A control method and system for electric vehicle charging facilities

Abstract

The application belongs to the technical field of data processing, and specifically discloses a control method and system for electric vehicle charging facilities, which comprises the following steps: collecting real-time voltage and current during the charging process, calculating the contact resistance through difference calculation, and comparing the contact resistance deviation AR(t) with the preset nominal value; performing sliding window processing on AR(t), extracting nonlinear characteristic parameters such as average deviation, fluctuation variance and rising rate; based on these parameters, predicting the charging contact point deterioration rate λ(t) by means of a time series prediction model; calculating the virtual life consumption L(t) according to the cumulative λ(t), and then determining the normal, transition and key three charging power control stages according to L(t), and respectively performing high-power maintenance, graded power reduction and forced power limitation control. The application solves the problems of response lag and single control of traditional temperature monitoring, balances safety and user experience, and prolongs the service life of the facility.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of data processing, and particularly relates to a control method and system for electric vehicle charging facilities. BACKGROUND

[0002] In the running process of the existing charging facilities, the contact point is prone to local deterioration due to current conduction and temperature rise effect, which is manifested as abnormal increase of contact resistance. The traditional monitoring method mainly depends on temperature sensors, and only triggers power limitation when the temperature exceeds the threshold value. However, the temperature signal has a response lag, and it is difficult to reflect the early trend of deterioration, which leads to lag of protection measures and increases the life risk of charging facilities and user equipment. In addition, the traditional control method usually reduces power or stops running, ignoring the dynamic change of the deterioration rate, and lacks flexibility and prediction ability, which is easy to cause contradiction between safety and user experience. SUMMARY

[0003] To solve the above technical problems, the present application provides a control method and system for electric vehicle charging facilities.

[0004] The present application provides a control method for electric vehicle charging facilities, comprising the following method steps:

[0005] Collecting real-time voltage and real-time current in the charging process;

[0006] Calculating the contact resistance based on the difference between the real-time voltage and the real-time current, comparing the contact resistance with the preset nominal value under normal conditions to obtain the contact resistance deviation ΔR(t);

[0007] Sliding window processing the contact resistance deviation ΔR(t) to extract nonlinear feature parameters, the nonlinear feature parameters including average deviation, fluctuation variance and rising rate;

[0008] Based on at least the nonlinear feature parameters, predicting the charging contact point deterioration rate λ(t) through a time series prediction model;

[0009] According to the deterioration rate λ(t), calculating the virtual life consumption L(t) of the charging contact point;

[0010] Based on the virtual life consumption L(t), determining the charging power control stage, and performing control of maintaining high power, step-down power and forced power limitation respectively based on the control stage, wherein the control stage includes normal stage, transition stage and critical stage.

[0011] Further, the transition stage adopts a step-down power, each time the power reduction amplitude is a preset percentage of the rated power, and is gradually adjusted in minutes, wherein the power reduction amplitude is dynamically adjusted based on the degradation rate λ(t).

[0012] Further, based on the combined relationship of the virtual life consumption L(t) and the degradation rate λ(t), the charging control is performed, specifically including: when L(t) is low consumption and λ(t) is low rate, normal charging is performed; when L(t) is high consumption and λ(t) is low rate, conservative power reduction is performed; when L(t) is low consumption and λ(t) is high rate, fast power limiting is performed; when L(t) is high consumption and λ(t) is high rate, forced shutdown or alarm is performed, wherein the low consumption and the high consumption are determined based on preset thresholds.

[0013] Further, based on the service life of the charging facility, the low consumption determination criterion in the combined relationship is dynamically adjusted, the longer the service life, the lower the preset threshold corresponding to the low consumption.

[0014] Further, the degradation rate λ(t) of the charging contact point is predicted by the time series prediction model, and the predicted value of λ(t) is updated based on a preset time period, wherein the update frequency is adaptively adjusted based on the stability of the charging process, and the greater the charging current fluctuation, the higher the update frequency.

[0015] Further, the virtual life consumption L(t) of the charging contact point is calculated according to the degradation rate λ(t), and further includes introducing a temperature correction factor to adjust the accumulation speed of L(t), wherein the temperature correction factor is related to a reference temperature and a temperature coefficient of the contact point material.

[0016] In another aspect, the application also provides a control system for an electric vehicle charging facility, comprising:

[0017] The acquisition module is configured to acquire real-time voltage and real-time current during the charging process.

