Method and system for multi-stage power distribution of a DC charging pile

Abstract

The present invention discloses a method and system for multi-stage power distribution of a DC charging pile, which relates to the technical field of DC charging pile power distribution, and includes the following steps: According to historical charging data and monitoring requirements, set an initial monitoring frequency in the first charging stage of the DC charging pile, and collect current data when the electric vehicle battery is charging at the initial monitoring frequency; On the basis of data collection, introduce a time-sliding monitoring window, and form a current data set by arranging the current data within the monitoring window in chronological order. The present invention realizes intelligent monitoring and abnormal prediction of current data by setting the initial monitoring frequency, combining the sliding window and the neural network model. In the potential abnormal stage, dynamically increase the monitoring frequency and reduce the charging power, accurately capture abnormal current, protect the battery performance, extend the service life, optimize the system efficiency while ensuring charging safety, and provide reliable technical support for high-power charging scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of DC charging pile power distribution, and particularly to a method and system for multi-stage power distribution of DC charging piles. Background Art

[0002] Multi-stage power distribution of DC charging piles refers to dividing the charging process into multiple stages according to the charging requirements and status of the electric vehicle battery (such as SOC status, temperature, charging curve, etc.), and adopting different power output strategies in different stages to optimize the charging efficiency, protect the battery life, and ensure safety. For example, when the battery SOC is relatively low (such as 0 - 30%), a relatively high power is used for rapid charging; in the middle stage (such as 30 - 80%), the power is gradually reduced to reduce heat generation and battery loss; while in the high SOC stage (such as 80% - 100%), the charging is completed with a relatively low power or constant voltage mode to precisely control the power input. This method can balance the charging speed and battery life and is suitable for the optimized design of intelligent DC charging piles.

[0003] When the DC charging pile charges the battery in the low SOC stage (such as 0 - 30%), a constant current charging mode is usually adopted, that is, a relatively high constant current (high power) is used to quickly supplement the power. This is because the internal resistance of the battery is relatively low in this stage, and it can safely withstand a relatively large current, thus achieving efficient charging. However, high-power charging also brings potential risks. If the current control accuracy of the charging pile is insufficient, resulting in the current exceeding the maximum design tolerance of the battery, it may cause overheating inside the battery and material degradation, and then cause irreversible damage to the battery. In extreme cases, such excessive current may trigger thermal runaway, leading to serious consequences such as fire or explosion.

[0004] To solve the above problems, the prior art usually uses a current sensor to monitor the current during battery charging, and immediately takes measures when the current during battery charging exceeds the set value. However, when the prior art monitors the current, it usually monitors the current in real time at a fixed monitoring frequency, and the monitoring frequency is usually not set very high. The main advantage of doing this is to balance real-time performance and system resource consumption; however, when the current control of the charging pile is abnormal, continuing to use this monitoring frequency may miss the rapidly rising current peak, resulting in the abnormal current not being detected and controlled in time. Continuing excessive current will cause overheating inside the battery, exacerbate material decomposition and degradation, lead to irreversible damage, shorten the battery life, and even make it scrapped.

[0005] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and therefore it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0006] The object of the present invention is to provide a method and system for multi-stage power distribution of a DC charging pile. By setting an initial monitoring frequency, efficient utilization of system resources is achieved. Combining a sliding window and a neural network model, current data is dynamically analyzed and the operation trend is accurately predicted, effectively distinguishing the normal operation stage and the potential abnormal stage, thereby realizing intelligent monitoring. In the potential abnormal stage, dynamically increasing the monitoring frequency significantly improves the ability to capture abnormal currents, avoiding battery damage or safety accidents caused by response delays. At the same time, the charging power is appropriately reduced to reduce the current load, protect the battery performance and extend its service life. Thus, on the basis of ensuring charging safety, the monitoring accuracy and operation efficiency of the system are comprehensively optimized, providing a reliable and intelligent technical solution for high-power charging scenarios to solve the problems in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solution: A method for multi-stage power distribution of a DC charging pile, comprising the following steps:

[0008] According to historical charging data and monitoring requirements, set an initial monitoring frequency in the first charging stage of the DC charging pile, and collect current data when the electric vehicle battery is charging at the initial monitoring frequency.

[0009] On the basis of data collection, introduce a time-sliding monitoring window, and form a current data set by arranging the current data within the monitoring window in chronological order, so that the current data set is continuously updated as time progresses.

[0010] Within the monitoring window, after performing abnormal analysis on the current data in the current data set, input the analyzed features into a pre-trained neural network model, and predict the current operation trend when the electric vehicle battery is charging through the neural network model.

[0011] Based on the output result of the neural network model, divide the current situation within the monitoring window into a normal operation stage and a potential abnormal operation stage.

[0012] For the current situation classified as the normal operation stage, continue to monitor the current when the electric vehicle battery is charging at the set initial monitoring frequency.

[0013] For the current situation classified as the potential abnormal operation stage, according to the prediction result of the neural network model, dynamically increase the actual monitoring frequency of current monitoring to achieve accurate monitoring of abnormal current situations, and reduce the charging power to reduce the current load.

[0014] Preferably, the initial monitoring frequency refers to a fixed data collection frequency preset when the DC charging pile performs constant current charging on the electric vehicle battery, and is used to monitor the change of the battery charging current.

[0015] Preferably, a sliding window refers to, during the data acquisition process, taking a fixed time length as the window range, extracting a part of the data from the continuous current data stream for processing and analysis, and as time goes by, the window continuously moves forward, and the data contained therein is updated each time it moves.

