New energy automobile charging safety data processing method based on machine learning

Abstract

The invention provides a new energy automobile charging safety data processing method based on machine learning. The new energy automobile charging safety data processing method comprises the steps that a cross-session tracking unit is constructed for multiple pairs of adjacent charging sessions; from a cross-session tracking unit, a reproduction trace type physical degradation feature capable of representing accumulative degradation in the connector is extracted, and an evolution track reflecting monotone increasing of the degradation feature along with time is constructed; generating a fatigue accumulation label for quantifying the internal fatigue accumulation degree of the connector contact based on the evolution trajectory; according to the grade of the fatigue accumulation label, a differentiated advanced intervention strategy is mapped and generated; according to the method, an intervention strategy is issued to a station operation and maintenance side for execution in advance, an intervention effect is verified based on recollected charging data after the strategy is executed, if a verification result shows that degradation is not effectively inhibited, the intervention strategy is upgraded, the limitation of single session analysis is broken through, cross-session data continuous tracking is realized, and an internal accumulative degradation trend can be accurately identified.

Description

Technical Field

[0001] This disclosure relates to the field of safety monitoring and fault prediction technology for new energy vehicles, and in particular to a data processing method for charging safety of new energy vehicles based on machine learning. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the safety issues during vehicle charging have become increasingly prominent. As a core component of new energy vehicles, the safety performance of the power battery directly affects the reliability of the entire vehicle. During high-power fast charging, the various components of the charging system are subjected to extremely high electrical loads, which can easily lead to potential risks such as connector contact aging and material fatigue. The accumulation of these hidden dangers may result in serious consequences such as local overheating or even thermal runaway.

[0003] Currently, the industry's commonly used charging safety monitoring methods are primarily based on data analysis from a single charging session. This method monitors parameters such as temperature, current, and voltage in real time during charging and uses preset thresholds for safety assessment. When the monitored data exceeds safe limits, the system triggers an alert or interrupts charging. This monitoring method is effective for obvious anomalies that develop continuously during a single charging session, but it still has significant limitations in practical applications. In high-frequency fast charging scenarios, charging connector contacts suffer microscopic damage after experiencing high-current surges. During the intervals between charging sessions, the surface temperature returns to normal due to natural cooling, masking the internal degradation trend. This apparent recovery caused by cooling makes it difficult for traditional monitoring methods to identify progressive faults. Existing technologies lack cross-session data analysis capabilities and cannot capture long-term degradation signals fragmented by intervals, resulting in insufficient early warning capabilities for potential hazards. Systems often only identify anomalies when damage accumulates to a critical state, at which point the optimal intervention time has been missed, leading to significant safety risks.

[0004] Therefore, there is an urgent need for a charging safety data processing method that can overcome the limitations of single-session analysis, achieve continuous tracking of cross-session data, and accurately identify internal cumulative degradation trends. Summary of the Invention

[0005] In view of this, in order to solve the problems caused by the existing technology, this application provides a new energy vehicle charging safety data processing method based on machine learning.

[0006] In a first aspect, this disclosure provides a method for processing new energy vehicle charging safety data based on machine learning, the method comprising: Construct cross-session tracking units for multiple pairs of adjacent charging sessions; From the cross-session tracking unit, recurring trace-type physical degradation features that can characterize the cumulative degradation inside the connector are extracted, and an evolution trajectory reflecting the monotonically increasing degradation features over time is constructed. Based on the evolution trajectory, a fatigue accumulation label is generated to quantify the degree of fatigue accumulation inside the connector contacts; Based on the level of the fatigue accumulation label, a differentiated early intervention strategy is mapped and generated; The aforementioned early intervention strategy is sent to the station operation and maintenance side for execution, and the intervention effect is verified based on the charging data re-collected after the strategy is executed. If the verification results show that degradation has not been effectively suppressed, the intervention strategy is upgraded.

[0007] Optionally, the construction of a cross-session tracking unit for multiple pairs of adjacent charging sessions includes: Based on the monitoring data of each charging session, the steady-state characteristics at the end of the session, which characterize the thermal state of the connector as it tends to stabilize at the end of charging, are extracted. Pair up charging sessions that are adjacent in time and calculate the cooling interval between the two sessions and the residual temperature difference after the cooling interval. Based on the steady-state characteristics of the session end, the cooling interval, and the residual temperature difference, a tracking unit for cross-session performance evolution analysis is constructed.

[0008] Optionally, the residual temperature difference is obtained by the difference between the average port temperature and the average ambient temperature at the start of the next charging session.

[0009] Optionally, the extraction of recurring trace-type physical degradation features capable of characterizing cumulative degradation within the connector includes: The residual heat amplitude is obtained based on the residual temperature difference; The impact amplitude between charging sessions is calculated based on the peak current data of adjacent charging sessions, serving as a characterization of mechanical impact. By combining the thermal residual amplitude and the mechanical impact amplitude, a one-time degradation trace is formed that reflects the amount of irreversible degradation within a single session cycle.

[0010] Optionally, the construction of the evolution trajectory reflecting the monotonically increasing degradation features over time is achieved by recursively accumulating the sequence of one-time degradation traces arranged in chronological order.

[0011] Optionally, generating fatigue accumulation labels for quantifying the degree of fatigue accumulation within the connector contacts includes: Analyze the growth trend of the evolutionary trajectory to obtain indicators of growth rate and trend persistence; By combining the aforementioned growth rate and trend persistence indicators, a trend strength characteristic is formed; Based on the trend intensity characteristics and a preset dynamic threshold, a graded fatigue accumulation label is generated.

[0012] Optionally, mapping and generating differentiated early intervention strategies based on the level of the fatigue accumulation label includes: Based on the pre-established strategy knowledge base, the fatigue accumulation tags of mild, moderate and high risk levels are mapped to standardized operation and maintenance instructions for basic maintenance registration, on-site inspection and mandatory component replacement, respectively.

[0013] Optionally, the intervention effect is verified based on the charging data re-collected after the strategy is executed. If the verification results show that degradation has not been effectively suppressed, the intervention strategy is upgraded. After the early intervention strategy is implemented, at least one complete charging cycle is re-monitored to obtain new performance data; Based on the new performance data, the cumulative degradation of the connector is recalculated; The effectiveness of the strategy is evaluated by comparing the cumulative degradation before and after the intervention; If the assessment results indicate that the degradation has not been effectively suppressed, a higher level of intervention strategy will be automatically triggered.

