New energy vehicle charging pile safety performance life prediction method

Abstract

The invention provides a new energy vehicle charging pile safety performance life prediction method. The method comprises the steps of collecting real-time operation monitoring equipment data of a new energy vehicle charging pile; constructing a key component degradation model to evaluate the life degradation trend of the current charging pile; adopting a time decay model based on the residual life of the current charging pile; constructing a group life prediction model, and predicting the overall residual life of the charging pile group in the micro-region according to the failure risk probability of the current charging pile; evaluating a current load optimization strategy by using the micro-region environment data in combination with the dynamic load characteristics; and designing a load optimization strategy to maximize the overall life of the charging pile group in the micro-region, and generating an active maintenance plan based on the priority maintenance list. According to the invention, full-process closed-loop management from data acquisition to load modeling, life prediction and active optimization is realized, and an efficient and accurate solution is provided for safety performance and life management of the charging pile.

Description

Technical Field

[0001] The present invention belongs to the field of new energy vehicle charging piles, and in particular relates to a method for predicting the safety performance life of new energy vehicle charging piles. Background Art

[0002] As the core facilities of the electric vehicle charging network, the operation safety and life prediction of new energy vehicle charging piles have always been the focus of operators. The core function of charging piles is to provide stable and efficient charging services for new energy vehicles. However, in the long-term use process, the safety performance and life of key components of charging piles will be affected by many factors, such as power quality fluctuations, user charging behavior, environmental conditions (such as temperature and humidity), and dynamic changes in load distribution. At present, many charging pile operators adopt life management methods based on fixed thresholds or single component monitoring, which has the following defects: Dynamic load factors are not fully considered: When multiple charging piles are running together in a micro-area, the load distribution of different charging piles is usually significantly different. For example, charging piles that are used frequently may age faster due to long-term overload operation, while charging piles that are used less frequently may not fully utilize their lifespan. However, existing technologies mostly analyze the static lifespan of a single charging pile, ignoring the overall impact of dynamic load changes on the lifespan of a group of charging piles.

[0003] Insufficient modeling of degradation of key components: Key components of charging piles (such as power modules, cooling systems, and connecting cables) will degrade in performance during long-term operation due to multiple factors such as voltage fluctuations, ambient temperature and humidity. Many current life prediction methods fail to accurately model the natural degradation process of components, but instead simply evaluate them using empirical thresholds. This method makes it difficult to predict the remaining life of components and cannot accurately identify high-risk components.

[0004] Lack of active optimization mechanism: The existing charging pile maintenance mode is mostly passive, that is, repair or replacement is carried out only after the equipment fails. This not only increases the risk of unplanned downtime, but also may lead to low utilization of resources within the group.

[0005] In view of the above defects, a new method is urgently needed to organically combine dynamic load, component degradation and active optimization, provide accurate life prediction and safety performance management capabilities, and provide effective decision-making support for charging pile operators. Summary of the invention

[0006] The purpose of the present invention is to propose a method for predicting the safety performance life of a new energy vehicle charging pile, which combines the micro-area dynamic load distribution and the key component degradation model of the new energy vehicle charging pile. By introducing a new collaborative analysis and active optimization mechanism, the limitations of traditional methods are overcome.