[0016] Preferably, within the monitoring window, perform anomaly analysis on the current data in the current data set. The specific steps are as follows:

[0017] Extract the ratio of the amplitude to the frequency of the sharp increase in the current waveform and the rate and cumulative amplitude of the continuous increase of the current over a period of time from the current data set. After analyzing the extracted features, generate a spike quantization reference value and a continuous increase quantization reference value respectively. The spike quantization reference value reflects the instantaneous abnormal sharp increase degree and occurrence frequency of the charging current; the continuous increase quantization reference value reflects the trend of the current deviating from the stable value for a long time.

[0018] Preferably, after obtaining the spike quantization reference value and the continuous increase quantization reference value obtained by performing anomaly analysis on the current data in the current data set, input the spike quantization reference value and the continuous increase quantization reference value into a pre-trained neural network model, generate a current potential anomaly index through the model, and predict the abnormal operating current situation during the charging of the electric vehicle battery through the current potential anomaly index.

[0019] Preferably, compare and analyze the current potential anomaly index generated when predicting the current operation trend during the charging of the electric vehicle battery through the neural network model under the monitoring window with a preset current potential anomaly index reference threshold, and divide the current situation within the monitoring window. The specific division steps are as follows:

[0020] If the current potential anomaly index is greater than the current potential anomaly index reference threshold, then divide the current situation within the monitoring window into the potential abnormal operation stage; if the current potential anomaly index is less than the current potential anomaly index reference threshold, then divide the current situation within the monitoring window into the potential abnormal operation stage.

[0021] Preferably, the specific steps for extracting the ratio of the amplitude to the frequency of the sharp increase in the current waveform from the current data set and generating a spike quantization reference value after analyzing the extracted features within the monitoring window are as follows:

[0022] Extract the sharp increase features in the current waveform from the current data set, including the sharp increase amplitude and the sharp increase frequency. The sharp increase amplitude is defined as the maximum instantaneous change value of the current between two adjacent time points, and the calculation expression is: , where is the current value at time and is the current value at time The current value, j represents the time point index, is the peak surge amplitude, which is used to quantify the severity of a single peak change, is the sampling time interval; the peak surge frequency is defined as the ratio of the number of peak surge events within the monitoring window to the window duration, and the calculation expression is: , where, is the peak surge frequency, indicating the density of abnormal fluctuations, is the duration of the monitoring window, is the count of peak surge events;

[0023] Introduce the time weighting coefficient , which is used to enhance the weight of recent peak events. The calculation expression of the time weighting coefficient is: , where, is the weight decay factor, which is used to control the time sensitivity;

[0024] Calculate the peak quantization reference value through a non-linear fusion formula in combination with the time weighting coefficient. The calculation expression is: , in the formula, is the peak quantization reference value, which quantifies the sudden increase degree and occurrence frequency of the instantaneous abnormality of the charging current, and are the weights of the normalized peak surge amplitude and the normalized peak surge frequency respectively, which adjust the influence of the peak surge amplitude and the peak surge frequency. The calculation expressions of the normalized peak surge amplitude and the normalized peak surge frequency are: ， where, and represent the normalized peak surge amplitude and the normalized peak surge frequency respectively, and represent the maximum values of the peak surge amplitude and the peak surge frequency within the monitoring window respectively.

[0025] Preferably, extract the continuous rising rate and cumulative amplitude of the current over a period of time from the current data set. The specific steps for generating the continuous rising quantization reference value after analyzing the extracted features within the monitoring window are as follows:

[0026] Within the monitoring window, analyze the current data set in the order of time series, and extract the continuous rising rate and cumulative amplitude of the current over a period of time. The continuous rising rate is defined as the ratio of the cumulative value of the current change rate within the monitoring window to the time, and the expression is: , in the formula, Represents the continuous rising rate of current over a period of time, and represent the current values at the k th and the k +1th time points within the monitoring window, n represents the total number of time points within the monitoring window, means only considering the rising part of the current, ignoring the decreasing or stable changes; the cumulative amplitude is defined as the cumulative sum of all current rising changes within the monitoring window, and the expression is: , where represents the cumulative amplitude of the current over a period of time. The cumulative amplitude represents the accumulation of all increments of the rising current and reflects the cumulative amplitude of the current deviation from the initial value;

[0027] By performing a non - linear combination analysis on the continuous rising rate of the current over a period of time and the cumulative amplitude , a continuous rising quantization reference value is generated to reflect the trend of the current deviating from the stable value in the long term. The generation expression of the continuous rising quantization reference value is: , where represents the continuous rising quantization reference value, which is used to quantify the degree of abnormality of the current deviating from the stable value in the long term, and are the weights of the continuous rising rate and the cumulative amplitude respectively, adjusting the influence of the continuous rising rate and the cumulative amplitude on the result.

[0028] Preferably, for the current situation classified as the potential abnormal operation stage, according to the prediction result of the neural network model, the actual monitoring frequency of current monitoring is dynamically increased. The specific steps are as follows:

[0029] In the potential abnormal operation stage, according to the current potential abnormal index generated by the neural network model prediction and the preset current potential abnormal index reference threshold, calculate the dynamic monitoring frequency adjustment factor. The dynamic monitoring frequency adjustment factor is determined by the amplitude of the current potential abnormal index exceeding the current potential abnormal index reference threshold. The calculation expression is: , where is the current potential abnormal index within the current monitoring window, which is predicted by the neural network and quantifies the degree of abnormal risk, is the current potential abnormal index reference threshold, which is used to distinguish the normal and potential abnormal operation stages, is the adjustment coefficient, which controls the sensitivity of the dynamic monitoring frequency adjustment factor, is the hyperbolic tangent function, which is used to smooth the change of the dynamic monitoring frequency adjustment factor;

