Charging pile operation and maintenance management method and system

Abstract

The invention provides a charging pile operation and maintenance management method and system. The method comprises the following steps: acquiring an operation state parameter and an environment parameter of the charging pile; inputting the operation state parameters and the environment parameters into a multi-task prediction model to obtain the residual life and confidence interval of each key component of the charging pile; wherein the multi-task prediction model comprises a shared feature extraction module and a plurality of life prediction branches, and each life prediction branch corresponds to one key component; and determining a fault risk level of the charging pile based on the residual life and the confidence interval of each key component, and performing operation and maintenance management on the charging pile based on the fault risk level. Accurate and reliable data basis can be provided for operation and maintenance management of the charging pile, conversion of operation and maintenance of the charging pile from passive response to active prevention can be achieved, and remarkable breakthrough in the aspects of prediction precision, cost efficiency and system reliability is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of charging pile management, and in particular to a charging pile operation and maintenance management method and system. Background Art

[0002] With the rapid popularization of new energy vehicles, the scale of charging piles as infrastructure continues to expand, but their operation and maintenance management faces severe challenges.

[0003] At present, traditional operation and maintenance methods rely on manual regular inspections and threshold alarm mechanisms, which have significant limitations: on the one hand, the failure mechanisms of key components inside charging piles, such as power modules, heat dissipation systems and charging guns, are complex, and single sensor monitoring is difficult to accurately capture early degradation signals; on the other hand, existing life prediction methods are mostly based on statistical analysis of independent components or single data source modeling, ignoring the coupling effects between parameters and components, resulting in the current operation and maintenance management of charging piles causing local failures that may trigger systemic risks and resource mismatches. Summary of the Invention

[0004] The embodiments of the present invention provide a charging pile operation and maintenance management method and system to solve the problem of improving the operation and maintenance management efficiency of the charging pile.

[0005] In a first aspect, an embodiment of the present invention provides a charging pile operation and maintenance management method, comprising: Obtain the operating status parameters and environmental parameters of the charging pile; Inputting operating state parameters and environmental parameters into a multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile; the multi-task prediction model includes a shared feature extraction module and multiple life prediction branches, each corresponding to a key component; The failure risk level of the charging pile is determined based on the remaining life and confidence interval of each key component, and the operation and maintenance management of the charging pile is carried out based on the failure risk level.

[0006] In one possible implementation, the operating status parameters include the temperature sequence of the power module, the vibration spectrum of the cooling system, the contact image of the charging gun, and the number of plug-in and plug-out times; the environmental parameters include ambient temperature, ambient humidity, and dust concentration; and the shared feature extraction module includes an LSTM layer, a convolutional layer, and a fully connected layer. Inputting operating state parameters and environmental parameters into the multi-task prediction model, the remaining life and confidence interval of each key component of the charging pile are obtained, including: Input the temperature sequence into the LSTM layer to obtain the time series feature vector; Input the vibration spectrum and contact image into the convolution layer to obtain the image feature vector; Input the ambient temperature, ambient humidity and dust concentration into the fully connected layer to obtain the environmental feature vector; Concatenate the time series feature vector, image feature vector and environment feature vector to obtain a shared feature vector; The shared feature vector is input into multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile.

[0007] In one possible implementation, the key component includes a power module, the multiple life prediction branches include a power module prediction branch, and the operating state parameters also include a current ripple sequence of the power module; the power module prediction branch includes a Coffin-Manson physical constraint module and an LSTM network; The shared feature vector is input into multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile, including: Input the current ripple sequence into the Coffin-Manson physical constraint module to obtain the physical remaining life of the power module; The shared feature vector is input into the LSTM network to obtain the time series remaining life of the power module; The physical remaining life and the sequential remaining life are weightedly integrated to obtain the remaining life and confidence interval of the power module.

[0008] In one possible implementation, the key components include a cooling system, the multiple life prediction branches include a cooling system prediction branch, and the operating status parameters also include a pressure sequence of the cooling system; the power module prediction branch includes a CNN network, an autoencoder, and an XGBoost model; The shared feature vector is input into multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile, including: Input the vibration spectrum into the CNN network to obtain the vibration feature vector of the cooling system; The vibration feature vector is input into the autoencoder to obtain the health index of the cooling system; The shared feature vector and pressure sequence are input into the XGBoost model to obtain the pressure fluctuation index of the cooling system; By integrating the health index and the pressure fluctuation index, the remaining life and confidence interval of the cooling system are obtained.

[0009] In one possible implementation, the key component includes a charging gun, the multiple life prediction branches include a charging gun prediction branch, and the operating status parameters also include a bending angle sequence of the charging gun cable; the charging gun prediction branch includes a ResNet-18 network, a physical wear model, and a cable fatigue model; The shared feature vector is input into multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile, including: The shared feature vector is input into the ResNet-18 network to obtain the contact oxidation rate of the charging gun; The number of plug-in and plug-out times is input into the physical wear model to obtain the contact resistance increase rate of the charging gun; The bending angle sequence is input into the cable fatigue model to obtain the cable fatigue damage of the charging gun; The contact oxidation rate, contact resistance increase rate and cable fatigue damage are combined to obtain the remaining life and confidence interval of the charging gun.