[0007] In order to achieve the above object, the present invention provides a method for predicting the safety performance life of a new energy vehicle charging pile, the method comprising the following steps: Collect the real-time operation monitoring equipment data of the new energy vehicle charging piles, use the load correlation of the charging piles to build a micro-region load interaction graph, use a custom graph embedding method to extract dynamic load features in the micro-region load interaction graph, and obtain dynamic load features. The dynamic load features are time series matrices that represent the dynamic load status of each charging pile in the micro-region, and each row corresponds to the comprehensive features of a charging pile; wherein the real-time operation monitoring equipment data includes the load rate of the charging pile, the usage frequency of the charging pile, the load peak distribution of the charging pile, the load correlation of two random charging piles, and the micro-region environmental data; Constructing a key component degradation model based on real-time operation monitoring equipment data to evaluate the life degradation trend of the current charging pile, obtaining the unit time degradation rate of the current charging pile, and predicting the remaining life of the current charging pile based on the unit time degradation rate of the current charging pile; Based on the remaining life of the current charging pile, a time decay model is used to calculate the failure risk probability of the current charging pile, and the failure risk probability of the current charging pile is obtained; Construct a group life prediction model to predict the overall remaining life of the charging pile group in the micro-area according to the failure risk probability of the current charging pile. At the same time, generate a priority maintenance list based on the failure risk probability of the current charging pile and the remaining life of the current charging pile to ensure that the operation risk of the charging piles in the group is minimized; Use micro-regional environmental data combined with dynamic load characteristics to evaluate the current load optimization strategy, calculate the potential improvement in group life due to load optimization, and obtain the potential for group life improvement; Design a load optimization strategy to maximize the overall life of the charging pile group in the micro-area, generate an active maintenance plan based on the priority maintenance list, select the charging piles that need to replace parts, and optimize them in combination with the dynamic load. According to the optimized load distribution, design a dynamic diversion strategy to adjust the user load to the low-load equipment, and maximize the group life through gradient optimization and regularization constraints.

[0008] Furthermore, the micro-area load interaction diagram is constructed as follows: node Indicates the first Charging stations; Edge Weight Load dependency , reflects the load correlation strength between charging piles and is defined as follows: ; in, Indicates Charging stations at the time Load rate and Charging stations at the time Load rate The covariance of Indicates Charging stations at the time Load rate The standard deviation of Indicates Charging stations at the time Load rate The standard deviation of A customized graph embedding method is used to extract dynamic load features in the micro-area load interaction graph to obtain dynamic load features, including: For each node , summarizing the characteristics of its first-order neighbors: ; in, It is Charging stations at the time The first-order neighborhood aggregation characteristics of express The set of neighboring nodes; Neighbor node The feature vector of , For the Charging stations at the time Frequency of use, For the Charging stations at the time Load peak distribution; Incorporating environmental data , normalize and update all node features: ; in, Is a node The final feature representation of is the weight matrix; is the bias term; Is the activation function, used for nonlinear mapping; The resulting load characteristics It is a time series matrix, which represents the dynamic load status of each charging pile in the micro area. Each row corresponds to the comprehensive characteristics of a charging pile. The output form is: ; in, is the micro-region load characteristic matrix; It is Charging stations at the time The comprehensive load characteristics.

[0009] Furthermore, the key component degradation model is expressed as: ; in, It is The degradation rate per unit time of key components of each charging pile; is the influence function of dynamic load characteristics on degradation, through the dynamic load characteristic matrix Nonlinear combination representation of ; is the voltage square term, reflecting the stress effect of high voltage on components; is the current term, reflecting the heat loss of components caused by large current; It is a combination of micro-regional environmental data, used to quantify the impact of external conditions such as temperature and humidity; , , and is the weight of the corresponding item; is a regularization additional term used to capture the special effect of voltage and current interaction on accelerated aging of components, expressed as: ; The remaining life of the current charging pile is predicted according to the unit time degradation rate of the current charging pile, which is expressed as: ; in, The remaining life of the current charging pile; is the initial design life; is the cumulative degradation value, obtained by integrating the degradation rate formula.

[0010] Furthermore, the failure risk probability , calculated as follows: ; in, For cumulative degradation.

[0011] Furthermore, it is characterized in that the group life prediction model is expressed as: ; in, is the overall remaining life of the charging pile group in the micro area; It is The failure risk of each charging pile is Adjusting the weighting of life expectancy; is the life improvement function of dynamic load and environmental conditions, used to capture the positive effect of load optimization; is the adjustment factor, controlling the proportion of load optimization’s impact on overall life; is the total number of charging posts in the group; The priority repair list is generated based on the priority score, calculated as follows: ; in, It is The priority score of each charging station; is the load anomaly degree, capturing the abnormal deviation of load characteristics; is the abnormality weight coefficient, which controls the impact of abnormal load on priority; according to Sort from high to low to generate a priority maintenance list , where a high score represents the necessity of priority repair.