[0030] After obtaining the dynamic monitoring frequency adjustment factor Based on the initial monitoring frequency, the actual monitoring frequency is calculated, and the calculation formula is: , where is the initial monitoring frequency, representing the basic monitoring frequency of the current during the normal operation stage, is the actual monitoring frequency after dynamic adjustment, which is used for current monitoring in the current potential abnormal stage;

[0031] In the potential abnormal stage, based on the calculated actual monitoring frequency , optimize the time interval of data acquisition, and adjust the data update strategy of the monitoring window in real time, so that the data points in the window are denser and can capture the abnormal change trend more precisely. The expression of the optimized data acquisition time interval is: , where is the optimized data acquisition time interval;

[0032] Based on the updated monitoring window data, further analyze the real-time collected data, and optimize the power output of the charging pile. The actual charging power of the optimized charging pile is: , where is the actual charging power after dynamic adjustment, which decreases as the abnormal risk increases to protect the battery safety, is the rated charging power, representing the charging power value under normal conditions, is the sensitivity parameter of power adjustment.

[0033] A system for multi-stage power distribution of a DC charging pile includes an initial monitoring module, a data update module, an anomaly prediction module, a stage division module, a normal monitoring module, and a dynamic adjustment module;

[0034] The initial monitoring module, according to historical charging data and monitoring requirements, sets an initial monitoring frequency in the first charging stage of the DC charging pile, and collects current data when the electric vehicle battery is charging at the initial monitoring frequency;

[0035] The data update module, based on data acquisition, introduces a time-sliding monitoring window, forms a current data set with the current data in the monitoring window in chronological order, and continuously updates the current data set as time progresses;

[0036] The anomaly prediction module, within the monitoring window, after performing anomaly analysis on the current data in the current data set, inputs the analyzed features into a pre-trained neural network model, and predicts the current operation trend when the electric vehicle battery is charging through the neural network model;

[0037] The stage division module divides the current situation within the monitoring window into a normal operation stage and a potential abnormal operation stage based on the output result of the neural network model;

[0038] The normal monitoring module continues to monitor the current during the charging of the electric vehicle battery in real time at the set initial monitoring frequency for the current situation classified as the normal operation stage;

[0039] The dynamic adjustment module, for the current situation classified as the potential abnormal operation stage, dynamically increases the actual monitoring frequency of current monitoring according to the prediction result of the neural network model, realizes accurate monitoring of abnormal current conditions, and reduces the charging power to reduce the current load.

[0040] In the above technical solution, the technical effects and advantages provided by the present invention are as follows:

[0041] By setting the initial monitoring frequency, the present invention realizes the efficient utilization of system resources, and combines the sliding window and the neural network model to dynamically analyze current data and accurately predict the operation trend, effectively distinguishing the normal operation stage and the potential abnormal stage, so as to realize intelligent monitoring. In the potential abnormal stage, dynamically increasing the monitoring frequency significantly improves the ability to capture abnormal current, avoiding battery damage or safety accidents caused by response delay. At the same time, appropriately reducing the charging power to reduce the current load, protecting the battery performance and extending its service life, thus comprehensively optimizing the monitoring accuracy and operation efficiency of the system on the basis of ensuring charging safety, and providing a reliable and intelligent technical solution for high-power charging scenarios. Description of the Drawings

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings.

[0043] Figure 1 It is a method flow chart of a multi-stage power distribution method for a DC charging pile of the present invention.

[0044] Figure 2 It is a module schematic diagram of a multi-stage power distribution system for a DC charging pile of the present invention. Detailed Embodiments

[0045] Now, the example embodiments will be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.

[0046] The present invention provides a method for multi-stage power distribution of a DC charging pile as Figure 1 shown, comprising the following steps:

[0047] According to historical charging data and monitoring requirements, set an initial monitoring frequency in the first charging stage (when the SOC is relatively low) of the DC charging pile, and collect current data during the charging of the electric vehicle battery at the initial monitoring frequency;

[0048] The initial monitoring frequency refers to a fixed data collection frequency preset when the DC charging pile conducts constant current charging on the electric vehicle battery, and is used to monitor the change of the battery charging current. This initial monitoring frequency determines the time interval for current data collection. For example, if data is collected once per second, the initial monitoring frequency is 1 Hz. It is a basic parameter for real-time monitoring during the charging process and is usually set to a moderate fixed value at the beginning of charging or when the system is operating normally to balance the real-time nature of data and resource usage efficiency.

[0049] Based on the data collection, introduce a time-sliding monitoring window, and organize the current data within the monitoring window into a current data set in chronological order, so that the current data set is continuously updated as time progresses;

[0050] Introducing a time-sliding monitoring window can dynamically organize the current data within a certain time range into a current data set and continuously update it as time goes by. The function of this mechanism is to capture the latest change trend of the current in real time during the charging process, ensuring that subsequent anomaly analysis and prediction are based on the current charging state rather than lagging or outdated data. The dynamic movement of the monitoring window enables the system to continuously obtain the latest data characteristics without increasing the additional data storage and processing burden, thereby improving the real-time nature and accuracy of the analysis and providing basic support for quickly responding to potential anomalies.

[0051] The sliding window refers to, during the data collection process, using a fixed time length as the window range, extracting a part of the data from the continuous current data stream for processing and analysis, and as time goes by, the window continuously moves forward, and the data contained therein is updated each time. The setting of the sliding window needs to comprehensively consider the real-time nature and data representativeness: the window length should be short enough to capture the latest current change trend to meet the real-time monitoring requirements, and at the same time, it should be long enough to cover the typical fluctuation period of the charging current to avoid missing key patterns or abnormal features. The step size of the window sliding determines the data update frequency and is usually set to a part of the window length to ensure continuity and efficiency.