[0014] In a second aspect, this disclosure provides an electronic device including a memory and at least one processor, the memory storing a computer program, and the processor executing the computer program to implement the method of the first aspect described above.

[0015] Thirdly, this disclosure provides a computer storage medium storing a computer program that, when executed, implements the method described in the first aspect.

[0016] The beneficial effects of this disclosure are that, compared with the prior art, this disclosure has the following advantages: 1) By constructing charging tracking unit data fragments across sessions and retaining cooling interval history, an analysis framework covering multiple consecutive charging cycles was established. This framework can effectively correlate performance degradation signals that are interrupted by the natural cooling process, enabling continuous tracking of the cumulative fatigue evolution inside the connector and providing a complete data foundation for early identification of progressive faults.

[0017] 2) A method for extracting physical degradation features of reproducible traces is proposed. By extracting features such as residual temperature difference and dynamic impact amplitude from cross-session data and constructing a monotonically increasing evolution trajectory, the latent degradation trend that is covered by apparent restoration can be made explicit, which significantly improves the ability to detect early hidden dangers.

[0018] 3) By converting the quantified degradation trajectory into fatigue accumulation labels and generating graded intervention strategies accordingly, a precise mapping from condition diagnosis to maintenance actions is achieved. Differentiated operation and maintenance instructions can be triggered based on the actual health status of the connector, realizing true predictive maintenance and effectively avoiding the phenomenon of insufficient or excessive maintenance.

[0019] 4) By establishing a complete process management mechanism from strategy generation to on-site execution and effect verification, it ensures that every identified defect can be transformed into traceable and verifiable operation and maintenance actions, effectively advancing charging safety management from the algorithm level to the physical level of problem solving, and significantly improving the reliability and safety of charging system operation. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0021] Figure 1 A flowchart of a machine learning-based data processing method for new energy vehicle charging safety provided in an embodiment of this disclosure is shown.

[0022] The accompanying drawings have illustrated specific embodiments of this disclosure, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this disclosure to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0023] The present disclosure will be further described below with reference to the accompanying drawings. The following embodiments are only used to illustrate the technical solutions of the present disclosure more clearly, and should not be used to limit the scope of protection of the present disclosure.

[0024] The components of the embodiments of the invention described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0025] In the following, the terms “comprising,” “having,” and their cognates, which may be used in various embodiments of the invention, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as excluding, firstly, the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations thereof.

[0026] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0027] In fast-charging scenarios for new energy vehicles, the performance degradation of charging connectors is often not immediately apparent during a single charging session, but rather accumulates gradually over multiple charging cycles. Especially in high-frequency fast-charging stations, due to the short charging intervals, the minor damage to connector contacts after experiencing high-current surges is masked by the recovery of surface temperature during natural cooling. To overcome this technical challenge, a data analysis framework capable of spanning a single charging session is needed to capture the long-term degradation trends interrupted by the cooling process.

[0028] Figure 1 A flowchart of the machine learning-based new energy vehicle charging safety data processing method provided in this disclosure is shown below. Figure 1 As shown, the process may include the following steps: S1: Construct cross-session tracking units for multiple pairs of adjacent charging sessions.

[0029] By systematically processing monitoring data during the charging process, charging trigger segments across sessions are established and cooling interval history is retained, providing a complete data foundation for subsequent analysis.

[0030] S1.1: Charging session slice and calibration.

[0031] In this implementation, monitoring data of the charging process needs to be obtained from the vehicle battery management system and the charging pile monitoring system, and each charging event needs to be independently classified and identified. The data to be collected includes information from multiple dimensions: Timestamp data: Records the precise time points of each key operation during the charging process, including the time of plugging in the charging gun, the time of charging start, the time of charging end, and the time of plugging out the charging gun. These time points are synchronized through the CAN bus communication protocol or Ethernet protocol between the vehicle and the charging station to ensure time consistency between different systems; Insertion and removal trigger indicators: These are digital signals generated by a mechanical microswitch or a Hall effect magnetic field sensor on the charging connector. When the charging gun is inserted into place, the mechanical switch triggers a closing signal; when the charging gun is fully inserted, the magnetic field sensor detects a change in the magnet's position and generates a trigger signal. These indicator signals are acquired through a digital input channel and have distinct rising and falling edge characteristics, accurately identifying the start and end of the charging process. Charging current sequence: Acquired by a closed-loop Hall effect sensor built into the charging pile, with a sampling frequency of 1Hz, a measurement range typically of 0-500A, and an accuracy of ±1%. This sequence fully records the current changes during the charging process, including the current rise during the start-up phase, the maximum current value during the steady-state charging phase, and the current drop during the end phase, reflecting the real-time intensity of the charging load. Port temperature sequence: acquired via a K-type thermocouple or digital temperature sensor installed near the charging gun contacts, with a sampling frequency of 1Hz, a measurement range of -40℃ to 100℃, and an accuracy of ±0.5℃. The sensor is directly attached to the metal contact surface via thermally conductive silicone, enabling accurate monitoring of temperature changes during connector operation, including abnormal temperature rises caused by contact oxidation and heat generation due to increased contact resistance. Insulation status sequence: This sequence is derived by monitoring the insulation resistance in the charging circuit. The voltage-current method is used for measurement, with a test voltage of 500VDC, a measurement range of 0-10MΩ, and a sampling frequency of 1Hz. This sequence reflects the electrical safety status of the charging system. When the insulation resistance value is below the safety threshold (typically 1MΩ), it indicates a risk of leakage current and is an important parameter for assessing charging safety. Contact pressure sequence: Acquired by micro-piezoresistive sensors embedded inside the charging gun. The sensors are distributed at 6 key locations around the contacts, with a sampling frequency of 1Hz, a measurement range of 0-20N, and an accuracy of ±0.1N. The sensors employ the Wheatstone bridge principle, enabling precise measurement of the contact pressure distribution during the mating process, and assessment of the stability and reliability of the mechanical connection. Ambient temperature sequence: Acquired by a temperature and humidity sensor installed inside the charging station, with a sampling frequency of 0.1Hz, a measurement range of -30℃ to 50℃, and an accuracy of ±0.3℃. This sequence provides a reference temperature of the surrounding environment during charging, used to analyze the natural heat exchange process between the connector and the environment, and to provide an environmental benchmark for the analysis of temperature changes.