[0010] In one possible implementation, before inputting the operating state parameters and environmental parameters into the multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile, the following is further included: Obtain the operating status parameters, environmental parameters, operating time of each key component and the time when the fault occurred of multiple charging piles; The operating status parameters and environmental parameters of each charging pile are used as training samples. The remaining life of each key component is calculated based on the operating time and fault occurrence time of each key component and used as sample labels to construct a training dataset. The initial multi-task prediction model is trained based on the training dataset and the multi-task joint loss function; the weights of each key component in the multi-task joint loss function are determined based on the fault impact coefficient.

[0011] In one possible implementation, the failure risk level of a charging pile is determined based on the remaining life and confidence interval of each key component, including: For each key component, calculate the risk score of the key component based on its design life, remaining life and confidence interval; The risk score of the charging pile is obtained by taking the failure impact coefficient of each key component as the weight and summing the risk scores of each key component. The fault risk level of the charging pile is determined based on the risk score of the charging pile and the risk score ranges corresponding to multiple fault risk levels; wherein the risk score ranges corresponding to the multiple fault risk levels are determined based on historical fault data.

[0012] In a second aspect, an embodiment of the present invention provides a charging pile operation and maintenance management system, including: An acquisition module is used to obtain the operating status parameters and environmental parameters of the charging pile; A prediction module, configured to input operating state parameters and environmental parameters into a multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile; wherein the multi-task prediction model includes a shared feature extraction module and multiple life prediction branches, each life prediction branch corresponding to a key component; The operation and maintenance module is used to determine the failure risk level of the charging pile based on the remaining life and confidence interval of each key component, and to perform operation and maintenance management of the charging pile based on the failure risk level.

[0013] In a third aspect, an embodiment of the present invention provides a terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the method described in the first aspect or any possible implementation of the first aspect are implemented.

[0014] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the method described in the first aspect or any possible implementation of the first aspect.

[0015] An embodiment of the present invention provides a charging pile operation and maintenance management method and system. Through a multi-task prediction model and a shared feature extraction module, cross-modal common features of time series, images, and environmental parameters are extracted, the dependency relationship between key components of the charging pile is modeled, and multiple life prediction branches are used to differentiate the application of shared feature vectors. The health status of key components is evaluated from a holistic level, providing accurate and reliable data basis for the operation and maintenance management of the charging pile, which can realize the transformation of charging pile operation and maintenance from passive response to active prevention, and achieve significant breakthroughs in prediction accuracy, cost efficiency and system reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 This is a flowchart of an implementation method for charging pile operation and maintenance management provided by an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a charging pile operation and maintenance management system provided by an embodiment of the present invention; Figure 3 is a schematic diagram of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0019] In order to make the purpose, technical solutions and advantages of the present invention more clear, specific embodiments will be described below with reference to the accompanying drawings.

[0020] See also Figure 1 , which shows a flow chart of an implementation method of a charging pile operation and maintenance management method provided by an embodiment of the present invention, which is described in detail as follows: Step 101: Obtain the operating status parameters and environmental parameters of the charging pile.

[0021] In this embodiment, operating status parameters may include temperature and vibration, and environmental parameters may include temperature, humidity and dust concentration. Temperature, vibration, humidity and other sensors and edge computing equipment are deployed near the charging pile to monitor the operating status parameters of the charging pile in real time. At the same time, images of charging gun head wear and heat sink dust accumulation are obtained through manual inspections and other methods to ensure the comprehensiveness of parameter acquisition.

[0022] In step 102, the operating state parameters and environmental parameters are input into a multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile; wherein the multi-task prediction model includes a shared feature extraction module and multiple life prediction branches, each life prediction branch corresponding to a key component.

[0023] In this embodiment, the shared feature extraction module is composed of a CNN and LSTM network, which outputs a fused shared feature vector. This shared feature extraction module can extract common features across components, capturing the global impact of the environment and operating conditions. For example, high temperatures can accelerate both thermal fatigue of the power module and bearing wear of the cooling fan. Shared features model this cross-component correlation using parameters such as ambient temperature and humidity. Furthermore, the shared feature module reduces model degrees of freedom through parameter reuse, improving generalization capabilities for sparse data and small sample sizes.

[0024] The shared feature vectors are applied differently in each prediction branch to predict the remaining life of each key component.

[0025] The key components of a charging pile may be a power module, a heat dissipation system, and a charging gun. The corresponding power module branch integrates the thermal fatigue constraints of the Coffin-Manson equation to output the first remaining life (RUL1). The heat dissipation system branch reconstructs the health indicators of the vibration spectrum through an autoencoder and outputs the second remaining life (RUL2). The charging gun branch combines the contact image features with the wear model of the number of plugging and unplugging times to output the third remaining life (RUL3).

[0026] Step 103 : determining the fault risk level of the charging pile based on the remaining life and confidence interval of each key component, and performing operation and maintenance management on the charging pile based on the fault risk level.

