.3℃ / s, cable temperature rise rate 0.25℃ / s, and voltage instantaneous drop count 2 (fitting abnormal operating conditions), after inputting them into the model, 100 decision trees are traversed. Among them, the single path length of 70 trees is 1, and the single path length of 30 trees is 2. The sum of all path lengths is 70×1+30×2=130, and the average isolated path length is 130÷100=1.3. This value is less than the preset abnormal critical standard value of 2.4, so it can be clearly determined that it deviates from the normal sample.

[0033] Then, through normalization, the path length is converted into a thermal accumulation anomaly score within the range of 0-1. The higher the score, the more severe the current thermal accumulation anomaly. The current change slope, interface temperature rise rate, cable temperature rise rate, and number of instantaneous voltage drops are input into the pre-trained and saved isolated forest model in the feature order during model training. The model performs inference calculations on the input features and outputs the thermal accumulation anomaly score at the current monitoring time.

[0034] As can be seen, this embodiment uses an isolated forest model, which does not require a large number of labeled fault samples and is suitable for real-world scenarios where charging gun fault data is scarce. It also quickly quantifies the current degree of thermal accumulation anomaly through normalization processing, providing an intuitive and quantifiable basis for subsequent corrections and risk assessments.

[0035] In a preferred embodiment of the present invention, S4 specifically includes: S401. Based on the slope of ambient temperature change and the preset heat dissipation condition language value, determine the correction coefficient through the mapping relationship between the pre-stored ambient temperature change slope, heat dissipation condition language value and correction coefficient. S402. The thermal cumulative anomaly score is proportionally corrected based on the correction factor to obtain the corrected thermal cumulative anomaly score.

[0036] In this application, in order to eliminate the interference of environmental heat dissipation conditions on the heat accumulation anomaly score, make the anomaly score adapt to the current actual operating conditions, and avoid misjudgment or omission of anomalies caused by differences in heat dissipation conditions, the heat accumulation anomaly score can be corrected.

[0037] The slope of ambient temperature change can intuitively reflect the quality of on-site heat dissipation conditions: a negative slope indicates that the ambient temperature is continuously decreasing and the heat dissipation conditions are good; a positive slope and a larger value indicate that the ambient temperature is continuously increasing and the heat dissipation conditions are worse. Through the mapping relationship between the pre-stored range of ambient temperature change slope, heat dissipation condition language value and correction coefficient, corresponding correction coefficients can be matched for different heat dissipation conditions to achieve conditional adaptation and correction for abnormal scores.

[0038] First, a mapping relationship is pre-defined: the slope of the ambient temperature change is pre-divided into intervals, and each interval is matched with a corresponding heat dissipation condition linguistic value, such as good heat dissipation, average heat dissipation, poor heat dissipation, and extremely poor heat dissipation, as well as a correction coefficient. Specifically, the core logic of the mapping is that the better the heat dissipation, the smaller the correction coefficient; the worse the heat dissipation, the larger the correction coefficient. Heat dissipation can be intuitively reflected by the slope of the ambient temperature change, based on which a mapping relationship can be established between the ambient temperature change slope, the heat dissipation condition linguistic value, and the correction coefficient. For example, based on the sign and absolute value of the ambient temperature change slope, four reasonable intervals can be divided, each interval corresponding to a specific heat dissipation condition linguistic value and correction coefficient.

[0039] Interval 1: Ambient temperature change slope < -0.5℃ / s. In this interval, the slope is negative and the absolute value is large, corresponding to good heat dissipation. The correction coefficient ranges from 0.7 to 0.9, e.g., 0.8. A negative slope indicates that the ambient temperature is decreasing, and a large absolute value indicates that the ambient temperature is changing rapidly, meaning the ambient temperature is dropping quickly. At this time, heat dissipation efficiency is improved, and the heat generated by the charging gun can be dissipated quickly. However, the initial heat accumulation anomaly score is prone to being too high and needs to be corrected with a small coefficient to avoid misjudging it as an anomaly.