[0032] Based on this monitoring data, the system defines the start time of the k-th charging session. The time point at which the gun-triggered signal is first detected, and the end time. This is the time point when the trigger signal for drawing the gun is first detected. The data segment of the k-th charging session is then constructed from this point. This segment includes the start time of the charging session, the end time of the charging session, and the current sequence collected within this time interval. Port temperature sequence Insulation state sequence Contact pressure sequence and ambient temperature sequence Where time t belongs to the interval This process solidifies each insertion and removal of the gun into an independent analysis unit, avoiding data fragmentation caused by intermittent cooling and establishing the infrastructure for subsequent cross-session correlation analysis.

[0033] S1.2: Obtain the steady-state characteristics of the connector at the end of each charge to provide a reference benchmark for evaluating the effectiveness of the subsequent cooling process.

[0034] As the charging process nears its end, the connector's thermal state tends to stabilize due to the gradual decrease in current. The system selects a time window before the session ends; this duration parameter... Based on the charging power and connector thermal capacity characteristics, a recommended value range is 5 to 15 seconds. The specific value of this parameter is negatively correlated with the charging power level; for high-power scenarios, the lower limit of the range is used to capture the end-of-session transient, while for low-power scenarios, the upper limit is used to obtain more stable steady-state characteristics. Within this time window, the average value of the calculated current is used as the characteristic current value at the end of the session. This parameter reflects the load intensity of the battery management system when it reduces load to the end of the charging process, and is calculated using the following formula: Simultaneously, the average port temperature is calculated as the characteristic temperature value. This temperature value characterizes the peak operating temperature reached and maintained by the connector contacts during the current session before the charging load is removed. The calculation formula is as follows: In addition, the average contact pressure is calculated as the characteristic pressure value. This parameter reflects the stability of the mechanical connection state, and the calculation formula is: Through these calculations, the terminal feature vector is obtained. This vector fully characterizes the thermal boundary state of the connector at the end of the current charging session, providing an accurate starting state reference for subsequent analysis of the recovery effect of the cooling process, making cross-session performance comparison possible, while reducing noise interference through steady-state extraction and ensuring the reliability of feature data.

[0035] S1.3: Establish the correlation between adjacent charging events and quantify the impact of the cooling process.

[0036] The system sorts all charging sessions in chronological order and establishes a correspondence between two adjacent sessions. Calculate the time interval between two sessions. This parameter directly reflects the cooling time that the connector can achieve between two charging cycles, i.e., the cooling interval. The time interval is calculated based on precise timestamp data, ensuring the accuracy of the timing relationship. In the initial stage before the next charge begins, the system selects a time period... As the initial observation window, this parameter is recommended to be between 3 and 10 seconds to capture the initial state of the connector at the start of charging. The determination of this parameter needs to balance the requirements of measurement accuracy and response speed, and is usually set based on the connector's thermal response rate. Within this window period, the average port temperature is calculated. and the average ambient temperature Based on these two temperature values, the residual temperature difference was calculated. ,in, This represents the average port temperature at the start of the next session. This is the average ambient temperature at the start of the next session. This parameter... This reflects the amount of heat remaining inside the connector after the cooling interval. In the high-frequency fast charging scenario of new energy vehicles, the shorter charging interval leads to a residual temperature difference. This becomes a key indicator that the internal heat of the connector cannot be completely dissipated within a limited cooling time, and is an early and sensitive signal of cumulative performance degradation. Through this analysis process, the system can explicitly characterize the complete physical process of removing, cooling, and reinserting the connector, effectively capturing residual thermal signals masked by the self-cooling effect, and providing key input for subsequent degradation analysis.

[0037] S1.4: Build a complete tracking unit for each pair of adjacent charging sessions to support long-term performance evolution analysis.

[0038] For each pair Generate tracking unit This unit contains the session number. and Terminal characteristics Cooling interval Residual temperature difference and session load identifier Among them, session load identifier The peak current and key timestamps from the two sessions were combined and defined as follows: Peak current The calculation formula is obtained by analyzing the current sequence throughout the entire charging session. This parameter reflects the maximum electrical stress the connector experiences during this charging process and is an important indicator for assessing performance degradation. All tracking units established in adjacent sessions are organized into an ordered sequence to form a global index table. This index table establishes a complete cross-session data analysis framework, integrating the previous ending state, cooling effect, and next starting state into a unified traceable unit. This data organization not only ensures the temporal continuity of the data but also provides a standardized input format for subsequent fatigue evolution assessment, enabling machine learning algorithms to effectively extract long-term degradation features from this data, thereby overcoming the limitations of single-session analysis.

[0039] S1.5: Systematically store and manage the important parameters used in the analysis process and the generated datasets to ensure the repeatability and traceability of data analysis.

[0040] System settings for steady-state window duration Its value ranges from 5 to 15 seconds. The selection of this parameter is based on the thermal time constant of the connector to ensure the accuracy of steady-state feature extraction; the initial window duration... The value range is 3 to 10 seconds, and determining this parameter requires balancing measurement accuracy and response speed. The lower and upper limits of the cooling interval threshold are set to 30 to 900 seconds. These thresholds are based on the statistical distribution of data in actual operating scenarios to ensure that only representative charging intervals are analyzed. These parameters together constitute the parameter set. Synchronize and archive the cross-session tracking unit sequence L with the parameter set Q to define a complete archive. This process not only solidifies reproducible scenarios and adjustable ranges but also ensures that subsequent steps perform cross-session comparisons under the same parameter benchmark, avoiding analytical biases caused by parameter inconsistencies. Through systematic archive management, a complete data foundation is provided for subsequent data mining and model optimization, while also supporting reliability verification for long-term trend analysis and fault early warning.

[0041] The technical solution of this disclosure systematically solves the inherent limitations of single-charge session data analysis by constructing cross-session charging trigger segments and retaining cooling interval history. Its beneficial effect lies in establishing a data foundation covering the complete charge-discharge cycle and intermittent cooling process, linking the end state of the previous charge, the natural cooling effect, and the start state of the next charge into a unified traceable unit. This data organization method effectively overcomes the problem of performance degradation signals caused by cooling being fragmented and masked, providing continuous and structured data support for subsequently capturing the long-term cumulative degradation trend within the connector.

[0042] S2: Extract recurring trace-type physical degradation features that can characterize the cumulative degradation inside the connector from the cross-session tracking unit, and construct an evolution trajectory that reflects the monotonically increasing degradation features over time.