[0012] Furthermore, the life improvement function of the dynamic load and environmental conditions is The design is as follows: ; in, Load abnormality, which quantifies the part of the load characteristic that deviates from the normal range and is used to deduct the negative factors of life extension; It is the abnormal load deduction coefficient, which adjusts the weight of the impact of abnormal load on life.

[0013] Furthermore, the current load optimization strategy is evaluated using micro-regional environmental data combined with dynamic load characteristics, and the potential improvement of group life due to load optimization is calculated to obtain the group life improvement potential, which is calculated as follows: ; in, Potential improvement in group lifespan for load optimization.

[0014] Furthermore, the optimization goal of the load optimization strategy is: ; in, is the overall lifespan of the optimized group; is the current group lifespan; After optimization The load variance of each charging pile is used to quantify the load balance; It is the load balancing adjustment coefficient, which weighs the priority between load balancing and life improvement.

[0015] Furthermore, the active maintenance plan is generated based on the priority maintenance list, which is expressed as: ; in, For the The maintenance decision of each charging pile, 1 means maintenance is needed, and 0 means no maintenance is needed; The threshold ratio of the remaining life is used to determine whether the equipment with a short life needs to be replaced proactively; is the risk threshold to determine whether the equipment with a higher risk of failure needs to replace key components; Press the charging pile that meets the maintenance conditions Sort them, include them into the maintenance plan one by one, and output the final maintenance list.

[0016] Furthermore, the dynamic diversion strategy is expressed as: ; in, For optimized load distribution; is the learning rate, which is used to control the gradient adjustment step size; The potential for life improvement is related to the gradient of load distribution, guiding the direction of load optimization; is a load smoothing regularization term used to reduce load jump fluctuations and is defined as: ; in yes The set of neighbor nodes.

[0017] The beneficial technical effects of the present invention are at least as follows: In view of the shortcomings of the existing technology, the present invention can accurately identify faults in complex and changing scenarios and adopt intelligent response strategies. Its main innovations are as follows: The present invention regards multiple charging piles in a micro-area as a collaborative group, collects and models the real-time load distribution characteristics of each charging pile, and uses graph neural network (GCN) to build a dynamic load interaction model between charging piles to analyze the collaborative impact of load distribution on the life of a single charging pile in the group. This innovation solves the problem of ignoring the overall impact of dynamic load changes on the life of charging piles in existing methods, and provides a more accurate group life prediction capability.

[0018] The present invention introduces multi-dimensional monitoring data (such as current, voltage, ambient temperature and humidity, etc.), and combines it with maintenance history records, using a combination of physical drive and data drive to accurately fit the degradation curve of key components of charging piles and predict their remaining life and failure risk. Different from traditional empirical evaluation methods, this method improves the accuracy of life prediction through degradation modeling and provides operators with specific guidance for high-risk components.

[0019] Based on the prediction of group life and component degradation results, the present invention generates active maintenance plans and load control strategies through optimization algorithms. For example, by adjusting the load distribution, the pressure on high-load charging piles is reduced, and critical components are replaced in advance to avoid unplanned downtime. This innovation solves the problem of lack of active optimization mechanism in the prior art, and greatly improves the resource utilization and operational reliability of charging piles. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0021] Figure 1 This is a flow chart of a method for predicting the safety performance life of a new energy vehicle charging pile according to the present invention. DETAILED DESCRIPTION

[0022] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0023] like Figure 1 As shown, an embodiment of the present invention provides a method for predicting the safety performance life of a new energy vehicle charging pile, the method comprising: S1. Collect real-time operation monitoring equipment data of new energy vehicle charging piles, use the load correlation of charging piles to build a micro-area load interaction diagram, use a customized graph embedding method to extract dynamic load features in the micro-area load interaction diagram, and obtain dynamic load features. The dynamic load features are time series matrices, which represent the dynamic load status of each charging pile in the micro-area, and each row corresponds to the comprehensive features of a charging pile; wherein the real-time operation monitoring equipment data includes the load rate of the charging pile, the frequency of use of the charging pile, the load peak distribution of the charging pile, the load correlation of two random charging piles, and the micro-area environmental data.