[0040] Interval 2: -0.5℃ / s ≤ Ambient temperature change slope ≤ 0.5℃ / s. In this range, the slope is close to 0 or fluctuates slightly, corresponding to a general heat dissipation condition. The correction factor is 1.0, meaning no correction is needed. This interval indicates a relatively stable ambient temperature, such as a normal indoor temperature environment. Heat dissipation conditions are at a normal level, and the original heat accumulation anomaly score accurately reflects the actual situation, requiring no additional correction. Interval 3: 0.5℃ / s < Ambient temperature change slope ≤ 1.0℃ / s. In this interval, the slope is positive and the absolute value is large, corresponding to poor heat dissipation. The correction coefficient ranges from 1.1 to 1.3, e.g., 1.2. A positive slope indicates that the ambient temperature is rising in this interval, and a large absolute value indicates that the ambient temperature is changing rapidly, meaning the ambient temperature is rising quickly. At this time, the heat dissipation capacity is significantly reduced, and the heat generated by the charging gun easily accumulates. The initial heat accumulation anomaly score is likely to be low, requiring a larger correction coefficient to avoid misjudging it as an anomaly.

[0041] Interval 4: Ambient temperature change slope > 1.0℃ / s. In this interval, the slope is positive and the absolute value is large, corresponding to a limited heat dissipation condition. The correction factor ranges from 1.4 to 1.6, e.g., 1.5. A positive slope indicates that the ambient temperature is rising in this interval, and a large absolute value indicates that the ambient temperature is changing extremely rapidly, meaning the ambient temperature is rising quickly. At this point, heat dissipation fails, the risk of heat accumulation is extremely high, and the anomaly score needs to be significantly increased to trigger a warning.

[0042] The above method establishes a mapping relationship by dividing the range of the slope of the ambient temperature change and presetting the corresponding heat dissipation condition language value and correction coefficient for each range.

[0043] Then, coefficient matching is performed: the slope of the current ambient temperature change calculated by S201 is extracted, the slope interval to which it belongs is determined, and the correction coefficient corresponding to the interval can be determined according to the pre-stored mapping relationship.

[0044] When making corrections, the proportional correction principle can be used to adjust the heat accumulation anomaly score using a correction coefficient. The corrected heat accumulation anomaly score = heat accumulation anomaly score × correction coefficient. Specifically, when heat dissipation is good, the correction coefficient is less than 1 to reduce the proportion of the heat accumulation anomaly score and avoid misjudgment; when heat dissipation is poor, the correction coefficient is greater than 1 to increase the proportion of the heat accumulation anomaly score and avoid missed judgment, ensuring that the score can truly reflect the degree of anomaly under different operating conditions. After the calculation is completed, it is necessary to ensure that the corrected score remains within the range of 0-1. If it exceeds the range, normalization processing should be performed to ensure the rationality of the heat accumulation anomaly score.

[0045] In this embodiment, determining the correction coefficient based on the slope of the ambient temperature change can eliminate the interference of environmental factors on the abnormal score and avoid misjudgment; by correcting the thermal accumulation abnormal score through the correction coefficient, the accuracy of the thermal accumulation abnormal score is optimized, making the thermal accumulation abnormal score more consistent with the actual operating conditions of the charging gun, and providing a more reliable quantitative basis for subsequent risk level reasoning.

[0046] S5. Input the corrected thermal accumulation anomaly score, dynamic temperature gradient, and temperature coupling strength into the fuzzy inference model to infer the risk level.