[0043] From the constructed cross-session data, physical features that can penetrate the cooling masking effect and reflect the cumulative degradation inside the connector are extracted. Traditional monitoring methods rely on continuous abnormal signals during a single charging process, but in high-frequency fast charging scenarios, such abnormalities are often masked by the cooling process. This step establishes a feature index system that can characterize the fatigue accumulation inside the connector by analyzing recurring degradation traces in the cross-session data, providing a reliable basis for subsequent fault early warning.

[0044] S2.1: Analyze the thermally related degradation traces using the residual temperature difference data and terminal temperature characteristics obtained in step S1.

[0045] The system extracts each tracking unit from the cross-session data archive D. Included residual temperature difference This residual temperature difference is defined as the residual heat amplitude. Its calculation formula is Residual heat amplitude The physical significance lies in the fact that it quantifies the accumulated heat inside the connector that failed to be completely dissipated through the cooling process. When Larger values ​​indicate that although the connector surface temperature has recovered during cooling, significant heat accumulation remains inside, which is a crucial precursor to material fatigue and performance degradation. The system organizes the residual thermal amplitude of each tracking unit into a set of residual thermal characteristics. This set constitutes the foundational dataset for thermal fatigue analysis. In practical applications, this parameter can effectively identify connector degradation cases that perform normally during a single charge but experience gradual and escalating heat accumulation over long-term operation.

[0046] S2.2: Based on the peak current variation characteristics between adjacent charging sessions, analyze the degradation traces related to mechanical stress.

[0047] The system extracts each tracking unit from the cross-session data archive D. Includes peak current data from two consecutive charging sessions, including the peak current from the previous session. and the peak current of the next session The peak current is obtained by analyzing the current sequence throughout the entire charging session, and the calculation formula is as follows: This reflects the maximum electrical load the connector withstands during a single charging cycle. Based on these two peak current values, the surge amplitude between adjacent sessions is calculated. The calculation formula is: The physical significance of this parameter lies in quantifying the intensity of the current step change experienced by the connector between two adjacent fast charging sessions due to differences in vehicle battery demand or charging pile output power. This drastic current change can trigger Joule thermal shock and electromechanical shock at the connector contacts. The system organizes the shock amplitude of each tracking unit into a shock amplitude set. This set reflects the dynamic mechanical stress changes experienced by the connector during cross-session use, providing an important basis for assessing the fatigue accumulation of contact materials.

[0048] S2.3: By integrating thermal fatigue characteristics with mechanical impact characteristics, a composite index can be constructed that can comprehensively reflect the degradation state of the connector.

[0049] The system will use the thermal residual characteristics obtained in S2.1 Impact amplitude characteristics obtained in S2.2 By organically combining and weighting the data, a one-time degradation trace can be formed. ,in and The weighting coefficients are determined based on the relative importance of thermal and mechanical effects in connector degradation, and are typically set to 0.6 and 0.4, respectively. The magnitude of this value directly reflects the degree of new, irreversible degradation of the connector during the k-th inter-session cycle. A larger value indicates that thermal effects and mechanical shocks have a significant impact simultaneously during this cycle, exhibiting clear initial degradation characteristics. The advantage of this composite index is that it can simultaneously capture degradation information from both thermal and mechanical dimensions, avoiding the limitations of single-dimensional analysis and providing a more comprehensive basis for accurately assessing the health status of the connector.

[0050] S2.4: Establish the long-term evolution trajectory of connector degradation, forming a continuous performance degradation curve by accumulating historical degradation traces.

[0051] The system is based on the one-time degradation trace sequence calculated in S2.3 { Evolutionary trajectories are constructed through recursive accumulation. The specific calculation formula is as follows: ,in This represents the total degradation level of the connector up to the k-th charging session. This cumulative process has a clear physical meaning: each one-time degradation trace... This represents irreversible damage incurred within a single session cycle, while the cumulative value... This reflects the total damage from the start of use to the current moment. During the cumulative calculation, the system ensures that all degradation traces are strictly aligned chronologically and employs a monotonically increasing processing method. Even if apparent performance temporarily recovers due to cooling effects in certain cycles, the cumulative degradation value still maintains an increasing trend. It still maintains a non-decreasing growth trend, that is This characteristic ensures that even if apparent performance temporarily recovers due to cooling effects in certain cycles, the accumulated irreversible degradation will not decrease, thus effectively revealing the internal long-term degradation trend. This approach effectively overcomes the limitations of traditional monitoring methods that only focus on performance changes in a single session, capturing latent faults that perform normally during short-term cooling but continuously deteriorate during long-term operation, thereby directionally accumulating the irreversible micro-damage to the connector's internal structure with each charging session. Evolutionary Trajectory This constitutes a complete record of the connector's performance degradation throughout its entire lifecycle, providing crucial data support for predicting remaining service life and developing preventative maintenance strategies. By analyzing the changing trends of this trajectory, accelerated degradation stages of the connector can be identified early, allowing for timely intervention before failure occurs and significantly improving charging safety performance.

[0052] In the technical solution of this disclosure, by extracting reproducible trace-type physical degradation features from cross-session segments, an effective characterization of the implicit degradation within the connector is achieved. Its beneficial effect lies in its ability to penetrate the apparent recovery caused by intermittent cooling, extracting key features reflecting thermal fatigue and mechanical stress accumulation from physical quantities such as residual temperature difference and peak current surges. By fusing these features and constructing a monotonically increasing evolution trajectory, irreversible micro-damage to the connector material can be directionally accumulated and made explicit, thereby providing a stable and reliable quantitative basis for early identification of progressive performance degradation.

[0053] S3: Based on the evolution trajectory, generate fatigue accumulation labels for quantifying the degree of fatigue accumulation inside the connector contacts.

[0054] The degradation evolution trajectory obtained in step S2 is transformed into fatigue accumulation tags with clear engineering significance, enabling a graded assessment of the fatigue state of the connector's internal structure. Traditional monitoring methods can only provide continuous performance parameters and cannot directly guide maintenance decisions. This step establishes a mapping relationship from quantitative data to qualitative tags, transforming the abstract degradation trajectory into actionable early warning information, providing a precise basis for subsequent preventative maintenance.

[0055] S3.1: Analyze the dynamic changes in the degradation trajectory and quantify the contribution of each charging session to the overall degradation.