[0024] Specifically, this step is the basis of the entire patent solution, which aims to provide accurate load characteristic data for subsequent life prediction and optimization. By constructing the load interaction diagram of the charging pile and extracting the dynamic load characteristics , capturing the load distribution and interaction relationship of charging piles in a micro-area. The load characteristic matrix serves as an important input for subsequent steps, reflecting the synergy and dynamic change characteristics among charging piles.

[0025] The input data includes the following parts, all of which come from the real-time operation monitoring equipment of the charging pile: : No. Charging stations at the time Load rate (ratio of charging power to maximum power); : No. Charging stations at the time Frequency of use (number of charges per unit time); : No. Charging stations at the time Load peak distribution; : No. and Two charging stations at the time The load correlation is calculated as the correlation coefficient of the load time series of the two; : Micro-area environmental data (such as multi-dimensional vectors of temperature and humidity).

[0026] The input data has been cleaned, normalized, and time-aligned through preprocessing steps to ensure data continuity and availability.

[0027] Furthermore, using the load correlation of the charging pile Constructing micro-area load interaction diagrams : node Indicates the first Charging stations; Edge Weight for , reflects the load correlation strength between charging piles and is defined as follows: ; in, express and The covariance of express The standard deviation of express Load Interaction Plot It dynamically reflects the coordination of load distribution among different charging piles and is the core basis for subsequent analysis.

[0028] Furthermore, a custom graph embedding method is used to extract dynamic load features. Graph embedding consists of the following steps: Neighborhood aggregation: For each node , summarizing the characteristics of its first-order neighborhood (i.e., directly connected charging pile nodes): ; in, It is Charging stations at the time The first-order neighborhood aggregation characteristics of express The set of neighboring nodes; Neighbor node The feature vector of .

[0029] Further, global feature update: combining environmental data , normalize and update all node features: ; in, Is a node The final feature representation of is the weight matrix; is the bias term; Is an activation function used for nonlinear mapping.

[0030] The resulting load characteristics It is a time series matrix, which represents the dynamic load status of each charging pile in the micro area. Each row corresponds to the comprehensive characteristics of a charging pile. The output form is: ; in, is the micro-region load characteristic matrix; It is Charging stations at the time The comprehensive load characteristics.

[0031] Understandably, load characteristics The dynamic load behavior data of the charging pile will be provided in the key component degradation model used in the subsequent steps as an important influencing factor for component life assessment.

[0032] S2. Construct a key component degradation model based on real-time operation monitoring equipment data to evaluate the life degradation trend of the current charging pile, obtain the unit time degradation rate of the current charging pile, and predict the remaining life of the current charging pile based on the unit time degradation rate of the current charging pile.

[0033] Among them, in order to evaluate the life degradation trend of key components of charging piles, the following targeted formula is designed to fully reflect the complex relationship between the dynamic load characteristics of charging piles and the performance attenuation of components: Degradation rate formula: ; in, For the The degradation rate per unit time of key components of each charging pile; is the influence function of dynamic load characteristics on degradation, through Nonlinear combination representation of ; is the voltage square term, reflecting the stress effect of high voltage on components; is the current term, reflecting the heat loss of components caused by large current; It is a combination of environmental data, used to quantify the impact of external conditions such as temperature and humidity; is a regularization additional term used to capture the special effect of voltage and current interaction on accelerated aging of components; ; This design takes into account the complex effects of voltage and current nonlinearity on component degradation.

[0034] Further, based on the above degradation rate formula, calculate the The remaining life of the key components of the charging pile : ; in, For the The key components at the moment remaining lifespan; is the initial design life; is the cumulative degradation value, obtained by integrating the degradation rate formula.

[0035] To ensure the reliability of calculations, the accumulated degradation value triggers dynamic adjustment every time a critical environmental change occurs (such as extreme high temperature or sudden high load) to reflect the actual life status.

[0036] S3. Based on the remaining life of the current charging pile, a time decay model is used to calculate the failure risk probability of the current charging pile to obtain the failure risk probability of the current charging pile.

[0037] Specifically, based on the remaining life , defines the failure risk probability of the component , calculated using the time decay model: ; in, For the The key components at the moment The probability of failure risk; For cumulative degradation; is the initial design life. This formula combines the calculation results of the degradation rate formula to associate the remaining life of the component with the failure probability, which is used to evaluate the failure risk in real-time operation.