[0047] In order to achieve a refined classification of overheating risk of charging guns, based on the corrected thermal accumulation anomaly score, dynamic temperature gradient, and temperature coupling strength, and to abandon the one-sidedness of single feature judgment, this application adapts the gradual and fuzzy evolution characteristics of overheating risk of charging guns by using fuzzy set theory and fuzzy reasoning principles. First, model preparation is performed: a fuzzy inference model adapted to the overheating monitoring scenario of charging guns is pre-built, clarifying the fuzzy classification criteria for the three input features and the classification criteria for the output risk level; a comprehensive fuzzy rule base is pre-built, covering all fuzzy combinations of the three input features. For example, if the corrected anomaly score is low, the dynamic temperature gradient is small, and the temperature coupling strength is strong, then the risk level is extremely low; if the corrected anomaly score is low, the dynamic temperature gradient is medium, and the temperature coupling strength is medium, then the risk level is low; if the corrected anomaly score is medium, the dynamic temperature gradient is medium, and the temperature coupling strength is medium, then the risk level is medium; if the corrected anomaly score is high, the dynamic temperature gradient is medium, and the temperature coupling strength is weak, then the risk level is high; if the corrected anomaly score is high, the dynamic temperature gradient is large, and the temperature coupling strength is weak, then the risk level is extremely high. The rule base can be flexibly supplemented and adjusted later according to the operating characteristics of different charging gun models and the requirements of on-site working conditions. After completing the rule base, feature input is performed: the corrected thermal accumulation anomaly score and dynamic temperature gradient are synchronously input into the pre-built fuzzy inference model to ensure that the input sequence of the three features is consistent.

[0048] The fuzzy inference model first fuzzifies the three input features and calculates the membership degree of each feature value to its corresponding fuzzy set: low, medium, high / small, medium, large / weak, medium, and strong. Then, it calls a preset fuzzy rule library to match all inference rules that conform to the current fuzzy membership degrees of the three features. The matched rules are then logically synthesized to obtain a comprehensive fuzzy output set. Finally, the fuzzification is defuzzified using the centroid method, transforming the fuzzy output set into a unique and explicit linguistic risk level. This outputs the overheating risk level of the charging gun at the current moment, which can be one of five levels: extremely low, low, medium, high, and extremely high. The inference process and the membership degrees of each feature are recorded simultaneously for subsequent traceability and rule optimization.

[0049] As can be seen, this embodiment takes into account the synergistic effect of multiple features and avoids the one-sidedness of single feature judgment: it can comprehensively and accurately depict the thermal operating state of the charging gun; it has strong versatility and is suitable for multiple application scenarios. The fuzzy rule base supports flexible adjustment and can modify and improve the inference rules according to the heat dissipation characteristics and rated parameters of different models of charging guns, such as DC charging guns and AC charging guns, as well as the differences in working conditions of different application scenarios, such as outdoor high temperature scenarios and indoor normal temperature scenarios.

[0050] In a preferred embodiment of the present invention, S6 specifically includes: S601. Combine the risk levels corresponding to the current monitoring cycle and the historical monitoring cycles within the preset time window into a risk evolution sequence. S602. Calculate the similarity between it and each mode template in the preset overheating mode library using the dynamic time warping algorithm to obtain the similarity vector.

[0051] In this application, to avoid the randomness of risk levels in a single monitoring period, a time window can be preset, such as the previous 9 monitoring periods. The risk level of the current monitoring period and the risk levels of all historical monitoring periods within the time window are extracted. These risk levels are arranged in chronological order of monitoring time to form a risk evolution sequence. Then, the sequence is standardized to convert the verbalized risk levels into corresponding values, such as extremely low risk level = 0, low = 1, medium = 2, high = 3, extremely high = 4, which facilitates subsequent similarity calculation.

[0052] Then, the current risk evolution sequence is precisely matched with preset typical overheating patterns to quantify the similarity between the current risk trend and each typical pattern. A typical overheating pattern library can be pre-established, storing pattern templates for different overheating trends, such as normal pattern, slow heating pattern, rapid heating pattern, and fluctuating heating pattern. Each template is a numerical sequence with the same length as the risk evolution sequence. Then, similarity calculation is performed. Using a dynamic time warping algorithm, the similarity distance between the current risk evolution sequence and each pattern template in the pattern library is calculated, and the similarity distance is converted into a similarity score. The smaller the similarity distance, the higher the similarity score. Finally, vector generation is performed. The similarity scores of the current sequence and each pattern template are arranged in a preset order to form a similarity vector. Each element in the vector corresponds to the similarity score of a typical pattern.

[0053] In detail, data from the current monitoring period and its nine preceding historical monitoring periods, totaling ten periods, were selected. The risk level for each period was recorded in natural language, such as extremely low risk level = 0, low = 1, medium = 2, high = 3, and extremely high = 4, arranged chronologically to form the original risk sequence. Subsequently, this sequence was standardized, mapping the verbal risk levels to preset integer values.