[0056] The system directly utilizes the one-time degradation trace sequence obtained from step S2.3. As an additional degradation magnitude for each session cycle. Each in this sequence It directly characterizes the amount of irreversible degradation that occurs during the k-th cross-session cycle, and its magnitude directly reflects the rate of degradation within each cycle. Sequence This constitutes the differential characteristics of connector performance degradation. By analyzing the statistical features of this sequence, key inflection points of accelerated degradation can be identified, providing fundamental data support for subsequent trend analysis.

[0057] S3.2: Assess the persistence and stability of degradation trends, and distinguish between occasional fluctuations and genuine structural degradation.

[0058] Based on the obtained growth rate sequence This involves constructing a persistence indicator by comparing the development trends of adjacent periods. The specific implementation process is as follows: Define the persistence indicator. For Boolean variables, when hour, This indicates that the degradation trend remains stable or accelerates; when hour, This indicates a temporary suppression of the degradation trend. This binary division is based on the fact that true material fatigue degradation is cumulative and irreversible, while occasional performance recovery often stems from changes in environmental factors or operating conditions. In engineering practice, when... When this occurs repeatedly, it indicates that continuous damage is accumulating inside the connector; while when At that time, although the apparent performance showed a temporary improvement, the underlying internal damage was not truly restored. The system maintains a sliding window of length N and counts the data within the window. The frequency of occurrence is used to further assess the stability of the trend. This persistence indicator transforms discrete degradation magnitude data into trend judgments with clear physical meaning, providing a directional basis for subsequent grading.

[0059] S3.3: Integrate the magnitude of degradation and the persistence of the trend to construct a composite index that can comprehensively reflect the severity of degradation.

[0060] Based on the obtained growth rate sequence and persistence indicators A trend strength index is constructed through product operations. This construction method has clear engineering significance: when At any time, regardless What value should be taken? This indicates that the degradation trend is temporarily suppressed by external factors; when hour, This indicates that both trend persistence and a clear degree of degradation exist within the current session cycle, and its magnitude directly quantifies the degradation intensity of that cycle. This trend intensity feature takes into account both the directionality and severity of degradation development and is a key input parameter for generating fatigue accumulation labels.

[0061] S3.4: Complete the transformation from quantitative characteristics to engineering labels to establish an intuitive representation of the connector's health status.

[0062] Based on the obtained trend strength index Fatigue accumulation labels are generated through hierarchical mapping. The specific implementation process employs a dynamic threshold mechanism, setting three key thresholds based on the distribution characteristics of historical data. ,in These thresholds were determined by quantile statistics of historical trend intensity indices from a large number of normal and abnormal connector samples: It is usually set to the upper limit of the normal performance fluctuation range, such as the 85th percentile, to distinguish between normal aging and slight degradation. Set as a significant inflection point where degradation enters a stable accumulation phase, such as the 95th percentile; This is associated with the accelerated degradation critical point of connector material thermal fatigue, such as the 99.5th percentile, which indicates the emergence of structural risks. When At that time, the label = Mild, indicating that degradation exists but is slow, falling within the normal aging range; when At that time, the label =Moderate, indicating that degradation has entered a stable accumulation stage, requiring enhanced monitoring; when At that time, the label =Accelerated type, indicating that internal damage has formed a self-reinforcing effect, and the degradation rate is significantly accelerated; when At that time, the label =High risk indicates that the thermal degradation of the contacts has entered a structural risk stage, and even cooling cannot restore the original performance. These thresholds are not fixed values, but are dynamically adjusted based on the connector type, usage environment, and historical performance data. For example, in high-power fast charging scenarios, and The value will be reduced accordingly to improve early warning sensitivity. The system also establishes a tag credibility assessment mechanism, calculating the current tag's credibility score by analyzing the consistency of tags over the most recent M periods, thus avoiding misjudgments due to single fluctuations. The final generated fatigue accumulation tag... It provides clear and actionable input for subsequent maintenance decisions, enabling a complete process from data monitoring to engineering applications.

[0063] In the technical solution of this disclosure, by generating fatigue accumulation tags inside the contact based on the degradation trajectory, a precise mapping from continuous quantitative data to discrete engineering states is achieved. Its beneficial effect lies in transforming abstract degradation evolution values ​​into fatigue levels with clear engineering significance, such as mild, moderate, and high-risk. This transformation process comprehensively considers the magnitude and persistence of the degradation trend, forming an intuitive and operable assessment conclusion of the connector's health status. This provides a clear decision-making basis for subsequently formulating differentiated maintenance strategies, achieving standardization and objectivity in condition assessment.

[0064] S4: Based on the level of the fatigue accumulation label, map and generate differentiated early intervention strategies.

[0065] The fatigue accumulation tags generated in step S3 are transformed into specific, actionable engineering interventions, establishing a complete process from condition diagnosis to maintenance actions. Traditional maintenance strategies are often based on fixed time periods or severe fault manifestations, lacking specificity for early, latent degradation. This step, by establishing a precise mapping relationship between tags and strategies, enables accurate maintenance based on the actual health status of the connector, effectively preventing potential safety incidents.

[0066] S4.1: Develop corresponding basic maintenance measures for mild fatigue.

[0067] When fatigue accumulation label When the level is mild, it indicates that the connector has begun to degrade, but the degradation rate is relatively slow and has not yet resulted in actual structural damage. At this point, the system executes the basic maintenance registration strategy and generates a mild intervention command. The specific implementation process includes three core steps: contact status recording, maintenance number registration, and periodic review marking. First, contact status recording is performed, where the system completely inputs the current trend analysis results, key parameter values, and environmental conditions into the maintenance database, forming the initial health profile of the connector. Second, maintenance number registration is performed, assigning a unique tracking identifier to the connector and establishing a lifelong maintenance record to ensure the traceability of historical data. Finally, periodic review marking is set, dynamically calculating the shortest review cycle based on the connector's usage frequency and degradation rate, typically set to half of the normal inspection cycle. The core value of this basic maintenance strategy lies in establishing an early degradation tracking mechanism. Through systematic data recording and periodic monitoring, it provides continuous benchmark data for subsequent condition assessments. In engineering practice, this strategy achieves continuous monitoring of potential risks with minimal maintenance costs, effectively preventing minor degradation from escalating into severe conditions.

[0068] S4.2: Develop on-site maintenance plans for moderate fatigue conditions.