[0038] S4. Construct a group life prediction model to predict the overall remaining life of the charging pile group in the micro-area according to the failure risk probability of the current charging pile. At the same time, generate a priority maintenance list according to the failure risk probability of the current charging pile and the remaining life of the current charging pile to ensure that the operation risk of the charging piles in the group is minimized.

[0039] Specifically, when predicting the life of a group of charging piles in a micro-area, it is necessary to comprehensively consider the individual life , Failure Risk The synergistic effect of dynamic load on life. The following model is proposed: Overall lifespan formula: ; in, The overall remaining life of the charging pile group in the micro area; For the The remaining life of each charging station; For the The failure risk of each charging pile is Adjusting the weighting of life expectancy; is the life improvement function of dynamic load and environmental conditions, used to capture the positive effect of load optimization; is the adjustment factor, controlling the proportion of the impact of load optimization on the overall life; The total number of charging piles in the group.

[0040] Lifespan enhancement function design: Taking into account the effects of load characteristics and environmental conditions on life, the specific definition is: ; in, For the Dynamic load characteristics of each charging pile; is the environmental condition vector; Load abnormality, which quantifies the part of the load characteristic that deviates from the normal range and is used to deduct the negative factors of life extension; The abnormal load deduction coefficient is used to adjust the weight of the abnormal load on the life. This formula dynamically adjusts the positive and negative effects of optimization in the life assessment by combining load distribution and environmental conditions.

[0041] Further, according to and , integrate dynamic load characteristics to generate a priority maintenance list , to ensure the minimization of the operating risks of charging piles within the group: Priority Rating: ; in, For the The priority score of each charging station; is the load anomaly degree, which captures the abnormal deviation of load characteristics; is the abnormality weight coefficient, which controls the impact of abnormal load on priority.

[0042] according to Sort from high to low to generate a priority maintenance list , where a high score represents the necessity of priority repair.

[0043] S5. Use micro-area environmental data combined with dynamic load characteristics to evaluate the current load optimization strategy, calculate the potential improvement of group life by load optimization, and obtain the potential for improvement of group life.

[0044] Specifically, to optimize the life of micro-area charging piles, the potential for life improvement brought by load optimization is evaluated: Potential formula: ; in, Potential improvement in group lifespan for load optimization; is the life-promoting function. By calculating , can dynamically evaluate the actual value of the current load optimization strategy and provide data support for subsequent proactive optimization and planning.

[0045] S6. Design a load optimization strategy to maximize the overall life of the charging pile group in the micro-area, generate an active maintenance plan based on the priority maintenance list, select the charging piles that need to replace parts, and optimize them in combination with the dynamic load. According to the optimized load distribution, design a dynamic diversion strategy to adjust the user load to low-load equipment, and maximize the group life through gradient optimization and regularization constraints.

[0046] Specifically, in load optimization, it is necessary to balance load balancing and reasonable allocation of equipment utilization on the basis of maximizing group life. To this end, the following innovative formula is designed: Optimization goal: maximize the overall life after optimization : ; in, is the overall lifespan of the optimized group; is the current group lifespan; Load-optimized lifespan improvement potential; After optimization The load variance of each charging pile is used to quantify the load balance; is the load balancing adjustment coefficient, which weighs the priority between balancing and lifespan improvement. In this formula, load balancing is achieved by introducing the load variance term This is done to avoid the problem of overloading of individual charging piles during the optimization process. value, so that it can both achieve a longer service life and avoid excessive uneven load distribution.

[0047] Furthermore, proactive maintenance planning is based on a priority maintenance list and equipment status, select the charging pile that needs to replace parts, and optimize it in combination with dynamic load: Maintenance decision formula: ; in, For the The maintenance decision of each charging pile, 1 means maintenance is needed, and 0 means no maintenance is needed; The threshold ratio of the remaining life is used to determine whether the equipment with a short life needs to be replaced proactively; As the risk threshold, determine whether the equipment with a high risk of failure needs to replace key components. Sort them, include them into the maintenance plan one by one, and output the final maintenance list.