[0018] The contact resistance acquisition module is configured to calculate the contact resistance based on the difference between the real-time voltage and the real-time current, compare the contact resistance with a preset nominal value under normal conditions, and obtain a contact resistance deviation ΔR(t).

[0019] The nonlinear feature extraction module is configured to perform sliding window processing on the contact resistance deviation ΔR(t) to extract nonlinear feature parameters, including average deviation, fluctuation variance, and rising rate.

[0020] The degradation rate prediction module is configured to predict the degradation rate λ(t) of the charging contact point based on at least the nonlinear feature parameters by a time series prediction model.

[0021] a virtual life consumption amount calculation module configured to calculate a charging contact point virtual life consumption amount L(t) according to the degradation rate λ(t);

[0022] a power control module configured to determine a charging power control phase based on the virtual life consumption amount L(t), and perform control of maintaining high power, step-down power and forced limited power respectively based on the control phase, wherein the control phase includes a normal phase, a transition phase and a critical phase.

[0023] In another aspect, the present application provides an electronic device, comprising: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the control method for an electric vehicle charging facility as described above.

[0024] In another aspect, the present application provides a computer readable storage medium having computer program instructions stored thereon, wherein the computer program instructions are executable by a processor to implement the control method for an electric vehicle charging facility as described above.

[0025] In another aspect, the present application provides a computer program product comprising a computer program, wherein the computer program is executed by a processor to implement the control method for an electric vehicle charging facility as described above.

[0026] The present application solves the problems of traditional charging facilities relying on temperature monitoring response lag and single control by collecting voltage and current differential contact resistance deviation, extracting features by sliding window, combining GRU-D model to predict degradation rate, accumulating virtual life consumption amount, and performing power control in stages, and further introducing temperature correction factor and dynamic threshold adjustment. Early prediction and accurate quantification of contact point degradation are realized, protection lag is avoided, safety and user experience are balanced through step-down power and other strategies, no additional hardware is needed, cost is reduced, facility life is extended, and charging safety and practicality are improved. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be considered as limiting the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0028] Other features, objects and advantages of the present application will become more apparent from the following detailed description of the non-limiting embodiments, made with reference to the accompanying drawings.

[0029] Figure 1 This is a flowchart of a control method for electric vehicle charging facilities provided in an embodiment of the present invention.

[0030] Figure 2 A schematic diagram of the process for calculating contact resistance is provided for embodiments of the present invention.

[0031] Figure 3 This is a schematic diagram of a control system for electric vehicle charging facilities provided in an embodiment of the present invention.

[0032] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0034] The technical solutions of this application will be described in detail below with reference to various embodiments.

[0035] like Figure 1 As shown, this embodiment of the invention discloses a control method 100 for electric vehicle charging facilities, comprising the following method steps:

[0036] S101, collects real-time voltage and real-time current during the charging process;

[0037] S102, the contact resistance is calculated based on the difference between the real-time voltage and the real-time current. The contact resistance is compared with the preset nominal value under normal conditions to obtain the contact resistance deviation ΔR(t).

[0038] S103, perform sliding window processing on the contact resistance deviation ΔR(t) to extract nonlinear characteristic parameters, including average deviation, fluctuation variance and rise rate;

[0039] S104, based at least on the aforementioned nonlinear characteristic parameters, the degradation rate λ(t) of the charging contact point is predicted using a time series prediction model;

[0040] S105, calculate the virtual lifetime consumption L(t) of the charging contact point based on the degradation rate λ(t);

[0041] S106, based on the virtual life consumption L(t), determine the charging power control stage, based on the control stage, respectively, maintain high power, hierarchical power and forced power control, wherein the control stage includes normal stage, transition stage and key stage.

[0042] In some embodiments, for S101, according to the requirement of local deviation calculation of contact resistance according to Ohm's law, the collected parameters include two types of core electrical parameters, specifically including,

[0043] Real-time voltage U(t), collect the voltage signal between the contact points in the charging loop (such as the contact part of the charging gun and the vehicle interface, the contact part of the charging module output and the bus), instead of only collecting the overall loop voltage, to avoid the interference of the line voltage drop of non-contact points on the calculation of contact resistance, the voltage collection range needs to cover the rated voltage interval of the charging facility (such as 200V-1000V for DC fast charging facility, 220V / 380V for AC slow charging), and the accuracy reaches 0.1%FS (full scale), to ensure that the small change of ΔU(t) can be captured.