[0052] In the monitoring window, after abnormal analysis of the current data in the current data set, the analyzed features are input into the pre-trained neural network model, and the current operation trend of the electric vehicle battery during charging is predicted through the neural network model;

[0053] In the monitoring window, the current data in the current data set is analyzed for abnormalities. The specific steps are as follows:

[0054] The ratio of the amplitude and frequency of the sudden increase in the peak value in the current waveform, the rate of continuous increase of the current over a period of time, and the cumulative amplitude are extracted from the current data set. After analyzing the extracted features, the peak quantization reference value and the continuous increase quantization reference value are generated respectively. The peak quantization reference value is used to reflect the degree and frequency of the instantaneous abnormal increase in the charging current, and the possible overload risk or control disorder problem of the system is evaluated. The continuous increase quantization reference value is used to reflect the long-term deviation of the current from the stable value, and to evaluate whether the battery or charging system has entered a potential overload or unstable state.

[0055] After obtaining the peak quantization reference value and the continuous rise quantization reference value obtained by abnormal analysis of the current data in the current data set, the peak quantization reference value and the continuous rise quantization reference value are input into a pre-trained neural network model, and a current potential abnormality index is generated through the model. The abnormal operating current situation of the electric vehicle battery during charging is predicted through the current potential abnormality index.

[0056] A pre-trained neural network model refers to a model obtained after sufficient training with a historical current data set and its features (such as spike quantization reference values, continuous rise quantization reference values, etc.). The model has a certain generalization ability and can learn the complex laws and potential patterns of current anomalies from input features. During the training process, the neural network continuously optimizes its parameters by processing a large amount of labeled data (including normal current and abnormal current samples), so that it can accurately distinguish between normal operating conditions and abnormal operating conditions. The structure of the model usually includes multiple layers of neuron units (such as multilayer perceptrons or convolutional neural networks), which mine deep associations in the input data through nonlinear transformations, thereby realizing the recognition of complex abnormal patterns.

[0057] In the above application, the peak quantization reference value and the continuous rise quantization reference value are used as inputs to the neural network model. Combined with historical training data, the model can learn how to map these indices to the current potential anomaly index. This mapping relationship takes into account the interaction between features, such as whether a short-term surge in the peak quantization reference value is accompanied by a continuous upward trend, or the contribution of different feature combinations to the anomaly. Through the deep structure of the neural network, the model can capture the nonlinear patterns hidden in the input features, making the prediction of complex current anomalies more accurate.

[0058] The core significance of this neural network model lies in enhancing the accuracy and real-time performance of anomaly prediction. First of all, through pre-training, the neural network can identify complex anomaly patterns that are difficult to capture by traditional rules or statistical methods. For example, a single anomaly such as a spike quantization reference value and a continuously rising quantization reference value may not be significant, but their combined anomaly may indicate that the charging current is entering a dangerous zone. Through the non-linear combination analysis of the neural network, the model can extract key patterns from multi-dimensional features, thereby generating an accurate potential current anomaly index.

[0059] Secondly, the adaptive learning ability of the neural network enables it to dynamically adjust the prediction performance according to actual data. For example, in the case of changes in electric vehicle battery types, charging pile models, or charging environmental conditions, the model can maintain high efficiency through continuous optimization. This flexibility is particularly applicable to environments with complex and variable charging scenarios. The potential current anomaly index generated by the neural network model provides a quantitative reference value for the charging system. Therefore, the pre-trained neural network model can not only accurately predict anomalies, but also provide core technical support for the operation of the intelligent charging system through in-depth analysis and decision-making support.

[0060] When charging an electric vehicle battery through a DC charging pile, a relatively high ratio of the amplitude to the frequency of the spike increase in the current waveform usually indicates that there may be a potential current anomaly when the DC charging pile is charging the electric vehicle battery. The amplitude of the spike increase reflects the degree of drastic change in the current within a short period of time, and the frequency ratio refers to the frequency of such drastic changes. If this ratio is relatively high, it may indicate problems with the current output stability of the charging system, such as response delays in the control module of the charging pile, unstable current regulation, or external interference (such as grid fluctuations) affecting the charging process. Continuous current spikes will not only increase the heat accumulation in the battery, affecting its lifespan, but also pose a threat to the safety of the charging equipment and the battery.

[0061] The specific steps for extracting the ratio of the amplitude to the frequency of the spike increase in the current waveform from the current data set and generating a spike quantization reference value after analyzing the extracted features within the monitoring window are as follows:

[0062] Extract the spike increase features in the current waveform from the current data set, including the spike increase amplitude and the spike increase frequency. The spike increase amplitude is defined as the maximum instantaneous change value of the current between two adjacent time points, and the calculation expression is: , where is the current value at time , is the current value at time , j represents the time point index, is the spike increase amplitude, used to quantify the severity of a single spike change, is the sampling time interval; the spike increase frequency is defined as the ratio of the number of spike increase events within the monitoring window to the window duration, and the calculation expression is: , where is the spike increase frequency, indicating the density of abnormal fluctuations, is the duration of the monitoring window, is the count of spike increase events;

[0063] Introduce the time weighting coefficient , which is used to enhance the weight of recent spike events. The calculation expression of the time weighting coefficient is: , where is the weight decay factor, which is used to control the time sensitivity;

[0064] Combine the time weighting coefficient to calculate the spike quantization reference value through a non-linear fusion formula. The calculation expression is: , in the formula, is the spike quantization reference value, which quantifies the sharp increase degree and its occurrence frequency of the instantaneous abnormal charging current, and are the weights of the normalized spike increase amplitude and the normalized spike increase frequency respectively, which adjust the influence of the spike increase amplitude and the spike increase frequency. The calculation expressions of the normalized spike increase amplitude and the normalized spike increase frequency are: ， where and represent the normalized spike increase amplitude and the normalized spike increase frequency respectively, and represent the maximum values of the spike increase amplitude and the spike increase frequency within the monitoring window respectively;

[0065] The non-linear fusion formula enhances the sensitivity to high-amplitude and high-frequency spikes through feature power operations, improving the accuracy of anomaly detection.