[0027] In this embodiment, each key component can be classified into a fault risk level based on its RUL confidence interval. For example, if the RUL is lower than threshold T1 and the confidence interval is greater than 90%, a preventive replacement work order is generated; if the RUL is lower than threshold T2 and the confidence interval is less than 70%, an immediate maintenance work order is generated. Thresholds T1 and T2 are dynamically adjusted based on historical fault data, and the adjustment formula is: T 1= α ⋅MTBF⋅(1−confidence interval) T2=β⋅MTTR⋅failure impact factor in, α , β is the weight factor, MTBF is the mean time between failures, MTTR is the mean time to repair, the failure impact coefficient is calculated by the historical usage frequency of the pile position and the user rating, and the confidence interval represents the uncertainty range of the model's predicted remaining useful life (RUL), usually presented in the form of an interval (for example, RUL = 1000 hours, the 95% confidence interval is [800,1200] hours), which means that the true RUL has a 95% probability of falling within this interval, and the wider the interval, the lower the confidence of the model in fitting the current data or working conditions.

[0028] T1 represents the preventive maintenance threshold. A longer MTBF indicates a higher component reliability, allowing for a looser T1 setting (i.e., a larger RUL threshold). A higher confidence interval (more certain predictions) and a smaller T1 value result in a more conservative maintenance triggering.

[0029] T2 represents the emergency maintenance threshold. The shorter the MTTR (faster repair) and the lower the fault impact coefficient (for example, the charging gun has a smaller impact on the overall charging pile), the smaller the T2 value, and maintenance is triggered only in the event of severe degradation.

[0030] It is also possible to integrate the remaining life and confidence interval of each key component to represent the failure risk level of the charging pile according to the importance of each key component, and adopt operation and maintenance measures with different costs and efficiencies according to different failure risk levels to achieve dynamic and differentiated operation and maintenance management.

[0031] The embodiment of the present invention uses a multi-task prediction model and a shared feature extraction module to extract cross-modal common features of time series, images, and environmental parameters, model the dependencies between key components of the charging pile, and use multiple life prediction branches to differentiate the application of shared feature vectors, evaluate the health status of key components from a holistic level, and provide accurate and reliable data basis for the operation and maintenance management of the charging pile. It can realize the transformation of charging pile operation and maintenance from passive response to active prevention, and achieve significant breakthroughs in prediction accuracy, cost efficiency and system reliability.

[0032] In one possible implementation, the operating status parameters include the temperature sequence of the power module, the vibration spectrum of the cooling system, the contact image of the charging gun, and the number of plug-in and plug-out times; the environmental parameters include ambient temperature, ambient humidity, and dust concentration; and the shared feature extraction module includes an LSTM layer, a convolutional layer, and a fully connected layer. Inputting operating state parameters and environmental parameters into the multi-task prediction model, the remaining life and confidence interval of each key component of the charging pile are obtained, including: Input the temperature sequence into the LSTM layer to obtain the time series feature vector; Input the vibration spectrum and contact image into the convolution layer to obtain the image feature vector; Input the ambient temperature, ambient humidity and dust concentration into the fully connected layer to obtain the environmental feature vector; Concatenate the time series feature vector, image feature vector and environment feature vector to obtain a shared feature vector; The shared feature vector is input into multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile.

[0033] In this embodiment, the operating status parameters include the power module temperature series, which reflects changes in device thermal load; the cooling system vibration spectrum, which captures mechanical wear characteristics; the charging gun contact image, which records surface oxidation and physical damage; and the number of plugging and unplugging cycles, which quantifies mechanical fatigue accumulation. Ambient temperature, humidity, and dust concentration are used to modify component degradation rates. High temperatures accelerate material aging, high humidity promotes metal oxidation, and dust accumulation reduces cooling efficiency.

[0034] In the shared feature extraction module, the LSTM layer can analyze the temporal dependencies of temperature sequences, thereby capturing the periodic thermal cycle patterns of the power module. The convolutional layer extracts abnormal frequency components (such as bearing fault characteristic peaks) in the vibration spectrum and texture degradation patterns (such as oxidation spot distribution) in the contact image through local perception characteristics. The fully connected layer maps environmental parameters into interpretable scalar factors, quantifying the weight of the impact of external conditions on component life.

[0035] The feature concatenation stage aligns temporal, image, and environmental features within a unified vector space, enabling the model to learn cross-modal association patterns. Finally, the shared features are fed into multiple lifespan prediction branches, mitigating overfitting risks through parameter sharing while preserving component-specific prediction capabilities.

[0036] In one possible implementation, the key component includes a power module, the multiple life prediction branches include a power module prediction branch, and the operating state parameters also include a current ripple sequence of the power module; the power module prediction branch includes a Coffin-Manson physical constraint module and an LSTM network; The shared feature vector is input into multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile, including: Input the current ripple sequence into the Coffin-Manson physical constraint module to obtain the physical remaining life of the power module; The shared feature vector is input into the LSTM network to obtain the time series remaining life of the power module; The physical remaining life and the sequential remaining life are weightedly integrated to obtain the remaining life and confidence interval of the power module.

[0037] In this embodiment, the power module is responsible for converting AC power from the grid into DC power suitable for charging electric vehicle batteries, or directly providing AC power. It is the core of the charging pile's power conversion and contains key power electronic components such as rectifiers and inverters.