[0044] Real-time current I(t), synchronously collect the loop current flowing through the contact point, the collection range needs to match the rated current of the facility (such as 100A-500A for DC fast charging, 10A-32A for AC slow charging), the accuracy needs to reach 0.05%FS, and it needs to be time-synchronized with the voltage collection to avoid the calculation deviation of ΔU(t) / ΔI(t) caused by the asynchronous sampling of current and voltage.

[0045] Similarly, for the sampling frequency, support the dynamic calculation of contact resistance, the deterioration of the contact point may show as "sudden increase-slow decline" instantaneous fluctuation, if the sampling frequency is too low (such as seconds), it is easy to miss the instantaneous fluctuation signal, resulting in the missed judgment of early deterioration trend; therefore, the sampling frequency needs to be set to 50ms-500ms (i.e. 2Hz-20Hz), for example, using 100ms sampling interval, collecting 10 groups of voltage and current data per second, which can capture the instantaneous change of ΔR(t) and will not cause data redundancy due to too high frequency.

[0046] At the same time, support the extraction of sliding window characteristics, and subsequently need to extract the average deviation and fluctuation variance of ΔR through sliding window regression (claim 1), if the sliding window is set to 10 sampling points (common window size), 100ms sampling interval can make the window cover 1 second time dimension, which can reflect the short-term fluctuation (such as the change of ΔR within 1 second), and can avoid the instability of characteristics caused by too short window time.

[0047] Optionally, due to the problems of electromagnetic interference, sampling noise and the like in the charging environment, the original collected data may have abnormal values, preferably, the pre-processing is performed before being used for contact resistance calculation.

[0048] Specifically, first, it is judged whether the data is within a reasonable range, including,

[0049] Voltage validity: if U(t) exceeds the rated voltage of the facility by ±20% (e.g., for a facility rated at 1000V, U(t)>1200V or <800V), it is determined to be an abnormal value, and the data is excluded;

[0050] Current validity: if I(t) exceeds the rated current by ±30% (e.g., for a facility rated at 500A, I(t)>650A or <0A, excluding normal impact during charging start and stop), it is determined to be an abnormal value, and the data is excluded;

[0051] Timestamp verification: if the timestamp interval between adjacent two sets of data exceeds ±50% of the sampling period (e.g., for a 100ms sampling period, interval>150ms or <50ms), it is determined to be a data packet loss, and the missing data is supplemented by linear interpolation (to avoid interruption in subsequent ΔU(t) / ΔI(t) calculation).

[0052] Next, a moving average filter (window size of 3-5 sampling points) is used to filter U(t) and I(t) to remove high-frequency noise (e.g., 10kHz-100kHz noise caused by charging module switching frequency), for example:

[0053] For continuously collected U1(t), U2(t), U3(t) (100ms interval), the filtered voltage U_filter(t)=(U1(t)+U2(t)+U3(t)) / 3;

[0054] Current filtering is the same, ensuring that the filtered U_filter(t) and I_filter(t) can reflect the true contact point voltage and current change trend, avoiding false fluctuations in ΔR(t) calculation caused by noise.

[0055] In addition, due to the possible small differences in the response speed of the voltage and current modules, it is preferred to align the two sets of data by timestamp: if the timestamp of U(t) is t1 and the timestamp of I(t) is t2, when |t1-t2|≤1ms, it is determined to be synchronous data and can be used for subsequent ΔU(t), ΔI(t) calculation; if the deviation is >1ms, the voltage timestamp is used as the reference, and the current data is linearly interpolated to obtain I(t) at t1, ensuring that the time interval Δt in ΔU(t)=U(t)-U(t-Δt), ΔI(t)=I(t)-I(t-Δt) is uniform (e.g., Δt=100ms, i.e., the difference between adjacent sampling points).

[0056] In some embodiments, for S102, differential calculation is the basis for obtaining contact resistance. First, the calculation rules for voltage change ΔU(t) and current change ΔI(t) are defined to ensure that the two are synchronized in the time dimension and the changes are accurately captured. For example, the specific rules are as follows:

[0057] The time window and differential interval settings allow for data sampling frequencies ranging from tens to hundreds of milliseconds, and the instantaneous fluctuation characteristics of contact resistance exhibiting a "sudden increase—slow decrease" pattern are matched with the differential interval and sampling frequency.