[0054] The system establishes a library of typical overheating patterns, containing several typical risk evolution templates. Each template is a numerical sequence of the same length as the aforementioned risk evolution sequence. The typical templates are categorized into the following four types: Normal mode =[0, 0, 0, 0, 0, 0, 0, 0, 0, 0] Slow heating mode =[0, 0, 1, 1, 2, 2, 3, 3, 3, 4] Rapid heating mode =[0, 0, 0, 0, 1, 2, 3, 4, 4, 4] Fluctuating heating mode =[0, 1, 0, 2, 1, 3, 2, 3, 4, 3].

[0055] The risk levels of the 10 cycles are arranged in chronological order as follows: “Low”, “Low”, “Medium”, “Medium”, “High”, “High”, “Extremely High”, “Extremely High”.

[0056] According to the preset mapping rules: "extremely low"=0, "low"=1, "medium"=2, "high"=3, "extremely high"=4, it is converted into a numerical sequence: The current risk evolution sequence S = [1, 1, 2, 2, 3, 3, 3, 4, 4, 4].

[0057] Then, the similarity with each template is calculated item by item using a dynamic time warping algorithm, in order to achieve a rapid heating pattern with similar trends. For example, let's compare the current sequence S with... Inputting the dynamic time warping algorithm, the algorithm automatically aligns the time axes of the two sequences, finding that they highly overlap in the latter half, while the first half shows a slight offset, but the overall upward rhythm is consistent. Assuming the algorithm outputs a warped distance of 0.6, according to the conversion rule "similarity = 1 / (1 + distance)", the similarity is calculated as 1 / (1 + 0.6) = 0.625. Similarly, the similarity to the normal pattern is calculated... The normalized distance is 5, and the similarity is 0.167; similar to the slow heating pattern. The normalized distance is 2.5, and the similarity is 0.286; similar to the fluctuating heating pattern. The normalized distance was 3.0, and the similarity was 0.250. Finally, the similarity scores were arranged in the preset order: normal, slow heating, rapid heating, and fluctuating heating. The similarity vector V = [0.167, 0.286, 0.625, 0.250].

[0058] As can be seen, this embodiment can avoid the randomness of single-period data by integrating multiple periodic risk levels; the dynamic time warping algorithm can adapt to overheating patterns with different rhythms, solve the problem of inconsistent length and rhythm in time series matching, and improve the accuracy of pattern matching.

[0059] In a preferred embodiment of the present invention, S7 specifically includes: S701. Determine the overheating mode corresponding to the maximum value in the similarity vector; S702. Read the trigger margin coefficient corresponding to the overheating mode from the preset protection margin mapping table, and adjust the preset abnormal protection threshold according to the trigger margin coefficient to obtain the final adaptive protection threshold.

[0060] Specifically, this embodiment can determine the overheating mode of the current charging gun overheating risk, thereby clarifying the development speed and severity of the risk and providing a clear direction for the adaptive adjustment of the subsequent protection threshold. Each element in the similarity vector corresponds to the similarity between the current sequence and an overheating mode, and the typical mode corresponding to the maximum value is the mode most similar to the current risk evolution trend.

[0061] Because different overheating modes correspond to different risk development speeds and severity levels—modes with higher risks and faster development require earlier protection actions and correspond to lower protection thresholds—a pre-defined protection margin mapping table is used to match specific trigger margin coefficients for each typical mode. Higher risks result in smaller margin coefficients, and the basic anomaly protection threshold is adjusted proportionally to obtain an adaptive protection threshold suitable for the current mode. First, a protection margin mapping table is pre-established, covering the correspondence between each typical overheating mode and its corresponding trigger margin coefficient. Then, based on the determined current overheating mode, the corresponding trigger margin coefficient is read from the mapping table. Based on the pre-defined anomaly protection threshold, the anomaly protection threshold is multiplied by the trigger margin coefficient to calculate the final adaptive protection threshold. This ensures that the adjusted threshold remains within a reasonable range to match the range of values ​​for the corrected thermal accumulation anomaly score.