[0066] From the peak quantization reference value, it can be seen that under the monitoring window, the larger the performance value of the peak quantization reference value generated after analyzing the ratio of the amplitude to the frequency of the sharp increase in the current waveform extracted from the current data set, the greater the potential hidden danger of abnormal current when the DC charging pile charges the electric vehicle battery. On the contrary, it indicates that the hidden danger is smaller. The peak quantization reference value comprehensively evaluates the degree of violent current fluctuation and its frequency of occurrence during the charging process by quantifying the ratio of the amplitude to the frequency of the sharp increase in the current waveform. When the peak quantization reference value is large, it means that the current has increased sharply and fluctuated multiple times in a short period, which may be caused by abnormal control system response, circuit fluctuation or hardware failure, indicating that the potential overload risk or system imbalance problem is more serious. When the peak quantization reference value is small, it means that the current fluctuation is small and relatively stable, and the charging process is safer and smoother. Therefore, the peak quantization reference value can be used as an important indicator to evaluate the stability of the charging current and provide a warning basis for potential abnormalities.

[0067] The rate of continuous increase and the cumulative amplitude of the current over a period of time are relatively high, indicating potential current abnormalities when the DC charging pile charges the electric vehicle battery. This phenomenon usually means that the current output of the charging system fails to remain constant, which may be due to abnormal controller of the charging pile, feedback delay of the battery management system (BMS), or out-of-control current regulation caused by line faults. The continuously increasing current will increase the thermal load of the battery, leading to safety hazards such as internal material deterioration and thermal runaway. At the same time, the excessive or too large accumulation of current may exceed the designed tolerance range of the battery, having an adverse impact on the battery life and the stability of the charging equipment.

[0068] The specific steps to extract the rate of continuous increase and the cumulative amplitude of the current over a period of time from the current data set and generate the continuous increase quantization reference value after analyzing the extracted features within the monitoring window are as follows:

[0069] Within the monitoring window, for the current data set Analyze it in the order of time series, extract the continuous increase rate and the cumulative amplitude of the current over a period of time. The continuous increase rate Is defined as the ratio of the cumulative value of the current change rate within the monitoring window to the time, and the expression is: , where Represents the continuous increase rate of the current over a period of time, And Represent the current values at the k rd and the k +1th time points within the monitoring window, n Represents the total number of time points within the monitoring window, It means only considering the rising part of the current, ignoring the falling or steady changes. The continuous rising rate reflects the rate at which the current continuously rises per unit time. The higher the rate, the more likely there may be problems with the system's regulation ability; the cumulative amplitude is defined as the cumulative sum of all current rising changes within the monitoring window, and the expression is: , where represents the cumulative amplitude of the current over a period of time. The cumulative amplitude represents the accumulation of all increments of the current rise and reflects the cumulative amplitude of the current deviation from the initial value. The larger the cumulative amplitude, the more significant the fluctuation amplitude of the current during this time period, which may indicate potential abnormalities in the charging system.

[0070] By performing non - linear combination analysis on the continuous rising rate and the cumulative amplitude of the current over a period of time, a continuous rising quantization reference value is generated to reflect the trend of the current deviating from the stable value in the long term. The generation expression of the continuous rising quantization reference value is: , where represents the continuous rising quantization reference value, which is used to quantify the degree of abnormality of the current deviating from the stable value in the long term. The larger the value, the more obvious the abnormal trend. Logarithmic transformation is used to compress the exponential range, enhance the sensitivity to small - scale abnormalities, and at the same time suppress the influence of extreme values. and are the weights of the continuous rising rate and the cumulative amplitude respectively, which adjust the influence of the continuous rising rate and the cumulative amplitude on the result.

[0071] It can be seen from the continuous rising quantization reference value that under the monitoring window, the larger the performance value of the continuous rising quantization reference value generated by analyzing the characteristics of the continuous rising rate and the cumulative amplitude of the current over a period of time, the greater the potential hidden danger of abnormal current when the DC charging pile charges the electric vehicle battery. On the contrary, it indicates a smaller hidden danger. The continuous rising quantization reference value comprehensively quantifies the rising rate and the cumulative amplitude of the current within the monitoring window. A large performance value indicates that the trend of the current deviating from the stable value in a period of time is more significant, which may indicate insufficient control ability of the current regulation system, or the charging system enters an overloaded or abnormal operating state. When the continuous rising quantization reference value is small, it indicates that the current change tends to be stable, no significant deviation phenomenon occurs, and the charging system is in a relatively safe and normal operating state.

[0072] The neural network model is not specifically limited here. It can realize the comprehensive analysis of the spike quantization reference value obtained by performing abnormal analysis on the current data and the continuous rising quantization reference value to generate the current potential anomaly index Any neural network model can be used. To implement the technical solution of the present invention, the present invention provides a specific implementation; the current potential anomaly index The calculation formula for generation is: , where in the formula, and are respectively the preset proportionality coefficients of the spike quantization reference value and the continuously rising quantization reference value , and and are both greater than 0.

[0073] It can be seen from the current potential anomaly index that under the monitoring window, the larger the value of the spike quantization reference value generated by analyzing the ratio characteristic of the amplitude and frequency of the spike sudden increase in the current waveform extracted from the current data set, and the larger the value of the continuously rising quantization reference value generated by analyzing the rate and cumulative amplitude characteristics of the current continuously rising over a period of time. That is, when predicting the current operation trend during the charging of the electric vehicle battery through the neural network model under the monitoring window, the larger the value of the current potential anomaly index generated, it indicates that the potential current anomaly hidden danger during the charging of the DC charger to the electric vehicle battery is greater. On the contrary, it indicates that the potential current anomaly hidden danger during the charging of the DC charger to the electric vehicle battery is smaller.