[0058] If the sampling frequency is 50ms-500ms (i.e., 2-20 sets of voltage and current data are collected per second), the differential interval is uniformly set to a single sampling period (denoted as Δt), that is, ΔU(t) and ΔI(t) are both the difference between the current sampled value and the previous sampled value, expressed by the formula:

[0059] ΔU(t) = U(t) - U(t - Δt)

[0060] ΔI(t) = I(t) - I(t - Δt)

[0061] Choosing a single sampling period as the differential interval can preserve the instantaneous changes in contact resistance to the greatest extent (such as a sudden increase in resistance at the contact point due to temperature rise), and avoid smoothing out early deterioration signals due to an excessively large interval (such as multi-period differential) or introducing noise interference due to an excessively small interval (such as subsampling period).

[0062] Differential data synchronization verification is necessary because voltage U(t) and current I(t) need to be acquired through different acquisition modules, which may result in slight time deviations (such as sampling asynchrony caused by differences in module response speed).

[0063] Specifically, a millisecond-level timestamp (e.g., the acquisition time is accurate to 1ms) is added to each group of U(t) and I(t). If the current U(t) timestamp is T and the previous U(t-Δt) timestamp is T-Δt, the corresponding I(t) and I(t-Δt) timestamps must match T and T-Δt respectively, with a deviation range of ≤1ms.

[0064] If the time deviation is greater than 1ms, the current data is corrected by linear interpolation. For example, if the actual timestamp of I(t) is T+0.5ms, the theoretical current value I'(T) at time T needs to be calculated based on the changing trends of I(T) and I(T+Δt). Then, I'(T) is used to participate in the calculation of ΔI(t) to ensure that ΔU(t) and ΔI(t) correspond to the same time interval of change.

[0065] In this embodiment, as Figure 2As shown, the process for calculating contact resistance, based on verified ΔU(t) and ΔI(t), is exemplarily described by the following steps:

[0066] S201, basic calculation, directly substitute into Ohm's law local deviation formula, to get Rcontact(t)=ΔU(t) / ΔI(t);

[0067] S202, outlier filtering, sets a reasonable range for contact resistance (based on the material and design parameters of the charging facility contact points, such as the normal resistance of copper contacts is usually in the mΩ range, and the set range is 0.1mΩ-10mΩ). If the calculation result exceeds this range, it is judged as an anomaly (possibly because ΔI(t) is close to 0, resulting in an excessively small denominator, or ΔU(t) is abnormally disturbed). In this case, Rcontact(t) is not used, but the average value of the previous 3 valid calculation results is used instead to avoid the outlier affecting the subsequent deviation analysis.

[0068] S203, on the smoothed time series data, find the first position that meets the requirement that the improvement of the indicator value relative to the baseline value is ≥R% and the trend of change continues for at least K subsequent follow-up points as the candidate first significant improvement point;

[0069] S204, unify units to ensure that ΔU(t) is in volts (V), ΔI(t) is in amperes (A), and the calculated result Rcontact(t) is in ohms (Ω). In subsequent analysis, it can be converted to milliohms (mΩ) to simplify the values ​​(e.g., 1.2Ω = 1200mΩ).

[0070] In this embodiment, the nominal value Rref is the benchmark for judging whether the contact resistance has deteriorated. This embodiment is set based on the contact resistance characteristics under normal operating conditions to avoid misjudgment of ΔR(t) due to benchmark deviation.

[0071] Specifically, for the initial calibration of Rref, after the new charging facility or contact point is replaced, the initial calibration is performed under normal load conditions without degradation to obtain a reference Rref. For example,

[0072] The calibration conditions include an ambient temperature of 25℃±5℃ (to eliminate the influence of extreme temperatures on resistance), a charging power of 50%-80% of the rated power (to avoid abnormal resistance caused by no-load or full-load), and continuous stable operation for 30 minutes (to ensure that the contact point temperature reaches a stable state).

[0073] Under calibration conditions, U(t) and I(t) are collected at the normal sampling frequency, and Rcontact(t) is calculated for each time. A total of 300-500 sets of valid data are obtained (corresponding to 5-8 minutes of collection when the sampling frequency is 100ms).