[0062] For example, the normal mode corresponds to a trigger margin coefficient of 1.0. In this mode, the charging gun has extremely low risk and no overheating trend, so no threshold adjustment is needed. The slow heating mode corresponds to a trigger margin coefficient of 0.8. In this mode, the risk is moderate and the heating trend is stable. The threshold needs to be appropriately reduced to ensure timely triggering of mild protection. The rapid heating mode corresponds to a trigger margin coefficient of 0.6. In this mode, the risk is extremely high and the heating speed is fast. The threshold needs to be significantly reduced to ensure rapid triggering of severe protection and to avoid escalation of risk. The fluctuating heating mode corresponds to a trigger margin coefficient of 0.9. In this mode, the risk is low and the heating trend is unstable. The threshold needs to be slightly reduced to improve monitoring sensitivity. Taking a preset base threshold of 0.7 and a similarity vector V = [0.167, 0.286, 0.625, 0.250] as an example, the corresponding overheating mode is the rapid heating mode with a margin coefficient of 0.6. In this case, the calculated adaptive protection threshold is 0.7. 0.6 = 0.42. After the calculation is completed, the system automatically checks the threshold range. If the threshold exceeds the range of 0 to 1 due to the adjustment of the basic threshold or the margin coefficient, it will be automatically normalized to a reasonable range.

[0063] As can be seen, this embodiment can determine the dominant mode of the current overheating risk, clarify the development trend and hazard level of the risk, and provide a precise direction for threshold adjustment; it can realize dynamic adaptive adjustment of the protection threshold, with the threshold deeply matched with the current overheating trend. The higher the risk, the lower the threshold, which can trigger protection earlier and ensure the safety of the charging pile.

[0064] S8. Compare the corrected thermal accumulation anomaly score with the adaptive protection threshold, and determine the charging gun protection strategy based on the comparison results.

[0065] In this embodiment, the threshold comparison rule is first determined. If the corrected heat accumulation anomaly score is greater than the adaptive protection threshold, it is an abnormal state; if it is equal to or lower than the threshold, it is a normal state. A continuous anomaly cycle judgment standard can be preset. For example, if there are three consecutive monitoring cycles of anomalies, it can be judged as a mild anomaly; if there are five consecutive monitoring cycles of anomalies, it can be judged as a severe anomaly. Then, the protection strategy is determined according to different judgment results. For example, for mild anomalies, the charging power can be reduced: the charging power is reduced to 50% of the charging gun's rated power to suppress the charging heat generation rate and alleviate heat accumulation. After the power is reduced, the corrected heat accumulation anomaly score is continuously monitored. If the score subsequently falls below the adaptive protection threshold and remains below it for two consecutive cycles, a power recovery command is sent to restore the charging power to normal. The charging gun protection strategy for severe anomalies can be a power outage warning protection, which immediately sends an interrupt charging command to the charging gun control system to completely cut off the charging circuit, triggers an alarm on the site, and notifies maintenance personnel to troubleshoot the fault in a timely manner.

[0066] As can be seen, this embodiment can effectively filter false abnormal signals of 1-2 cycles caused by sensor jitter, electromagnetic interference, instantaneous voltage fluctuations, etc. by performing anomaly judgment in continuous cycles; for mild overheating, power reduction is used to suppress heat accumulation without interrupting charging, which avoids the risk of escalation and maximizes the user's charging experience; for severe overheating, charging is interrupted to completely cut off the heat source and prevent serious equipment failures such as charging gun burnout and circuit short circuit.