[0038] As the core power conversion unit of charging piles, the power module's lifespan prediction must account for both thermal fatigue mechanisms and dynamic operating conditions. The current ripple sequence reflects the switching losses and load fluctuations of the power components within the power module. Increased high-frequency ripple amplitude accelerates insulation material aging and generates additional heat dissipation. The Coffin-Manson physical constraint module, built on the thermal fatigue life equation, calculates the theoretical lifespan by analyzing the junction temperature fluctuations caused by the number of temperature cycles and ripple current. The material constants C and m in the equation are calibrated through accelerated aging tests to ensure the model conforms to the physical degradation laws of power device solder joints. The shared feature vectors received by the LSTM network incorporate multi-dimensional information such as historical temperature trends, cooling system status, and ambient dust concentration. This allows it to capture the nonlinear effects of complex operating conditions (such as sudden temperature rise caused by a sudden load increase) on lifespan.

[0039] Confidence intervals are generated using the Monte Carlo Dropout method, which randomly drops some neurons during LSTM inference. The variance of the distribution of prediction results from multiple sampling is used to reflect uncertainty. The current ripple series is chosen because it directly correlates with IGBT / MOSFET switching losses and better characterizes device stress than average current. The Coffin-Manson equation, due to its universal applicability in the field of thermomechanical fatigue, provides a physical benchmark for power module life prediction. The LSTM network is suitable for learning long-term coupled patterns between temperature and ripple (for example, the cumulative effect of ripple amplitude on life degradation during low nighttime temperatures).

[0040] The physical remaining life and the temporal remaining life are weightedly fused through residual connection. The formula is:

[0041] in, is the remaining life of the power module, is the remaining lifetime of the sequence, is a learnable parameter, is the physical remaining lifespan.

[0042] The residual connection linearly superimposes the physical remaining life output by the physical model and the time series remaining life predicted by LSTM, and its weight coefficient By automatically optimizing the training data, the interpretability of physical laws and the generalization ability of data-driven learning can be balanced. For example, when the ambient temperature rises abnormally, LSTM may capture the heat dissipation deterioration effect beyond the Coffin-Manson equation assumption. Lower to enhance data model contribution.

[0043] In one possible implementation, the key components include a cooling system, the multiple life prediction branches include a cooling system prediction branch, and the operating status parameters also include a pressure sequence of the cooling system; the power module prediction branch includes a CNN network, an autoencoder, and an XGBoost model; The shared feature vector is input into multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile, including: Input the vibration spectrum into the CNN network to obtain the vibration feature vector of the cooling system; The vibration feature vector is input into the autoencoder to obtain the health index of the cooling system; The shared feature vector and pressure sequence are input into the XGBoost model to obtain the pressure fluctuation index of the cooling system; By integrating the health index and the pressure fluctuation index, the remaining life and confidence interval of the cooling system are obtained.

[0044] In this embodiment, the cooling system removes heat from the charging pile through fans and liquid cooling, ensuring that the power module and other electronic components operate within a safe temperature range. Poor heat dissipation can lead to overheating, accelerated aging, and potentially failures of the power module and charging connector. Real-time monitoring of temperature, fan speed, and coolant flow is necessary to monitor the health of the cooling system and promptly clean dust from the heat sink or replace failed components.

[0045] The vibration spectrogram extracts frequency-domain features from the raw acceleration signal through Fourier transform. Increased amplitudes of characteristic frequencies of bearing faults (such as the cage pass frequency (BPFI)) indicate wear on mechanical components. The CNN, with its advantages in extracting local image features, is well-suited for frequency-domain visualization of the vibration spectrum. Convolutional neural networks (CNNs) capture localized abnormal patterns in the spectrogram using convolution kernels. Their hierarchical structure automatically identifies early, weak fault signals.

[0046] The autoencoder uses an encoder-decoder architecture that preserves the spatial local correlation of the spectrogram and avoids the smoothing effect of fully connected networks on high-frequency details. During training, the normal operating vibration spectrum is reconstructed. During actual operation, the health index (HI) is generated by calculating the mean squared error between the input spectrum and the reconstructed result. The health index is defined as the sliding window mean of the reconstruction error. This metric quantifies the degree of deviation from normal conditions (for example, increased high-frequency energy due to bearing wear).

[0047] The XGBoost model was selected for its efficient processing of structured features (such as statistics and environmental scalars) and its ability to rank feature importance, resulting in superior interpretability compared to neural networks. The XGBoost model leverages shared feature vectors and time-series statistics (mean and variance) of the cooling system's pressure, ambient dust concentration, and power module load factor. Using a gradient boosting tree, the model models the nonlinear relationship between pressure fluctuations and cooling efficiency. For example, dust accumulation increases pipe resistance, leading to increased pressure fluctuations.

[0048] The health index (HI) characterizes accumulated mechanical wear, while the pressure fluctuation index reflects fluid system performance degradation. A dynamic weighting strategy is used to integrate the health and pressure fluctuation indices: when the HI is greater than 0.5 (significant mechanical anomaly), the weighting shifts toward the HI; when the pressure fluctuation exceeds 2σ of the historical baseline, the weighting shifts toward the pressure index. Confidence intervals are calculated using quantile regression. This embodiment, by separating the mechanical and fluid degradation paths and introducing an environmentally adaptive fusion mechanism, addresses the insensitivity of traditional single indicators to complex faults, enabling refined assessment of cooling system lifespan.