[0074] Based on the output result of the neural network model, the current situation within the monitoring window is divided into a normal operation stage and a potential anomaly operation stage;

[0075] Compare and analyze the current potential anomaly index generated when predicting the current operation trend during the charging of the electric vehicle battery through the neural network model under the monitoring window with the pre-set current potential anomaly index reference threshold, and divide the current situation within the monitoring window. The specific division steps are as follows:

[0076] If the current potential anomaly index is greater than the current potential anomaly index reference threshold, it indicates that the current during the charging of the DC charger to the electric vehicle battery may have a potential anomaly. Then, divide the current situation within this monitoring window into the potential anomaly operation stage; if the current potential anomaly index is less than the current potential anomaly index reference threshold, it indicates that the current change during the charging of the DC charger to the electric vehicle battery meets the expectation and there is no obvious anomaly sign. Then, divide the current situation within this monitoring window into the potential anomaly operation stage.

[0077] Dividing the stage helps the system adopt corresponding monitoring strategies for different operation states, improving the flexibility and effectiveness of monitoring.

[0078] For the current situation divided into the normal operation stage, continue to perform real-time monitoring on the current during the charging of the electric vehicle battery at the set initial monitoring frequency;

[0079] For the current situation classified as the normal operation stage, continue to monitor the current during the charging of the electric vehicle battery in real time at the set initial monitoring frequency. The function is to achieve efficient resource management and stable charging monitoring. By maintaining the initial monitoring frequency, the system can meet the requirements of tracking normal current fluctuations while avoiding the computational pressure and increased energy consumption caused by too high a monitoring frequency, thereby optimizing the overall system performance. The current fluctuation is small during the normal operation stage. Maintaining the initial frequency can accurately capture the necessary change trends, ensure the stable operation of the charging system, and reduce the burden on the hardware and data processing modules.

[0080] For the current situation classified as the potential abnormal operation stage, according to the prediction results of the neural network model, dynamically increase the actual monitoring frequency of current monitoring, achieve accurate monitoring of abnormal current situations, and reduce the charging power to reduce the current load;

[0081] For the current situation classified as the potential abnormal operation stage, according to the prediction results of the neural network model, dynamically increase the actual monitoring frequency of current monitoring. The specific steps are as follows:

[0082] In the potential abnormal operation stage, according to the current potential abnormal index generated by the prediction of the neural network model and the preset reference threshold of the current potential abnormal index, calculate the dynamic monitoring frequency adjustment factor. The dynamic monitoring frequency adjustment factor is determined by the amplitude of the current potential abnormal index exceeding the reference threshold of the current potential abnormal index. The calculation expression is: , where is the current potential abnormal index within the current monitoring window, which is predicted by the neural network and quantifies the degree of abnormal risk. is the reference threshold of the current potential abnormal index, which is used to distinguish the normal and potential abnormal operation stages. is the adjustment coefficient, which controls the sensitivity of the dynamic monitoring frequency adjustment factor. The larger the value of , the more obvious the change in the monitoring frequency. is the hyperbolic tangent function, which is used to smooth the change of the dynamic monitoring frequency adjustment factor and avoid too large a frequency adjustment amplitude due to too high a current potential abnormal index;

[0083] Dynamically calculate the degree of increase in the monitoring frequency through the dynamic monitoring frequency adjustment factor to ensure flexible response to potential abnormalities.

[0084] After obtaining the dynamic monitoring frequency adjustment factor , calculate the actual monitoring frequency according to the initial monitoring frequency. The calculation expression is: , where is the initial monitoring frequency, representing the basic monitoring frequency of the current during the normal operation stage, usually a fixed value. is the actual monitoring frequency after dynamic adjustment, used for current monitoring in the current potential abnormal stage;

[0085] Dynamic actual monitoring frequency directly determines the monitoring fineness. The higher the monitoring frequency, the more accurately it can capture the subtle changes in current abnormalities.

[0086] In the potential abnormal stage, according to the calculated actual monitoring frequency , optimize the time interval of data acquisition, and adjust the data update strategy of the monitoring window in real time, making the data points in the window denser and capturing the abnormal change trend more finely. The expression of the optimized data acquisition time interval is: , where is the optimized data acquisition time interval;

[0087] By adjusting the time interval, the data acquisition frequency dynamically responds to potential abnormal situations, providing a denser sampling basis for subsequent real-time analysis.

[0088] Based on the updated monitoring window data, further analyze the real-time collected data, and optimize the power output of the charging pile. The actual charging power of the optimized charging pile is: , where is the actual charging power after dynamic adjustment, which decreases as the abnormal risk increases to protect the battery safety. is the rated charging power, representing the charging power value under normal conditions. Sensitivity parameter for power adjustment;

[0089] While accurately monitoring, reduce the current load by reducing the charging power to prevent the battery from overheating or overloading, forming a closed-loop strategy for monitoring and control.

[0090] For the current situation divided into the potential abnormal operation stage, the steps of dynamically increasing the monitoring frequency and reducing the charging power can effectively improve the ability to capture current anomalies, while reducing the battery load and preventing further deterioration. The core function of this strategy is to obtain detailed data on current changes in a timely manner through precise monitoring, so as to quickly identify the nature and trend of potential anomalies and provide high-resolution input for system decision-making. At the same time, by reducing the charging power to reduce the current load, it is possible to avoid overheating inside the battery and material degradation, reduce the risk of thermal runaway and damage, and protect the battery life. Adopting a higher monitoring frequency and power limit strategy during the abnormal stage is the best choice to balance safety and efficiency, ensuring that the system always operates within the safety margin before potential risks are eliminated. Such a design not only improves the safety and stability of the charging process, but also provides reliable data support for subsequent anomaly handling and regulation.