[0069] When fatigue accumulation label When the level is moderate, it indicates that the connector degradation has entered a stable accumulation phase, and substantial maintenance measures must be taken. The system generates a moderate intervention command. This instruction comprises four key operational steps: on-site mechanical inspection, contact resetting, contact surface cleaning, and springback retesting. On-site mechanical inspection requires maintenance personnel to open the protective structure of the charging gun head or vehicle port to conduct a comprehensive visual inspection of the internal connectors, focusing on color changes, oxidation levels, and mechanical integrity of the contact surfaces. Contact resetting requires disassembling and re-inserting the connector, using a dedicated contact resistance tester to assess whether the crimping elasticity has deteriorated, and recording the resistance changes during insertion and removal. Contact surface cleaning uses specialized electronic contact cleaners and anti-static cleaning tools to thoroughly remove oxide layers and carbon deposits from the contact surfaces, restoring the original properties of the metal surface. Springback retesting uses precision pressure sensors to measure the static contact pressure and dynamic holding force of the contacts to confirm the presence of plastic deformation trends. Implementing this on-site inspection strategy requires specialized testing tools and trained technicians to ensure that each operational step conforms to standard operating procedures. Through systematic on-site inspection, not only can existing surface defects be eliminated, but more importantly, the remaining lifespan of the connector can be accurately assessed, providing a basis for subsequent usage decisions.

[0070] S4.3: Develop mandatory replacement plans for high-risk fatigue conditions.

[0071] When fatigue accumulation label A high-risk status indicates irreversible fatigue damage to the connector's internal elastic material, with surface microcracks and thermal damage evolving at an accelerated pace, necessitating immediate replacement of the entire connector. The system generates a high-risk intervention command. This directive comprises four rigorous execution phases: mandatory contact replacement, gun head shutdown and isolation, safety locking, and resumption of use after successful re-inspection. Mandatory contact replacement requires immediate replacement of the metal elastic contact spring, crimping body, or the entire port housing, ensuring the use of original equipment manufacturer (OEM) specified replacement parts. The replacement process must utilize specialized installation tools and torque control equipment. Gun head shutdown and isolation temporarily removes the charging gun or vehicle charging port from the service network and marks it as unavailable in the management system to prevent continued use from escalating the accident. The safety locking phase employs both physical locks and electronic access control to ensure that power cannot be accidentally switched on before replacement is complete, eliminating safety hazards. Resumption of use after successful re-inspection requires comprehensive performance testing after replacement, including contact resistance measurement, temperature rise testing, and insulation performance testing. Only after all indicators pass can the device be put back into operation. The core of this mandatory replacement strategy is to eliminate major safety hazards through rapid and decisive action, while ensuring repair quality through rigorous testing and acceptance. In engineering implementation, this strategy requires establishing a standard emergency response procedure to ensure that appropriate procedures can be initiated immediately upon identification of a high-risk situation.

[0072] S4.4: Establish a complete strategy scheduling mechanism to ensure the orderly implementation of intervention measures at all levels.

[0073] The system is based on fatigue accumulation labels The level determination is used to generate the final intervention strategy through a conditional branching structure. The specific mapping relationship is as follows: when When it is mild, ;when When it is moderate, ;when When it is high-risk, This mapping is not a simple conditional judgment, but rather a real-time binding model that establishes rigid contingency plans. The system maintains a strategy knowledge base, which stores standard operating procedures, required tool lists, personnel qualification requirements, and safety precautions for different scenarios. When the system identifies a specific fatigue level, it automatically retrieves the corresponding complete work plan from the knowledge base, including specific operating steps, quality acceptance standards, and emergency plans. Furthermore, the system establishes a priority mechanism for strategy execution, implementing the highest priority response for high-risk states to ensure that processing procedures are initiated in the shortest possible time after identification. The core value of this hierarchical mapping management lies in achieving automated transformation from condition diagnosis to maintenance action, eliminating the subjectivity and delays of human judgment, and ensuring that every identified fault symptom is handled promptly and appropriately. Through standardized strategy mapping and rigid execution requirements, predictive maintenance is truly transformed into engineering practice.

[0074] In the technical solution of this disclosure, an early intervention strategy is generated based on fatigue accumulation tags, achieving a precise conversion from condition diagnosis to maintenance actions. Its beneficial effect lies in establishing a graded maintenance plan rigidly bound to fatigue levels, matching differentiated measures such as basic registration, on-site inspection, or mandatory component replacement to different degrees of degradation. This mapping relationship ensures that every identified fault symptom triggers specific, executable engineering instructions, thereby transforming predictive warnings into proactive maintenance control, effectively avoiding insufficient or excessive maintenance, and improving the targeting and timeliness of operation and maintenance management.

[0075] S5: The early intervention strategy is sent to the station operation and maintenance side for execution, and the intervention effect is verified based on the charging data re-collected after the strategy is executed. If the verification result shows that the degradation has not been effectively suppressed, the intervention strategy is upgraded.

[0076] Establish a complete engineering process from strategy generation to on-site execution and effect verification to ensure that every identified fatigue degradation problem is effectively addressed. In traditional operation and maintenance models, detection and maintenance are often disconnected, resulting in early warning information not being promptly translated into actual action. This step, by constructing a full-process management system for strategy execution, process monitoring, and effect evaluation, achieves a complete mapping from the digital space to the physical world, truly advancing safety measures from the diagnostic layer to the maintenance control layer.

[0077] S5.1: Solve the problem of rapid policy issuance and task allocation, ensuring that intervention instructions can be transmitted to the execution terminal without loss.

[0078] The final intervention strategy generated in step S4 Automatically generate standardized scheduling instruction sets. The implementation process includes three key steps: First, instruction encoding conversion transforms abstract strategy descriptions into standardized work orders that the site maintenance system can recognize and process. These work orders include specific operational items, technical standards, safety requirements, and completion deadlines. Second, a task priority assessment mechanism is established, automatically setting execution priorities based on fatigue level, with high-risk work orders receiving the highest priority to ensure they are issued to the maintenance terminal within 15 minutes of identification. Finally, intelligent resource matching is implemented, with the system automatically allocating necessary technical personnel, spare parts, and specialized tools based on the work order content, and generating detailed task lists and operation guides. The core advantage of this scheduling mechanism is the elimination of manual approval in traditional maintenance. Through a pre-set rule engine, strategies are automatically distributed, significantly shortening the response time from problem discovery to initiation of processing. In actual operation, the system also establishes an instruction confirmation mechanism, requiring maintenance personnel to confirm receipt of instructions within a specified time to ensure that every task is effectively undertaken.