[0049] In one possible implementation, the key component includes a charging gun, the multiple life prediction branches include a charging gun prediction branch, and the operating status parameters also include a bending angle sequence of the charging gun cable; the charging gun prediction branch includes a ResNet-18 network, a physical wear model, and a cable fatigue model; The shared feature vector is input into multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile, including: The shared feature vector is input into the ResNet-18 network to obtain the contact oxidation rate of the charging gun; The number of plug-in and plug-out times is input into the physical wear model to obtain the contact resistance increase rate of the charging gun; The bending angle sequence is input into the cable fatigue model to obtain the cable fatigue damage of the charging gun; The contact oxidation rate, contact resistance increase rate and cable fatigue damage are combined to obtain the remaining life and confidence interval of the charging gun.

[0050] In this embodiment, the charging cable is the component that the user directly operates. It is responsible for safely transmitting electrical energy to the electric vehicle battery and must possess high conductivity, wear resistance, and a seal. Wear or oxidation during insertion and removal can increase contact resistance, causing heat generation, reduced charging efficiency, and even sparks.

[0051] Predicting the remaining life of a charging gun requires considering both electrical contact degradation and mechanical fatigue. The environmental parameters and historical load data in the shared feature vector are analyzed through deep convolutional layers of the ResNet-18 network, focusing on the texture expansion patterns of oxidation spots on the contact surface. The output contact oxidation rate reflects the degree of chemical corrosion of the metal material. The physical wear model, based on Archard wear theory, maps the number of plugging and unplugging cycles to a mechanical wear volume. The contact resistance rise rate is calculated by combining contact material hardness and contact pressure to quantify the electrical performance degradation process of the plug interface.

[0052] The cable fatigue model uses the rain flow counting method to analyze bending angle sequences, identify typical bending cycles (e.g., the plastic deformation threshold for copper conductors corresponding to a 90-degree bend). Miner's linear cumulative damage theory is used to calculate the life loss ratio at each stress level, ultimately accumulating the total fatigue damage value. A dynamic attenuation factor is introduced by integrating the contact oxidation rate, contact resistance increase rate, and cable fatigue damage. When the contact resistance increase rate exceeds a threshold, the impact of oxidation on life is reduced (because the contact may be nearing failure), and the focus is shifted to assessing cable fatigue damage. ResNet-18 was selected because its residual structure effectively captures the multi-scale features of oxidation regions (e.g., micron-scale cracks and macroscopic discoloration), avoiding the limitations of manual feature design in traditional image algorithms. The physical wear model is based on the Archard equation due to its engineering universality in sliding contact wear prediction. The rain flow counting method, due to its ability to accurately extract valid cycles in random load sequences, is suitable for irregular bending scenarios in actual cable use.

[0053] In the charging gun prediction branch, ResNet-18 leverages the ambient humidity parameter in shared features to enhance the sensitivity of oxidation area detection (e.g., the color difference of oxidation texture is more pronounced in high-humidity environments). The physical wear model dynamically modifies the contact pressure parameter based on the current load data in the shared features (high current causes contact point temperature rise, exacerbating wear). The cable fatigue model utilizes the usage frequency data in the shared features to optimize rain flow count clustering (ignoring the impact of small bends in high-frequency usage scenarios). The confidence interval is generated using the Bootstrap resampling method, and the stability of life prediction is evaluated by perturbing the input parameters (e.g., oxidation rate detection error of ±5%). By decoupling the electrical contact degradation and mechanical fatigue mechanisms and introducing an environmentally adaptive fusion strategy, this method addresses the high misjudgment rate of traditional single indicators (e.g., plug-in / plug-out threshold) for complex failure modes, enabling accurate multi-dimensional assessment of charging gun life.

[0054] In one possible implementation, before inputting the operating state parameters and environmental parameters into the multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile, the following is further included: Obtain the operating status parameters, environmental parameters, operating time of each key component and the time when the fault occurred of multiple charging piles; The operating status parameters and environmental parameters of each charging pile are used as training samples. The remaining life of each key component is calculated based on the operating time and fault occurrence time of each key component and used as sample labels to construct a training dataset. The initial multi-task prediction model is trained based on the training dataset and the multi-task joint loss function; the weights of each key component in the multi-task joint loss function are determined based on the fault impact coefficient.

[0055] In this embodiment, multi-task joint training forces the shared layer to learn feature representations that are beneficial to multiple tasks. Historical operating data for multiple charging stations covers different regions, climate conditions, and usage intensities, ensuring the diversity and generalizability of the training samples.

[0056] Among the operating status parameters, the power module temperature series captures the complete thermal cycle before the failure. The cooling system vibration spectrum diagram annotates abnormal frequency intervals. Charging gun contact images are classified by oxidation level and matched with the number of plug-in and unplug cycles. The cable bend angle series extracts effective cycles through rain flow counting. Environmental parameters are strictly aligned with the operating data by timestamp to avoid seasonal interference. The operating time of each key component is calculated from the time of activation to the time of failure or the current time (for non-faulty samples). The remaining life label is converted into a relative life ratio through normalization. For example, the power module label is (fault occurrence time - data collection time) / design life. Non-faulty samples are right-truncated.

[0057] In the multi-task joint loss function, the loss weight of each component is dynamically allocated based on the fault impact coefficient: for example, the power module has the highest weight (0.5), the cooling system is second (0.3), and the charging gun has the lowest weight (0.2), ensuring that the prediction accuracy of high-impact components is optimized first.