[0091] Through this multi-stage power distribution scheme for DC charging piles, the problem of delayed capture of current anomalies caused by traditional fixed monitoring frequencies is solved, while improving the safety and efficiency of the charging process. First, set the initial monitoring frequency to ensure the efficient use of system resources. The combination of the sliding window and the neural network model can dynamically analyze current data and predict the operation trend, accurately distinguish between normal and potential abnormal stages, and achieve intelligent monitoring. For the potential abnormal stage, dynamically increasing the monitoring frequency significantly enhances the capture accuracy of abnormal currents, avoiding battery damage or safety accidents caused by delayed response. At the same time, by reducing the charging power to reduce the current load, the battery is effectively protected and its life is extended. This scheme not only improves the real-time monitoring ability of the charging process, but also optimizes the system operation efficiency on the premise of ensuring safety, providing reliable technical support for high-power charging scenarios.

[0092] The present invention provides a system for multi-stage power distribution of a DC charging pile as shown in Figure 2 including an initial monitoring module, a data update module, an anomaly prediction module, a stage division module, a normal monitoring module, and a dynamic adjustment module;

[0093] The initial monitoring module sets an initial monitoring frequency in the first charging stage of the DC charging pile according to historical charging data and monitoring requirements, and collects current data when the electric vehicle battery is charged at the initial monitoring frequency;

[0094] The data update module, on the basis of data collection, introduces a time-sliding monitoring window, forms a current data set by arranging the current data within the monitoring window in chronological order, and makes the current data set continuously updated as time progresses;

[0095] Anomaly prediction module, within the monitoring window, after performing anomaly analysis on the current data in the current data set, inputs the analyzed features into a pre-trained neural network model, and predicts the current operation trend during the charging of the electric vehicle battery through the neural network model;

[0096] Stage division module, based on the output result of the neural network model, divides the current situation within the monitoring window into a normal operation stage and a potential anomaly operation stage;

[0097] Normal monitoring module, for the current situation classified as the normal operation stage, continues to perform real-time monitoring of the current during the charging of the electric vehicle battery at the set initial monitoring frequency;

[0098] Dynamic adjustment module, for the current situation classified as the potential anomaly operation stage, according to the prediction result of the neural network model, dynamically increases the actual monitoring frequency of current monitoring, realizes precise monitoring of abnormal current situations, and reduces the charging power to reduce the current load;

[0099] A method for multi-stage power distribution of a DC charging pile provided by an embodiment of the present invention is implemented through the above-mentioned system for multi-stage power distribution of a DC charging pile. The specific method and process of the system for multi-stage power distribution of a DC charging pile are detailed in the embodiment of the above-mentioned method for multi-stage power distribution of a DC charging pile, and will not be elaborated here.

[0100] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0101] Only some exemplary embodiments of the present invention have been described by way of illustration above. Undoubtedly, for those of ordinary skill in the art, without departing from the spirit and scope of the present invention, the described embodiments can be modified in various different ways. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of the claims of the present invention.

[0102] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0103] 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 foregoing method embodiments and will not be elaborated herein.

[0104] In several embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, 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 displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0105] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0106] In addition, in each embodiment of the present application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0107] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0108] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims described above.