[0074] Remove the maximum and minimum values ​​from the valid Rcontact(t) data (to exclude random fluctuations), and take the average value of the remaining data as the initial nominal value Rref. For example, after removing 10 maximum and 10 minimum values ​​from 500 sets of data, the average value is 0.5mΩ, then Rref=0.5mΩ.

[0075] Preferably, as the charging facility is used for an extended period, slight normal wear (non-deterioration wear) may occur at the contact points. Rref should be periodically adjusted to prevent an outdated reference from causing an artificially high ΔR(t).

[0076] Optionally, the correction cycle is once every 3 months or once every 1000 charges (whichever comes first);

[0077] The correction method includes repeating the initial calibration conditions and data acquisition process, calculating a new average Rcontact(t), and if the deviation between the new average value and the current Rref is ≤10%, then the original Rref is maintained; if the deviation is >10%, then Rref is updated with the new average value (e.g., if the original Rref = 0.5mΩ, the new average value = 0.54mΩ, the deviation is 8%, so the original Rref is maintained; if the new average value = 0.56mΩ, the deviation is 12%, so Rref is updated to 0.56mΩ).

[0078] Once Rcontact(t) and Rref are determined, ΔR(t) can be calculated, which can be simplified to the following formula:

[0079] ΔR(t) = Rcontact(t) Rref

[0080] For example, if at a certain moment Rcontact(t) is calculated to be 0.7mΩ, and the current Rref is 0.5mΩ, then ΔR(t) = 0.7 - 0.5 = 0.2mΩ;

[0081] If Rcontact(t) < Rref (e.g., the contact resistance decreases due to improved cleanliness), then ΔR(t) is negative. This is marked as normal fluctuation and is not included in the degradation analysis (the core characteristic of degradation is an increase in contact resistance, i.e., ΔR(t) is positive and continues to increase).

[0082] Preferably, the original ΔR(t) may exhibit spurious fluctuations due to acquisition noise and instantaneous load fluctuations. In the preferred embodiment, verification and smoothing operations are performed.

[0083] Specifically, for validity verification, a normal fluctuation range for ΔR(t) is set (based on historical data statistics, such as ±0.1mΩ). If a single ΔR(t) exceeds this range (such as a sudden increase from 0.2mΩ to 1.0mΩ), it needs to be judged in conjunction with the concurrent ΔI(t) and temperature data: if ΔI(t) does not change significantly (≤5%) and the temperature is stable (≤±2℃), it is judged as a false fluctuation, and the sliding average of the previous 5 ΔR(t) is used instead; if ΔI(t) or temperature changes significantly, it is judged as a real change (such as a sudden increase in current causing instantaneous heating at the contact point), and the ΔR(t) is retained.

[0084] For smoothing, for example, a 3-point moving average is used to smooth the ΔR(t) sequence, as shown in the formula:

[0085] ΔR_smooth(t)=[ΔR(t-2Δt)+ΔR(t-Δt)+ΔR(t)] / 3

[0086] Smoothing can remove high-frequency noise while preserving the overall trend of ΔR(t) (such as the degradation characteristic of "sudden increase - slow decrease").

[0087] In some embodiments, for S103, the core parameters of the sliding window (window size, sliding step size, data type) can be set in combination with the sampling frequency and degradation characteristics of ΔR(t) to capture instantaneous fluctuations while avoiding data redundancy or feature distortion.

[0088] Specifically, the sliding window setting is based on the sampling frequency, determining the number of sampling points covered by the window. If the sampling frequency of ΔR(t) is 100ms / time (i.e., 10 data points per second), the window size is set to 10-30 sampling points (corresponding to a time dimension of 1-3 seconds). For example, selecting 20 sampling points (a 2-second window) can capture a complete "sudden increase - initial decrease" fluctuation (if degradation occurs instantaneously, the sudden increase peak and initial decrease trend can be captured within 2 seconds), without causing feature lag due to an excessively large window (e.g., 60 sampling points, 6 seconds) or severe noise interference due to an excessively small window (e.g., 5 sampling points, 0.5 seconds).

[0089] If ΔR(t) fluctuates frequently during charging (e.g., unstable current causes high-frequency changes in ΔR(t)), the window size is automatically reduced to 10 sampling points (1 second) to improve the feature's response speed to fluctuations; if ΔR(t) is stable (e.g., constant charging power causes ΔR(t) to change slowly), the window size is expanded to 30 sampling points (3 seconds) to enhance feature stability.