[0079] S5.2: Ensure that the strategy is executed accurately and flawlessly on-site, and digitize the entire execution process.

[0080] The system is based on a scheduling instruction set. The system guides maintenance personnel in completing specific on-site operations and collects execution process data in real time. The implementation process covers three main aspects: Standardized execution actions require maintenance personnel to strictly follow standard operating procedures. For replacement operations, specialized tools such as torque wrenches must be used and key parameters recorded. For cleaning operations, original equipment manufacturer (OEM) certified cleaning agents and tools must be specified. Accurate execution time recording utilizes the system's automatic timestamp function to accurately mark the start and end times of each key operation node, providing a time benchmark for subsequent performance analysis. Real-time updates of execution completion status require maintenance personnel to immediately confirm the status on their mobile terminals after completing each sub-task. The system automatically verifies the completeness and compliance of the operation. To ensure execution quality, the system also introduces a process supervision mechanism. For high-risk maintenance tasks, dual confirmation and photographic documentation are required, and key steps must be reviewed and signed by the on-site supervisor. All execution process data is uploaded to a central database in real time, forming a complete electronic maintenance file. This data includes specific information such as the batch number of spare parts used, the torque value of the operation, and the cleaning process parameters. This meticulous process management not only ensures the consistency of maintenance work quality, but more importantly, it establishes a full-chain data traceability system, providing rich basic data for subsequent effect analysis and strategy optimization.

[0081] S5.3: Evaluate the actual effectiveness of maintenance measures through quantitative indicators to verify the effectiveness of intervention strategies.

[0082] After maintenance is completed, the system restarts the monitoring program to collect performance data of the connector in its new state. Specifically, data acquisition begins during the first full charging cycle after maintenance, using the same sensor configuration and sampling frequency as in the fault diagnosis phase, and continuously monitors for at least three full charging sessions. Based on this new monitoring data, the system recalculates the connector's key performance indicators, particularly cumulative degradation. By comparing data before and after maintenance, recovery assessment indicators are calculated. If the intervention strategy implemented is mandatory component replacement, then the cumulative degradation after maintenance... Start calculating from zero; otherwise, Foundation before maintenance The calculation continues. Finally, ,in This indicates the cumulative degradation status before the intervention strategy was implemented. This indicates the level of degradation re-observed after the intervention strategy has been implemented. The scientific basis of this indicator lies in its ability to accurately quantify the actual inhibitory effect of maintenance measures on the degradation trend: given the cumulative degradation... The monotonic non-decreasing property: When the intervention strategy implemented is mandatory component replacement, Starting from scratch, therefore This indicates that the degradation has been completely reset; when other maintenance strategies are executed, if This indicates that maintenance measures completely suppressed degradation during the observation period. No growth; if This indicates that degradation is still ongoing and maintenance measures are not effective. The magnitude of the value directly reflects the rate of degradation and development.

[0083] The system also establishes a multi-dimensional evaluation framework, which, in addition to the core degradation index, includes auxiliary indicators such as contact resistance change rate and temperature rise slope correction, ensuring the comprehensiveness and reliability of the evaluation results. The value of this evaluation mechanism lies in transforming subjective experience-based judgments into objective data analysis, providing a scientific basis for the continuous optimization of maintenance strategies.

[0084] S5.4: Make the final decision on the entire management process and determine the direction of subsequent actions based on the results of the effectiveness evaluation.

[0085] The system is based on recovery assessment indicators Quantitative analysis automatically generates disposal judgment results. The specific decision-making logic is as follows: when At that time, the system determines =Process complete, indicating that the current intervention measures have effectively resolved the problem, and the connector can be restored to normal monitoring status; when At that time, the system determines An escalation intervention indicates that existing maintenance measures have failed to effectively curb the degradation trend, necessitating the immediate implementation of a higher-level handling plan. In cases requiring escalation intervention, the system automatically triggers an emergency response mechanism, first escalating the connector status to a high-risk level and implementing mandatory shutdown measures. Simultaneously, an upgraded maintenance plan is generated, adding more thorough overhaul items to the original measures, such as connector assembly replacement and preventative maintenance of adjacent components. Furthermore, the system initiates a root cause analysis program to delve into the specific reasons for the failure of routine maintenance measures, providing improvement directions for subsequent strategy optimization. The core value of this handling and judgment mechanism lies in establishing a virtuous cycle of continuous improvement. Through quantitative evaluation of each maintenance effect, the accuracy and effectiveness of intervention strategies are continuously calibrated and optimized. In practical engineering applications, the system regularly analyzes the success rate of all maintenance cases, continuously improving the plan settings and parameter settings in the strategy knowledge base, driving the entire operation and maintenance system towards a more precise and efficient direction.

[0086] In the technical solution of this disclosure, a complete engineering control process from strategy generation to execution verification is formed by reversing the early intervention strategy to the site operation and maintenance side. Its beneficial effects include the lossless distribution of maintenance strategies to standardized work orders, digital recording of the execution process, and quantitative evaluation of the post-maintenance status. By comparing the changes in performance indicators before and after intervention, the system can objectively determine the effectiveness of maintenance measures and automatically trigger strategy upgrades for unresolved issues. This closed-loop mechanism ensures that all identified defects can be effectively resolved and tracked, ultimately advancing charging safety management from the diagnostic level to the physical level of problem solving.

[0087] To verify the practical effectiveness of this disclosed method, an application verification was conducted at a public fast-charging station in a certain city. This station is equipped with 160kW dual-gun DC fast-charging piles, and one of the frequently used charging guns was selected as the implementation target. During implementation, the parameters were set as follows: steady-state window duration... Set the duration to 10 seconds, the initial window duration. Take 5 seconds, and set the degradation feature weight coefficient according to the typical configuration. The system continuously collected and analyzed 12 valid cross-session tracking units of the charging gun over two weeks.

[0088] Using the method disclosed herein, the system successfully constructed the connector degradation trajectory. Analysis shows that the trajectory exhibits a monotonically increasing trend. At the 10th tracking unit, the trend strength index... Exceeding the preset moderate fatigue threshold The system automatically generated a moderate fatigue accumulation tag, triggering standardized on-site maintenance instructions. Site maintenance personnel followed the instructions to clean and reset the charging gun contacts. After the intervention strategy was implemented, the system continued to monitor the subsequent three charging sessions. Data showed that the recalculated cumulative degradation amount... The growth rate has leveled off, validating the effectiveness of this on-site maintenance. This embodiment demonstrates that the method disclosed herein can effectively identify early cumulative degradation of connectors in actual fast-charging operation scenarios and successfully trigger precise predictive maintenance, thus avoiding potential safety hazards.