[0058] A curriculum learning strategy can be introduced into the training process, initially focusing on the convergence of the power module and the cooling system, and later gradually increasing the weight of the charging gun task to alleviate the optimization conflicts in multi-task learning.

[0059] The fault impact coefficient can be calculated based on historical fault downtime and repair costs. For example, the average downtime for a single power module failure is 48 hours, the cooling system 24 hours, and the charging gun 6 hours. The initial weight is determined after normalization and fine-tuned during training based on the performance of the validation set.

[0060] In one possible implementation, the failure risk level of a charging pile is determined based on the remaining life and confidence interval of each key component, including: For each key component, calculate the risk score of the key component based on its design life, remaining life and confidence interval; The risk score of the charging pile is obtained by taking the failure impact coefficient of each key component as the weight and summing the risk scores of each key component. The fault risk level of the charging pile is determined based on the risk score of the charging pile and the risk score ranges corresponding to multiple fault risk levels; wherein the risk score ranges corresponding to the multiple fault risk levels are determined based on historical fault data.

[0061] In this embodiment, the charging pile failure risk level assessment can be achieved by quantifying the degradation state of key components and their system-level impact. Specifically, for each key component, the risk score calculation comprehensively considers its life consumption rate and prediction uncertainty: the ratio of remaining life to design life reflects the current degree of degradation (for example, if the remaining life is 20% of the design life, the basic risk score is 0.8), and the width of the confidence interval measures the reliability of the prediction (for example, the interval [800, 1200] hours corresponds to an uncertainty factor of 0.4). The two are multiplied together to amplify the risk value of low-confidence predictions (0.8 × 1.4 = 1.12, truncated to 1.0).

[0062] The failure impact coefficient quantifies the global impact weight based on the historical consequences of component failures (for example, if a power module failure causes the entire charging pile to shut down, the coefficient is set to 0.5). The weighted summation results in the overall risk score of the charging pile. For example, the power module risk score is 1.0, the cooling system risk score is 0.6, and the charging gun risk score is 0.3. The total score is 1.0×0.5+0.6×0.3+0.3×0.2=0.68.

[0063] The risk level range division relies on statistical analysis of historical failure data: from the risk scores before and after the failure of a large number of charging piles, thresholds of low risk (0.0-0.3), medium risk (0.3-0.7), and high risk (0.7-1.0) are selected to ensure that the level division matches the actual failure rate (for example, the high-risk range covers 90% of sudden failure cases).

[0064] In this embodiment, the risk score calculation selects the design life as the benchmark because it standardizes the life dimensions of different components (such as the design life of the power module is 5000 hours and the charging gun is 10,000 plug-in and unplug cycles), making the scores comparable; the confidence interval width introduces a nonlinear amplification factor to avoid over-trust in low-quality predictions; the weighted summation mechanism combines the component importance (fault impact coefficient) with the real-time status (risk score), which can overcome the limitations of traditional methods that only rely on status or fixed weights.

[0065] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0066] The following is a system embodiment of the present invention. For details not described in detail, please refer to the corresponding method embodiment described above.

[0067] Figure 2 The following is a schematic diagram of the structure of a charging pile operation and maintenance management system provided by an embodiment of the present invention. For ease of explanation, only the parts related to the embodiment of the present invention are shown, which are detailed as follows: like Figure 2 As shown, a charging pile operation and maintenance management system 2 includes: Acquisition module 21, used to obtain the operating status parameters and environmental parameters of the charging pile; A prediction module 22 is configured to input operating state parameters and environmental parameters into a multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile; wherein the multi-task prediction model includes a shared feature extraction module and multiple life prediction branches, each life prediction branch corresponding to a key component; The operation and maintenance module 23 is used to determine the fault risk level of the charging pile based on the remaining life and confidence interval of each key component, and to perform operation and maintenance management of the charging pile based on the fault risk level.

[0068] In one possible implementation, the operating status parameters include the temperature sequence of the power module, the vibration spectrum of the cooling system, the contact image of the charging gun, and the number of plug-in and plug-out times; the environmental parameters include ambient temperature, ambient humidity, and dust concentration; and the shared feature extraction module includes an LSTM layer, a convolutional layer, and a fully connected layer. The prediction module 22 is specifically used for: Input the temperature sequence into the LSTM layer to obtain the time series feature vector; Input the vibration spectrum and contact image into the convolution layer to obtain the image feature vector; Input the ambient temperature, ambient humidity and dust concentration into the fully connected layer to obtain the environmental feature vector; Concatenate the time series feature vector, image feature vector and environment feature vector to obtain a shared feature vector; The shared feature vector is input into multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile.

[0069] In one possible implementation, the key component includes a power module, the multiple life prediction branches include a power module prediction branch, and the operating state parameters also include a current ripple sequence of the power module; the power module prediction branch includes a Coffin-Manson physical constraint module and an LSTM network; The prediction module 22 is specifically used for: Input the current ripple sequence into the Coffin-Manson physical constraint module to obtain the physical remaining life of the power module; The shared feature vector is input into the LSTM network to obtain the time series remaining life of the power module; The physical remaining life and the sequential remaining life are weightedly integrated to obtain the remaining life and confidence interval of the power module.