[0067] like Figure 2 As shown, this embodiment of the invention also proposes an intelligent charging gun adaptive overheat protection system based on multi-source data, including: The acquisition module 1 is used to acquire the thermal management data of the charging gun in real time. The thermal management data of the charging gun includes the temperature sequence of the charging gun interface, the temperature sequence of the cable, the ambient temperature sequence, the charging current sequence, and the voltage sequence. The module also processes the thermal management data of the charging gun to obtain a smoothed sequence of the thermal management data of the charging gun. The first calculation module 2 is used to extract a smooth sequence according to a preset length sliding window, and to calculate the current change slope, interface temperature rise rate, cable temperature rise rate, voltage instantaneous drop number, ambient temperature change slope, dynamic temperature gradient and temperature coupling strength according to the extracted smooth sequence. The second calculation module 3 is used to calculate the thermal cumulative anomaly score of the current monitoring cycle based on the current change slope, interface temperature rise rate, cable temperature rise rate and voltage instantaneous drop number. Correction module 4 is used to correct the heat accumulation anomaly score according to the slope of the ambient temperature change and the preset heat dissipation conditions, so as to obtain the corrected heat accumulation anomaly score. Inference module 5 is used to input the corrected thermal accumulation anomaly score, dynamic temperature gradient and temperature coupling strength into the fuzzy inference model to infer the risk level; The third calculation module 6 is used to form a risk evolution sequence by combining the risk levels corresponding to the current monitoring cycle and the historical monitoring cycles within the preset time window, and to calculate the similarity between the sequence and each mode in the preset overheating mode library to obtain a similarity vector. Module 7 is used to determine the overheating mode corresponding to the maximum value in the similarity vector and to determine the adaptive protection threshold. Output module 8 is used to compare the corrected thermal accumulation anomaly score with the adaptive protection threshold, and determine the charging gun protection strategy based on the comparison result.

[0068] In a preferred embodiment of the present invention, the acquisition module 1 is specifically used for: Real-time acquisition of charging gun thermal management data, including charging gun interface temperature sequence, cable temperature sequence, ambient temperature sequence, charging current sequence, and voltage sequence; The thermal management data of the charging gun is subjected to low-pass filtering to obtain a smoothed sequence of the thermal management data of the charging gun.

[0069] In a preferred embodiment of the present invention, the first calculation module 2 is specifically used for: The smooth sequence is extracted based on a preset length sliding window, and the current change slope, interface temperature rise rate, cable temperature rise rate, voltage instantaneous drop number, and ambient temperature change slope are calculated based on the extracted smooth sequence. The dynamic temperature gradient is calculated based on the average values ​​of the interface temperature sequence and the cable temperature sequence, and the temperature coupling strength is calculated using the Pearson correlation coefficient between the interface temperature and the cable temperature sequence.

[0070] In summary, this invention, by collecting multi-source data sequences, comprehensively covers the core parameters related to charging gun overheating, effectively avoiding monitoring blind spots caused by missing parameters. By eliminating noise interference, the data curves are smoother, clearly presenting the true trend of parameter changes and preventing false fluctuations caused by noise from affecting subsequent data processing.

[0071] This invention uses a sliding window to capture recent data, making it suitable for real-time monitoring scenarios and avoiding interference from historical data in judging the current state. The dynamic temperature gradient can directly reflect the temperature difference between the interface and the cable, and promptly capture potential local overheating risks. The temperature coupling strength can identify abnormal heat distribution, providing additional evidence for overheating risk assessment.

[0072] The embodiments of this invention employ an isolated forest model, which eliminates the need for a large number of labeled fault samples and is suitable for real-world scenarios where charging gun fault data is scarce. Through normalization processing, the current degree of thermal accumulation anomaly is quickly quantified, providing an intuitive and quantifiable basis for subsequent corrections and risk assessments.

[0073] The embodiments of the present invention determine the correction coefficient based on the slope of the change in ambient temperature, which can eliminate the interference of environmental factors on the abnormal score and avoid misjudgment; by adjusting the abnormal score of heat accumulation through the correction coefficient, the accuracy of the abnormal score is optimized, making the score more consistent with the actual operating conditions of the charging gun, and providing a more reliable quantitative basis for subsequent risk level reasoning.

[0074] The embodiments of this invention take into account the synergistic effects of multiple features and avoid the one-sidedness of single feature judgment: it can comprehensively and accurately depict the thermal operating state of the charging gun; it has strong versatility and is suitable for multiple application scenarios. The fuzzy rule base supports flexible adjustment and can modify and improve the inference rules according to the heat dissipation characteristics and rated parameters of different models of charging guns, such as DC charging guns and AC charging guns, as well as the differences in working conditions of different application scenarios, such as outdoor high temperature scenarios and indoor normal temperature scenarios.