[0089] In summary, this invention establishes an analytical framework covering multiple consecutive charging cycles by constructing cross-session charging tracking unit data fragments and retaining cooling interval history. This framework effectively correlates performance degradation signals fragmented by natural cooling processes, enabling continuous tracking of cumulative fatigue evolution within the connector and providing a complete data foundation for early identification of progressive faults. A method for extracting trace-type physical degradation features is proposed. By extracting features such as residual temperature difference and dynamic impact amplitude from cross-session data and constructing a monotonically increasing evolution trajectory, latent degradation trends masked by apparent recovery are made explicit, significantly improving the ability to detect early-stage problems. By transforming the quantified degradation trajectory into fatigue accumulation labels and generating graded intervention strategies accordingly, a precise mapping from condition diagnosis to maintenance actions is achieved. This triggers differentiated maintenance instructions based on the actual health status of the connector, realizing true predictive maintenance and effectively avoiding under- or over-maintenance. By establishing a complete process management mechanism from strategy generation to on-site execution and effect verification, it is ensured that every identified defect can be transformed into traceable and verifiable operation and maintenance actions, effectively advancing charging safety management from the algorithm level to the physical level of problem solving, and significantly improving the reliability and safety of charging system operation.

[0090] According to embodiments of this disclosure, an electronic device is also provided, which may include a processor, a communications interface, a memory, and a communication bus, wherein the processor, the communications interface, and the memory communicate with each other via the communication bus. The processor can invoke logical instructions stored in the memory to execute the methods provided by the above methods.

[0091] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0092] On the other hand, this disclosure also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the methods provided by the above methods.

[0093] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0094] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, 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 can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, 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 described in the various embodiments or some parts of the embodiments.

[0095] It should be understood that the above embodiments are only used to illustrate the technical solutions of this disclosure, and not to limit them; although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure.

Claims

1. A method for processing safety data of new energy vehicle charging based on machine learning, characterized in that, The method includes: Construct cross-session tracking units for multiple pairs of adjacent charging sessions; From the cross-session tracking unit, recurring trace-type physical degradation features that can characterize the cumulative degradation inside the connector are extracted, and an evolution trajectory reflecting the monotonically increasing degradation features over time is constructed. Based on the evolution trajectory, a fatigue accumulation label is generated to quantify the degree of fatigue accumulation inside the connector contacts; Based on the level of the fatigue accumulation label, a differentiated early intervention strategy is mapped and generated; The aforementioned early intervention strategy is sent to the station operation and maintenance side for execution. The intervention effect is verified based on the charging data re-collected after the strategy is executed. If the verification results show that degradation has not been effectively suppressed, the intervention strategy is upgraded.

2. The method for processing new energy vehicle charging safety data based on machine learning according to claim 1, characterized in that, The construction of a cross-session tracking unit for multiple pairs of adjacent charging sessions includes: Based on the monitoring data of each charging session, the steady-state characteristics at the end of the session, which characterize the thermal state of the connector as it tends to stabilize at the end of charging, are extracted. Pair up charging sessions that are adjacent in time and calculate the cooling interval between the two sessions and the residual temperature difference after the cooling interval. Based on the steady-state characteristics of the session end, the cooling interval, and the residual temperature difference, a tracking unit for cross-session performance evolution analysis is constructed.

3. The method for processing new energy vehicle charging safety data based on machine learning according to claim 2, characterized in that, The residual temperature difference is obtained by the difference between the average port temperature and the average ambient temperature at the start of the next charging session.

4. The method for processing new energy vehicle charging safety data based on machine learning according to claim 2, characterized in that, The extracted reproducible trace-type physical degradation features that characterize the cumulative degradation inside the connector include: The residual heat amplitude is obtained based on the residual temperature difference; The impact amplitude between charging sessions is calculated based on the peak current data of adjacent charging sessions, serving as a characterization of mechanical impact. By combining the thermal residual amplitude and the mechanical impact amplitude, a one-time degradation trace is formed that reflects the amount of irreversible degradation within a single session cycle.

5. The new energy vehicle charging safety data processing method based on machine learning according to claim 4, characterized in that, The construction of the evolution trajectory reflecting the monotonically increasing degradation features over time is achieved by recursively accumulating the sequence of one-time degradation traces arranged in chronological order.

6. The method for processing new energy vehicle charging safety data based on machine learning according to claim 1, characterized in that, The generation of fatigue accumulation labels for quantifying the degree of fatigue accumulation inside the connector contacts includes: Analyze the growth trend of the evolutionary trajectory to obtain indicators of growth rate and trend persistence; By combining the aforementioned growth rate and trend persistence indicators, a trend strength characteristic is formed; Based on the trend intensity characteristics and a preset dynamic threshold, a graded fatigue accumulation label is generated.

7. The method for processing new energy vehicle charging safety data based on machine learning according to claim 1, characterized in that, The step of mapping and generating differentiated early intervention strategies based on the level of the fatigue accumulation label includes: Based on the pre-established strategy knowledge base, the fatigue accumulation tags of mild, moderate and high risk levels are mapped to standardized operation and maintenance instructions for basic maintenance registration, on-site inspection and mandatory component replacement, respectively.

8. The method for processing new energy vehicle charging safety data based on machine learning according to claim 1, characterized in that, The intervention effect is verified based on the charging data re-collected after the strategy is executed. If the verification results show that degradation has not been effectively suppressed, the intervention strategy is upgraded as follows: After the early intervention strategy is implemented, at least one complete charging cycle is re-monitored to obtain new performance data; Based on the new performance data, the cumulative degradation of the connector is recalculated; The effectiveness of the strategy is evaluated by comparing the cumulative degradation before and after the intervention; If the assessment results indicate that the degradation has not been effectively suppressed, a higher level of intervention strategy will be automatically triggered.

9. An electronic device, characterized in that, The electronic device includes a memory and at least one processor. The memory stores a computer program, and the processor executes the computer program to implement the machine learning-based data processing method for charging safety of new energy vehicles as described in any one of claims 1-8.

10. A computer storage medium, characterized in that, It stores a computer program, which, when executed, implements the machine learning-based data processing method for new energy vehicle charging safety according to any one of claims 1-8.

Images