[0070] In one possible implementation, the key components include a cooling system, the multiple life prediction branches include a cooling system prediction branch, and the operating status parameters also include a pressure sequence of the cooling system; the power module prediction branch includes a CNN network, an autoencoder, and an XGBoost model; The prediction module 22 is specifically used for: Input the vibration spectrum into the CNN network to obtain the vibration feature vector of the cooling system; The vibration feature vector is input into the autoencoder to obtain the health index of the cooling system; The shared feature vector and pressure sequence are input into the XGBoost model to obtain the pressure fluctuation index of the cooling system; By integrating the health index and the pressure fluctuation index, the remaining life and confidence interval of the cooling system are obtained.

[0071] In one possible implementation, the key component includes a charging gun, the multiple life prediction branches include a charging gun prediction branch, and the operating status parameters also include a bending angle sequence of the charging gun cable; the charging gun prediction branch includes a ResNet-18 network, a physical wear model, and a cable fatigue model; The prediction module 22 is specifically used for: The shared feature vector is input into the ResNet-18 network to obtain the contact oxidation rate of the charging gun; The number of plug-in and plug-out times is input into the physical wear model to obtain the contact resistance increase rate of the charging gun; The bending angle sequence is input into the cable fatigue model to obtain the cable fatigue damage of the charging gun; The contact oxidation rate, contact resistance increase rate and cable fatigue damage are combined to obtain the remaining life and confidence interval of the charging gun.

[0072] In a possible implementation, the prediction module 22 is further configured to: Before inputting the operating status parameters and environmental parameters into the multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile, the operating status parameters, environmental parameters, operating time of each key component and fault occurrence time of multiple charging piles are obtained; The operating status parameters and environmental parameters of each charging pile are used as training samples. The remaining life of each key component is calculated based on the operating time and fault occurrence time of each key component and used as sample labels to construct a training dataset. The initial multi-task prediction model is trained based on the training dataset and the multi-task joint loss function; the weights of each key component in the multi-task joint loss function are determined based on the fault impact coefficient.

[0073] In a possible implementation, the operation and maintenance module 23 is specifically configured to: For each key component, calculate the risk score of the key component based on its design life, remaining life and confidence interval; The risk score of the charging pile is obtained by taking the failure impact coefficient of each key component as the weight and summing the risk scores of each key component. The fault risk level of the charging pile is determined based on the risk score of the charging pile and the risk score ranges corresponding to multiple fault risk levels; wherein the risk score ranges corresponding to the multiple fault risk levels are determined based on historical fault data.

[0074] The embodiment of the present invention uses a multi-task prediction model and a shared feature extraction module to extract cross-modal common features of time series, images, and environmental parameters, model the dependencies between key components of the charging pile, and use multiple life prediction branches to differentiate the application of shared feature vectors, evaluate the health status of key components from a holistic level, and provide accurate and reliable data basis for the operation and maintenance management of the charging pile. It can realize the transformation of charging pile operation and maintenance from passive response to active prevention, and achieve significant breakthroughs in prediction accuracy, cost efficiency and system reliability.

[0075] Figure 3 Schematic diagram of a terminal provided by an embodiment of the present invention. Figure 3 As shown, the terminal 3 of this embodiment includes: a processor 30, a memory 31, and a computer program 32 stored in the memory 31 and executable on the processor 30. When the processor 30 executes the computer program 32, the steps of each of the above-mentioned charging pile operation and maintenance management method embodiments are implemented. Alternatively, when the processor 30 executes the computer program 32, the functions of each module / unit in the above-mentioned system embodiments are implemented.

[0076] Exemplarily, the computer program 32 may be divided into one or more modules / units, which are stored in the memory 31 and executed by the processor 30 to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program 32 in the terminal 3.

[0077] The terminal 3 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The terminal 3 may include, but is not limited to, a processor 30 and a memory 31. Those skilled in the art will understand that Figure 3 It is only an example of terminal 3 and does not constitute a limitation on terminal 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal may also include input and output devices, network access devices, buses, etc.

[0078] The processor 30 may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0079] The memory 31 can be an internal storage unit of the terminal 3, such as a hard drive or memory of the terminal 3. The memory 31 can also be an external storage device of the terminal 3, such as a plug-in hard drive, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc. equipped with the terminal 3. Furthermore, the memory 31 can include both the internal storage unit of the terminal 3 and an external storage device. The memory 31 is used to store the computer program and other programs and data required by the terminal. The memory 31 can also be used to temporarily store data that has been output or is about to be output.

[0080] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0081] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0082] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0083] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For example, the division of the modules or units is merely a logical functional division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of devices or units, and can be electrical, mechanical, or other forms.

[0084] The units described as separate components may or may not be physically separate, and 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 these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0085] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0086] If the integrated module / unit is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can also implement all or part of the process steps in the above-mentioned method embodiments by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned charging pile operation and maintenance management method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunications signal, and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practices in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practices, computer-readable media does not include electrical carrier signals and telecommunication signals.

[0087] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A charging pile operation and maintenance management method, characterized in that: include: Obtain the operating status parameters and environmental parameters of the charging pile; Inputting the operating state parameters and the environmental parameters into a multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile; wherein the multi-task prediction model includes a shared feature extraction module and multiple life prediction branches, each life prediction branch corresponding to a key component; The failure risk level of the charging pile is determined based on the remaining life and confidence interval of each key component, and the operation and maintenance management of the charging pile is performed based on the failure risk level.