[0075] By integrating multiple periodic risk levels, this invention can avoid the randomness of single-period data; the dynamic time warping algorithm can adapt to overheating patterns with different rhythms, solve the problem of inconsistent length and rhythm in time series matching, and improve the accuracy of pattern matching.

[0076] The embodiments of the present invention can determine the dominant mode of the current overheating risk, clarify the development trend and hazard level of the risk, and provide a precise direction for threshold adjustment; it can realize dynamic adaptive adjustment of the protection threshold, with the threshold deeply matched with the current overheating trend. The higher the risk, the lower the threshold, which can trigger protection earlier and ensure the safety of the charging pile.

[0077] As can be seen, this embodiment can effectively filter false abnormal signals of 1-2 cycles caused by sensor jitter, electromagnetic interference, instantaneous voltage fluctuations, etc. by performing anomaly judgment in continuous cycles; for mild overheating, power reduction is used to suppress heat accumulation without interrupting charging, which avoids the risk of escalation and maximizes the user's charging experience; for severe overheating, charging is interrupted to completely cut off the heat source and prevent serious equipment failures such as charging gun burnout and circuit short circuit.

[0078] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, and of course, it can also be implemented by special hardware including application-specific integrated circuits, special CPUs, special memory, special components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of this application.

[0079] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive (SSD)).

[0080] Finally, it should be noted that the above are merely preferred embodiments of the present invention, used only to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A multi-source data-based intelligent charging gun adaptive overheat protection method, characterized in that, Includes the following steps: S1. Real-time acquisition of charging gun thermal management data, including charging gun interface temperature sequence, cable temperature sequence, ambient temperature sequence, charging current sequence and voltage sequence, and processing of the charging gun thermal management data to obtain a smoothed sequence of the charging gun thermal management data. S2. The smoothed sequence is extracted using a sliding window, and the current change slope, interface temperature rise rate, cable temperature rise rate, number of instantaneous voltage drops, ambient temperature change slope, dynamic temperature gradient, and temperature coupling strength are calculated based on the extracted smoothed sequence. S3. Calculate the cumulative thermal anomaly score for the current monitoring period based on the current change slope, interface temperature rise rate, cable temperature rise rate, and number of instantaneous voltage drops. S4. Correct the heat accumulation anomaly score according to the slope of the ambient temperature change and the preset heat dissipation conditions to obtain the corrected heat accumulation anomaly score. S5. Input the corrected thermal accumulation anomaly score, dynamic temperature gradient and temperature coupling strength into the fuzzy inference model to infer the risk level; S6. Combine the risk levels corresponding to the current monitoring cycle and the historical monitoring cycles within the preset time window into a risk evolution sequence, and calculate its similarity with each mode in the preset overheating mode library to obtain a similarity vector. S7. Determine the overheating mode corresponding to the maximum value in the similarity vector; and determine the adaptive protection threshold based on the overheating mode; S8. Compare the corrected thermal accumulation anomaly score with the adaptive protection threshold, and determine the charging gun protection strategy based on the comparison result.

2. The method for intelligent charging gun adaptive overheat protection based on multi-source data according to claim 1, wherein, S1 specifically includes: S101. Real-time acquisition of charging gun thermal management data, including charging gun interface temperature sequence, cable temperature sequence, ambient temperature sequence, charging current sequence, and voltage sequence. S102. Perform low-pass filtering on the charging gun thermal management data to obtain a smoothed sequence of the charging gun thermal management data.

3. The method of claim 1, wherein, S2 specifically includes: S201. Extract the smooth sequence according to the preset length sliding window, and calculate the current change slope, interface temperature rise rate, cable temperature rise rate, voltage instantaneous drop number, and ambient temperature change slope according to the extracted smooth sequence. S202. Calculate the dynamic temperature gradient based on the average values ​​of the interface temperature sequence and the cable temperature sequence, and calculate the temperature coupling strength using the Pearson correlation coefficient of the charging gun interface temperature sequence and the cable temperature sequence.