[0090] Regarding the sliding step size setting, the sliding step size determines the feature update frequency and matches the time granularity of subsequent degradation rate prediction. For example, using one sampling point as the sliding step size means that the window slides forward one step every time it collects a new ΔR(t) data point. The corresponding feature update frequency is consistent with the ΔR(t) sampling frequency (e.g., 100ms / sample, 100ms / feature update). This step size ensures that the features can track the changes in ΔR(t) in real time, avoiding missing abrupt degradation rate changes due to an excessively large step size (e.g., 5 sampling points, 500ms / update).

[0091] If the charging facility is in a low-power mode (such as trickle charging, where ΔR(t) changes very slowly), the sliding step size can be adjusted to 5 sampling points (500ms / update) to reduce computational resource consumption without affecting feature accuracy.

[0092] After determining the window parameters, the ΔR(t) data within each sliding window is preprocessed. For example, this may include window data alignment and timestamp marking, i.e., each sliding window corresponds to a window timestamp, and the timestamp of the last ΔR(t) within the window is used as the window marker (e.g., if the window contains 20 sampling points t1-t20, with timestamps from 100ms to 2000ms respectively, the window timestamp is recorded as 2000ms), to ensure that subsequent features correspond accurately to the time dimension, which facilitates the GRU-D model to call historical feature sequences.

[0093] Preferably, ΔR(t) within the window may contain high-frequency noise (such as small fluctuations caused by electromagnetic interference), which can be further denoised by moving average filtering, which will not be elaborated here.

[0094] Based on the preprocessed ΔR(t) data within the window, nonlinear characteristic parameters are extracted, including the calculation of mean deviation, fluctuation variance, and rise rate.

[0095] Specifically, for the average deviation It reflects the overall degradation level of ΔR(t) within the window. The average deviation is the arithmetic mean of all denoised ΔR(t) within the window, used to measure the overall magnitude of the contact resistance deviation within this time period, directly reflecting the cumulative trend of degradation. Its calculation can be simplified to the formula:

[0096]

[0097] Where N is the number of sampling points within the window (e.g., 20). Let ΔR(t) be the denoised data of the i-th data point within the window.

[0098] The variance of fluctuation (Var(ΔR)) reflects the stability of ΔR(t) within a window and identifies instantaneous degradation. Variance of fluctuation is the degree to which ΔR(t) deviates from the average deviation within the window; it measures the severity of fluctuations in ΔR(t) and can capture nonlinear fluctuations such as "sudden increase – slow decrease". Its calculation can be simplified to the following formula:

[0099]

[0100] That is, the sample variance is divided by N-1 to avoid underestimating the overall fluctuation. If Var(ΔR) suddenly increases (e.g., from 0.002mΩ² to 0.01mΩ²), it indicates that ΔR(t) fluctuates drastically, which may correspond to the instantaneous deterioration of the contact point due to temperature rise (e.g., a sudden increase); if Var(ΔR) remains at a low level for a long time (e.g., <0.003mΩ²), it indicates that the contact point is stable and there is no obvious fluctuation.

[0101] The rate of increase (d(ΔR) / dt) reflects the changing trend of ΔR(t) and predicts the risk of accelerated degradation. The rate of increase is the rate of change of ΔR(t) over time within a window, used to capture the upward or downward trend of ΔR(t). Its calculation process specifically includes...

[0102] Establish a linear regression model of ΔR(t) versus time within a window. Let each... The corresponding timestamp is t_i (e.g., t_1=100ms, t_2=200ms, ..., t_20=2000ms). With time t as the independent variable and ΔR as the dependent variable, a linear equation is fitted: ΔR=kt+b, where k is the slope of the regression line.

[0103] The least squares method is used to solve for k, and the formula is:

[0104]

[0105] Where k is in mΩ / ms or mΩ / s, usually converted to mΩ / s. If k > 0 and its absolute value increases (e.g., from 50 mΩ / s to 150 mΩ / s), it indicates that ΔR(t) is accelerating, and the contact point deterioration rate is accelerating; if k ≈ 0, it indicates that ΔR(t) is stabilizing; if k < 0, it indicates that ΔR(t) is decreasing, possibly corresponding to the resistance recovering after the contact point temperature decreases (not a deterioration trend).

[0106] The average deviation, variance, and rate of rise extracted from each sliding window are bound to the window timestamp to form a feature data set of timestamp-average deviation-variance-rate of rise.