[0048] Furthermore, according to the optimized load distribution, a dynamic diversion strategy is designed to adjust the user load to the low-load device: Load adjustment formula: ; in, For optimized load distribution; is the learning rate, which is used to control the gradient adjustment step size; The potential for life improvement is related to the gradient of load distribution, guiding the direction of load optimization; is a load smoothing regularization term used to reduce load jump fluctuations and is defined as: ; in yes The smooth regularization term ensures that the load distribution remains continuous between neighborhoods and avoids overly abrupt load transfers.

[0049] This step combines load optimization and active maintenance, and achieves the dual goals of life extension and load balancing through gradient optimization and regularization constraints. The innovativeness of the solution is reflected in the dynamic adjustment of load distribution and the accuracy of maintenance plans, which significantly improves the overall life and operating efficiency of the group.

[0050] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0051] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a division of logical functions. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0052] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0053] Although embodiments of the present invention have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for predicting the safety performance life of a new energy vehicle charging pile, characterized in that: The method comprises the following steps: Collect the real-time operation monitoring equipment data of the new energy vehicle charging piles, use the load correlation of the charging piles to build a micro-region load interaction graph, use a custom graph embedding method to extract dynamic load features in the micro-region load interaction graph, and obtain dynamic load features. The dynamic load features are time series matrices that represent the dynamic load status of each charging pile in the micro-region, and each row corresponds to the comprehensive features of a charging pile; wherein the real-time operation monitoring equipment data includes the load rate of the charging pile, the usage frequency of the charging pile, the load peak distribution of the charging pile, the load correlation of two random charging piles, and the micro-region environmental data; Constructing a key component degradation model based on real-time operation monitoring equipment data to evaluate the life degradation trend of the current charging pile, obtaining the unit time degradation rate of the current charging pile, and predicting the remaining life of the current charging pile based on the unit time degradation rate of the current charging pile; Based on the remaining life of the current charging pile, a time decay model is used to calculate the failure risk probability of the current charging pile, and the failure risk probability of the current charging pile is obtained; Construct a group life prediction model to predict the overall remaining life of the charging pile group in the micro-area according to the failure risk probability of the current charging pile. At the same time, generate a priority maintenance list based on the failure risk probability of the current charging pile and the remaining life of the current charging pile to ensure that the operation risk of the charging piles in the group is minimized; Use micro-regional environmental data combined with dynamic load characteristics to evaluate the current load optimization strategy, calculate the potential improvement in group life due to load optimization, and obtain the potential for group life improvement; Design a load optimization strategy to maximize the overall life of the charging pile group in the micro-area, generate an active maintenance plan based on the priority maintenance list, select the charging piles that need to replace parts, and optimize them in combination with the dynamic load. According to the optimized load distribution, design a dynamic diversion strategy to adjust the user load to the low-load equipment, and maximize the group life through gradient optimization and regularization constraints.

2. A method for predicting the safety performance life of a new energy vehicle charging pile according to claim 1, characterized in that: The micro-area load interaction diagram is constructed as follows: node Indicates the first Charging stations; Edge Weight Load dependency , reflects the load correlation strength between charging piles and is defined as follows: ； in, Indicates Charging stations at the time Load rate and Charging stations at the time Load rate The covariance of Indicates Charging stations at the time Load rate The standard deviation of Indicates Charging stations at the time Load rate The standard deviation of A customized graph embedding method is used to extract dynamic load features in the micro-area load interaction graph to obtain dynamic load features, including: For each node , summarizing the characteristics of its first-order neighbors: ； in, It is Charging stations at the time The first-order neighborhood aggregation characteristics of express The set of neighboring nodes; Neighbor node The feature vector of , For the Charging stations at the time Frequency of use, For the Charging stations at the time Load peak distribution; Incorporating environmental data , normalize and update all node features: ； in, Is a node The final feature representation of is the weight matrix; is the bias term; Is an activation function, used for nonlinear mapping; The resulting load characteristics It is a time series matrix, which represents the dynamic load status of each charging pile in the micro area. Each row corresponds to the comprehensive characteristics of a charging pile. The output form is: ； in, is the micro-region load characteristic matrix; It is Charging stations at the time The comprehensive load characteristics.