4. The intelligent charging gun adaptive overheat protection method based on multi-source data as described in claim 1, characterized in that, Specifically, S3 includes: S301, Pre-trained isolated forest model; S302. Input the current change slope, interface temperature rise rate, cable temperature rise rate, and voltage instantaneous drop count into the isolated forest model to calculate the thermal accumulation anomaly score at the current moment.

5. The method of claim 4, wherein, S4 specifically includes: S401. Based on the slope of the ambient temperature change and the preset heat dissipation condition language value, determine the correction coefficient through the mapping relationship between the pre-stored ambient temperature change slope, heat dissipation condition language value and correction coefficient. S402. Based on the correction coefficient, the heat accumulation anomaly score is proportionally corrected to obtain the corrected heat accumulation anomaly score.

6. The method of claim 1, wherein, S6 specifically includes: S601. Combine the risk levels corresponding to the current monitoring cycle and the historical monitoring cycles within the preset time window into a risk evolution sequence. S602. Calculate the similarity between it and each mode template in the preset overheating mode library using the dynamic time warping algorithm to obtain the similarity vector.

7. The method of claim 6, wherein, Specifically, S7 includes: S701. Determine the overheating mode corresponding to the maximum value in the similarity vector; S702. Read the trigger margin coefficient corresponding to the overheating mode from the preset protection margin mapping table, and adjust the preset abnormal protection threshold according to the trigger margin coefficient to obtain the final adaptive protection threshold.

8. An intelligent charging gun adaptive overheat protection system based on multi-source data, characterized in that, include: The acquisition module is used to acquire charging gun thermal management data in real time. The charging gun thermal management data includes charging gun interface temperature sequence, cable temperature sequence, ambient temperature sequence, charging current sequence and voltage sequence. The module processes the charging gun thermal management data to obtain a smoothed sequence of the charging gun thermal management data. The first calculation module is used to extract the smoothed sequence according to a preset length sliding window, and calculate the current change slope, interface temperature rise rate, cable temperature rise rate, voltage instantaneous drop number, ambient temperature change slope, dynamic temperature gradient and temperature coupling strength according to the extracted smoothed sequence. The second calculation module is used to calculate the thermal accumulation anomaly score for the current monitoring period based on the current change slope, interface temperature rise rate, cable temperature rise rate, and number of instantaneous voltage drops. The correction module is used to correct the heat accumulation anomaly score according to the slope of the ambient temperature change and the preset heat dissipation conditions, so as to obtain the corrected heat accumulation anomaly score. The inference module is used to input the corrected thermal accumulation anomaly score, dynamic temperature gradient, and temperature coupling strength into the fuzzy inference model to infer the risk level. The third calculation module is used to form a risk evolution sequence by combining the risk levels corresponding to the current monitoring cycle and the historical monitoring cycles within the preset time window, and to calculate the similarity between the sequence and each mode in the preset overheating mode library to obtain a similarity vector. A determination module is used to determine the overheating mode corresponding to the maximum value in the similarity vector; and to determine an adaptive protection threshold based on the overheating mode. The output module is used to compare the corrected thermal accumulation anomaly score with the adaptive protection threshold, and determine the charging gun protection strategy based on the comparison result.

9. The intelligent charging gun adaptive overheat protection system based on multi-source data of claim 8, wherein, The acquisition module is specifically used for: Real-time acquisition of charging gun thermal management data, including charging gun interface temperature sequence, cable temperature sequence, ambient temperature sequence, charging current sequence, and voltage sequence; The thermal management data of the charging gun is subjected to low-pass filtering to obtain a smoothed sequence of the thermal management data of the charging gun.

10. The intelligent, multi-source data based, gun adaptive overheat protection system of claim 8, wherein, The first calculation module is specifically used for: The smoothed sequence is extracted according to a preset length sliding window, and the current change slope, interface temperature rise rate, cable temperature rise rate, voltage instantaneous drop number, and ambient temperature change slope are calculated based on the extracted smoothed sequence. The dynamic temperature gradient is calculated based on the average values ​​of the charging gun interface temperature sequence and the cable temperature sequence, and the temperature coupling strength is calculated using the Pearson correlation coefficient of the interface temperature and cable temperature sequences.

Images