Claims

1. A method for multi-stage power distribution of a DC charging pile, characterized in that: The following steps are involved: According to historical charging data and monitoring requirements, an initial monitoring frequency is set in the first charging stage of the DC charging pile, and the current data of the electric vehicle battery during charging is collected at the initial monitoring frequency; On the basis of data collection, a time sliding monitoring window is introduced, and the current data in the monitoring window is organized into a current data set in time sequence, so that the current data set is continuously updated with the progress of time; In the monitoring window, after abnormal analysis of the current data in the current data set, the analyzed features are input into the pre-trained neural network model, and the current operation trend of the electric vehicle battery during charging is predicted through the neural network model; Based on the output results of the neural network model, the current situation in the monitoring window is divided into a normal operation stage and a potential abnormal operation stage; For the current situation classified as the normal operation stage, the current when the electric vehicle battery is charged continues to be monitored in real time at the set initial monitoring frequency; For current conditions that are classified as potential abnormal operation stages, the actual monitoring frequency of current monitoring is dynamically increased according to the prediction results of the neural network model, so as to accurately monitor abnormal current conditions and reduce charging power and current load. In the monitoring window, the current data in the current data set is analyzed for abnormalities. The specific steps are as follows: The ratio of the amplitude and frequency of the sudden increase of the peak in the current waveform and the rate and cumulative amplitude of the continuous increase of the current over a period of time are extracted from the current data set. After analyzing the extracted features, the peak quantization reference value and the continuous increase quantization reference value are generated respectively. The peak quantization reference value reflects the degree and frequency of the instantaneous abnormal increase of the charging current; the continuous increase quantization reference value reflects the trend of the current deviating from the stable value for a long time; After obtaining the peak quantization reference value and the continuous rise quantization reference value obtained by abnormal analysis of the current data in the current data set, the peak quantization reference value and the continuous rise quantization reference value are input into a pre-trained neural network model, and a current potential abnormality index is generated through the model, and the abnormal operating current situation of the electric vehicle battery during charging is predicted through the current potential abnormality index; For the current conditions classified as potential abnormal operation stages, the actual monitoring frequency of current monitoring is dynamically increased according to the prediction results of the neural network model. The specific steps are as follows: In the potential abnormal operation stage, the dynamic monitoring frequency adjustment factor is calculated according to the current potential abnormal index generated by the neural network model prediction and the preset current potential abnormal index reference threshold. The dynamic monitoring frequency adjustment factor is determined according to the magnitude of the current potential abnormal index exceeding the current potential abnormal index reference threshold. The calculation expression is: , where is the potential abnormality index of the current in the current monitoring window, which is predicted by the neural network and quantifies the degree of abnormal risk. is the reference threshold of the current potential abnormality index, which is used to distinguish between normal and potential abnormal operation stages. To adjust the coefficient, control the sensitivity of the dynamic monitoring frequency adjustment factor, is a hyperbolic tangent function, used to smoothly monitor the change of the frequency adjustment factor; In order to obtain the dynamic monitoring frequency adjustment factor After that, the actual monitoring frequency is calculated according to the initial monitoring frequency. The calculation expression is: , where is the initial monitoring frequency, indicating the basic monitoring frequency of the current during normal operation. The actual monitoring frequency after dynamic adjustment is used for current monitoring at the current potential abnormal stage; In the potential abnormal stage, according to the actual monitoring frequency calculated , optimize the time interval of data collection, and adjust the data update strategy of the monitoring window in real time, so that the data points in the window are denser and the abnormal change trend is captured more finely. The expression of the optimized data collection time interval is: , where is the optimized data collection time interval; Based on the updated monitoring window data, the real-time collected data is further analyzed to optimize the power output of the charging pile. The actual charging power of the optimized charging pile is: , where The actual charging power after dynamic adjustment will be reduced as the abnormal risk increases to protect the battery safety. is the rated charging power, which indicates the charging power value under normal conditions. Sensitivity parameters for power adjustment; The specific steps of extracting the ratio of the amplitude and frequency of the spike in the current waveform from the current data set and analyzing the extracted features within the monitoring window to generate the spike quantization reference value are as follows: Extracting spike features in the current waveform from the current data set, including spike amplitude and spike frequency, wherein the spike amplitude is defined as the maximum instantaneous change value of the current between two adjacent time points; The spike burst frequency is defined as the ratio of the number of spike burst events in the monitoring window to the window length; Introduce a time weighting coefficient to enhance the weight of recent peak events; The spike quantization reference value is calculated by combining the time weighting coefficient through a nonlinear fusion formula; The specific steps of extracting the rate and cumulative amplitude of the continuous increase of current over a period of time from the current data set and analyzing the extracted features within the monitoring window to generate the continuous increase quantitative reference value are as follows: In the monitoring window, the current data set is analyzed in time series order to extract the continuous rising rate and cumulative amplitude of the current over a period of time. The continuous rising rate is defined as the ratio of the cumulative value of the current change rate in the monitoring window to the time. The cumulative amplitude is defined as the cumulative sum of all current rise changes within the monitoring window; By performing nonlinear combination analysis on the continuous rising rate and cumulative amplitude of the current over a period of time, a continuous rising quantitative reference value is generated to reflect the long-term trend of the current deviating from the stable value.

2. A method for multi-stage power distribution of a DC charging pile according to claim 1, characterized in that: The initial monitoring frequency refers to the pre-set fixed data collection frequency when the DC charging pile charges the electric vehicle battery with a constant current, which is used to monitor the changes in the battery charging current.

3. A method for multi-stage power distribution of a DC charging pile according to claim 1, characterized in that: Sliding window means that during the data acquisition process, a fixed time length is used as the window range, a part of the data is extracted from the continuous current data stream for processing and analysis, and the window moves forward continuously over time, and each movement updates the data contained therein.

4. The method for multi-stage power distribution of a DC charging pile according to claim 1, characterized in that: The current potential abnormality index generated when predicting the current operation trend of the electric vehicle battery charging under the monitoring window through the neural network model is compared and analyzed with the preset current potential abnormality index reference threshold, and the current situation in the monitoring window is divided. The specific division steps are as follows: If the current potential abnormality index is greater than the current potential abnormality index reference threshold, the current situation in the monitoring window is divided into the potential abnormal operation stage; if the current potential abnormality index is less than the current potential abnormality index reference threshold, the current situation in the monitoring window is divided into the potential abnormal operation stage.

5. A system for multi-stage power distribution of a DC charging pile, used to implement the method for multi-stage power distribution of a DC charging pile as described in any one of claims 1 to 4, characterized in that: It includes initial monitoring module, data updating module, abnormal prediction module, stage division module, normal monitoring module and dynamic adjustment module; The initial monitoring module sets the initial monitoring frequency in the first charging stage of the DC charging pile according to the historical charging data and monitoring requirements, and collects the current data of the electric vehicle battery during charging at the initial monitoring frequency; The data update module introduces a time sliding monitoring window based on data acquisition, and organizes the current data in the monitoring window into a current data set in chronological order, so that the current data set is continuously updated over time; The abnormality prediction module, in the monitoring window, performs abnormal analysis on the current data in the current data set, inputs the analyzed features into a pre-trained neural network model, and predicts the current operation trend of the electric vehicle battery when charging through the neural network model; The stage division module divides the current situation in the monitoring window into a normal operation stage and a potential abnormal operation stage based on the output results of the neural network model; The normal monitoring module, for the current situation classified as the normal operation stage, continues to monitor the current when the electric vehicle battery is charged in real time at the set initial monitoring frequency; The dynamic adjustment module dynamically increases the actual monitoring frequency of current monitoring for current conditions classified as potential abnormal operation stages based on the prediction results of the neural network model, achieves accurate monitoring of abnormal current conditions, and reduces charging power to reduce current load.

Images