3. A method for predicting the safety performance life of a new energy vehicle charging pile according to claim 1, characterized in that: The key component degradation model is expressed as: ； in, It is The degradation rate per unit time of key components of each charging pile; is the influence function of dynamic load characteristics on degradation, through the dynamic load characteristic matrix Nonlinear combination representation of ; is the voltage square term, reflecting the stress effect of high voltage on components; is the current term, reflecting the heat loss of components caused by large current; It is a combination of micro-regional environmental data, used to quantify the impact of external conditions such as temperature and humidity; , , and is the weight of the corresponding item; is a regularization additional term used to capture the special effect of voltage and current interaction on accelerated aging of components, expressed as: ； The remaining life of the current charging pile is predicted according to the unit time degradation rate of the current charging pile, which is expressed as: ； in, The remaining life of the current charging pile; is the initial design life; is the cumulative degradation value, obtained by integrating the degradation rate formula.

4. A method for predicting the safety performance life of a new energy vehicle charging pile according to claim 1, characterized in that: The probability of failure risk , calculated as follows: ； in, For cumulative degradation.

5. A method for predicting the safety performance life of a new energy vehicle charging pile according to any one of claims 3 or 4, characterized in that: The group life prediction model is expressed as: ； in, is the overall remaining life of the charging pile group in the micro area; It is The failure risk of each charging pile is Adjusting the weighting of life expectancy; is the life improvement function of dynamic load and environmental conditions, used to capture the positive effect of load optimization; is the adjustment factor, controlling the proportion of load optimization’s impact on overall life; is the total number of charging posts in the group; The priority repair list is generated based on the priority score, calculated as follows: ； in, It is The priority score of each charging station; is the load anomaly degree, capturing the abnormal deviation of load characteristics; is the abnormality weight coefficient, which controls the impact of abnormal load on priority; according to Sort from high to low to generate a priority maintenance list , where a high score represents the necessity of priority repair.

6. A method for predicting the safety performance life of a new energy vehicle charging pile according to claim 5, characterized in that: The life improvement function of the dynamic load and environmental conditions The design is as follows: ； in, Load abnormality, which quantifies the part of the load characteristic that deviates from the normal range and is used to deduct the negative factors of life extension; It is the abnormal load deduction coefficient, which adjusts the weight of the impact of abnormal load on life.

7. A method for predicting the safety performance life of a new energy vehicle charging pile according to claim 5, characterized in that: The current load optimization strategy is evaluated by using micro-area environmental data combined with dynamic load characteristics, and the potential improvement of group life due to load optimization is calculated to obtain the potential improvement of group life, which is calculated as follows: ; in, Potential improvement in group lifespan for load optimization.

8. A method for predicting the safety performance life of a new energy vehicle charging pile according to claim 7, characterized in that: The optimization goal of the load optimization strategy is: ； in, is the overall lifespan of the optimized group; is the current group lifespan; After optimization The load variance of each charging pile is used to quantify the load balance; It is the load balancing adjustment coefficient, which weighs the priority between load balancing and life improvement.

9. A method for predicting the safety performance life of a new energy vehicle charging pile according to claim 5, characterized in that: The active maintenance plan is generated based on the priority maintenance list, which is expressed as: ； in, For the The maintenance decision of each charging pile, 1 means maintenance is needed, and 0 means no maintenance is needed; The threshold ratio of the remaining life is used to determine whether the equipment with a short life needs to be replaced proactively; is the risk threshold to determine whether the equipment with a higher risk of failure needs to replace key components; Press the charging pile that meets the maintenance conditions Sort them, include them into the maintenance plan one by one, and output the final maintenance list.

10. A method for predicting the safety performance life of a new energy vehicle charging pile according to claim 8, characterized in that: The dynamic diversion strategy is expressed as: ； in, For optimized load distribution; is the learning rate, which is used to control the gradient adjustment step size; The potential for life improvement with respect to the gradient of load distribution provides guidance for load optimization direction; is a load smoothing regularization term used to reduce load jump fluctuations and is defined as: ； in yes The set of neighbor nodes.

Images