2. A charging pile operation and maintenance management method according to claim 1, characterized in that: The operating status parameters include the temperature sequence of the power module, the vibration spectrum of the cooling system, the contact image of the charging gun, and the number of plug-in and plug-out times; the environmental parameters include ambient temperature, ambient humidity, and dust concentration; the shared feature extraction module includes an LSTM layer, a convolutional layer, and a fully connected layer; Inputting the operating state parameters and the environmental parameters into a multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile includes: Inputting the temperature sequence into the LSTM layer to obtain a time series feature vector; Inputting the vibration spectrum graph and the contact image into the convolution layer to obtain an image feature vector; Inputting the ambient temperature, the ambient humidity, and the dust concentration into the fully connected layer to obtain an environmental feature vector; splicing the time series feature vector, the image feature vector and the environment feature vector to obtain a shared feature vector; The shared feature vector is input into the multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile.

3. A charging pile operation and maintenance management method according to claim 2, characterized in that: The key component includes a power module, the multiple life prediction branches include a power module prediction branch, the operating state parameter also includes a current ripple sequence of the power module; the power module prediction branch includes a Coffin-Manson physical constraint module and an LSTM network; Inputting the shared feature vector into the multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile includes: Inputting the current ripple sequence into the Coffin-Manson physical constraint module to obtain the physical dimension remaining life of the power module; Inputting the shared feature vector into an LSTM network to obtain the time series remaining life of the power module; The physical remaining life and the sequential remaining life are weightedly integrated to obtain the remaining life and confidence interval of the power module.

4. A charging pile operation and maintenance management method according to claim 2, characterized in that: The key component includes a heat dissipation system, the multiple life prediction branches include a heat dissipation system prediction branch, and the operating state parameters also include a pressure sequence of the heat dissipation system; the power module prediction branch includes a CNN network, an autoencoder, and an XGBoost model; Inputting the shared feature vector into the multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile includes: Inputting the vibration spectrum graph into the CNN network to obtain a vibration eigenvector of the heat dissipation system; Inputting the vibration feature vector into the autoencoder to obtain a health indicator of the cooling system; Inputting the shared feature vector into the XGBoost model to obtain a pressure fluctuation index of the cooling system; The health index and the pressure fluctuation index are integrated to obtain the remaining life and confidence interval of the cooling system.

5. A charging pile operation and maintenance management method according to claim 2, characterized in that: The key component includes a charging gun, the multiple life prediction branches include a charging gun prediction branch, and the operating state parameters also include a bending angle sequence of the cable of the charging gun; the charging gun prediction branch includes a ResNet-18 network, a physical wear model, and a cable fatigue model; Inputting the shared feature vector into the multiple life prediction branches to obtain the remaining life and confidence interval of each key component of the charging pile includes: Inputting the shared feature vector into the ResNet-18 network to obtain the contact oxidation rate of the charging gun; Inputting the number of plugging and unplugging times into the physical wear model to obtain a contact resistance increase rate of the charging gun; Inputting the bending angle sequence into the cable fatigue model to obtain cable fatigue damage of the charging gun; The contact oxidation rate, the contact resistance increase rate, and the cable fatigue damage are combined to obtain the remaining life and confidence interval of the charging gun.

6. A charging pile operation and maintenance management method according to claim 1, characterized in that: Before inputting the operating state parameters and the environmental parameters into the multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile, the method further includes: Obtain the operating status parameters, environmental parameters, operating time of each key component and the time when the fault occurred of multiple charging piles; The operating status parameters and environmental parameters of each charging pile are used as training samples. The remaining life of each key component is calculated based on the operating time and fault occurrence time of each key component and used as sample labels to construct a training dataset. An initial multi-task prediction model is trained based on the training data set and the multi-task joint loss function; wherein the weight of each key component in the multi-task joint loss function is determined based on the fault influence coefficient.

7. A charging pile operation and maintenance management method according to claim 1, characterized in that: The determining of the failure risk level of the charging pile based on the remaining life and confidence interval of each key component includes: For each key component, calculate the risk score of the key component based on its design life, remaining life and confidence interval; Taking the failure impact coefficient of each key component as a weight, the risk score of each key component is weighted and summed to obtain the risk score of the charging pile; The fault risk level of the charging pile is determined based on the risk score of the charging pile and the risk score ranges corresponding to multiple fault risk levels; wherein the risk score ranges corresponding to the multiple fault risk levels are determined based on historical fault data.

8. A charging pile operation and maintenance management system, characterized in that: include: An acquisition module is used to obtain the operating status parameters and environmental parameters of the charging pile; a prediction module, configured to input the operating state parameters and the environmental parameters into a multi-task prediction model to obtain the remaining life and confidence interval of each key component of the charging pile; wherein the multi-task prediction model includes a shared feature extraction module and multiple life prediction branches, each life prediction branch corresponding to a key component; An operation and maintenance module is used to determine the failure risk level of the charging pile based on the remaining life and confidence interval of each key component, and to perform operation and maintenance management of the charging pile based on the failure risk level.

9. A terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images