[0107] In some embodiments, for S104, preferably, the time series prediction model is GRU-D. During the charging process, communication interference or temporary hardware failures may cause the loss of nonlinear characteristic parameters (such as the rate of rise d(ΔR) / dt), resulting in a sparse time series. GRU-D introduces a missing mask in the gating unit to quantify the degree of data loss and dynamically adjust the model weights, avoiding prediction bias caused by sparse data. Furthermore, compared to complex models (such as LSTM and Transformer), GRU-D has a simpler structure, containing only three core units: a reset gate, an update gate, and candidate hidden states. It can complete a single prediction in milliseconds, meeting real-time control requirements.

[0108] For example, the GRU-D model structure is designed with three layers: input layer, hidden layer, and output layer. Specifically, for the input layer, for example, the input dimension = 3 (non-linear features) + 2 (supplementary features) = 5; the time step = 10 (i.e., the feature sequence of the first 10 sliding windows is input, with each window spaced 100ms apart, covering 1 second of historical data); the historical feature sequence is converted into a vector form that the model can recognize, ensuring that it contains enough historical information to capture degradation trends;

[0109] For the hidden layer, it is a single GRU-D layer with 32 hidden units; the activation function is ReLU; the dropout probability is 0.2 (to prevent overfitting); the missing mask dimension is 5 (to match the input dimension); the time dependency of the input sequence is processed, the influence of sparse data is corrected by the missing mask, and the hidden state vector is output.

[0110] For the output layer, it is a fully connected layer with an output dimension of 1; the activation function is Sigmoid (which maps the output to the 0-1 interval, corresponding to the normalized value of λ(t)), which converts the hidden layer output into a single degradation rate value λ(t), ensuring that the output range is controllable and facilitating subsequent threshold judgment.

[0111] The supplementary features may include: real-time temperature T(t), as rising contact point temperature accelerates resistance degradation, and the temperature near the contact point is collected as an important influencing factor of λ(t); real-time charging power P(t), as higher power results in greater current, more severe contact point heating, and a higher degradation rate, and real-time power data (such as a percentage of rated power) is obtained from the charging facility control system.

[0112] The dimensions of different features vary greatly, so standardization is needed to eliminate the influence of dimensions and avoid the model being biased towards high numerical features. Z-score standardization can be used as a specific method, which will not be elaborated here.

[0113] The feature sequence is constructed according to the model time step (10 windows) to form the model input vector. Each input sample is a two-dimensional matrix of time step × feature dimension. For example, if the time step = 10 and the feature dimension = 5, If Var(ΔR), d(ΔR) / dt, T(t), P(t)), then the size of the input matrix is ​​10×5;

[0114] The sequence sliding rule allows the input sequence to slide forward one step each time a new sliding window feature is generated (at 100ms intervals), removing the feature from the oldest window and adding the feature from the newest window. This ensures that the input sequence always contains the latest historical information. For example, when t=1000ms, the input sequence can be 10 window features from t=100ms to t=1000ms, and when t=1100ms, the input sequence can be 10 window features from t=200ms to t=1100ms.

[0115] Optionally, the model needs to be trained using historical degraded data to learn the mapping relationship between features and λ(t), and at the same time, the generalization ability needs to be improved through optimization strategies. For example, the specific steps are as follows:

[0116] 1. Training dataset construction

[0117] Collect historical operating data for different charging facilities (covering DC fast charging and AC slow charging) and different contact point states (new contact, slightly deteriorated, moderately deteriorated, and severely deteriorated), and collect a sufficient number of samples (e.g., ≥100,000 sets) to ensure coverage of multiple deterioration scenarios;

[0118] The true value (label) of λ(t) is determined by the actual degree of degradation of the contact point. Specifically, for the contact points in historical data, the actual degradation level is obtained through offline detection (such as measuring contact resistance using the four-terminal method). Combined with the characteristic sequence at the corresponding time, λ(t) is divided into normalized values ​​of 0-1 (0 = no degradation, 1 = extremely severe degradation). For example, mild degradation corresponds to λ(t) = 0.2-0.4, and severe degradation corresponds to λ(t) = 0.8-1.0.

[0119] The dataset is divided into a training set (70%, used for model parameter learning), a validation set (20%, used for hyperparameter tuning), and a test set (10%, used for model accuracy evaluation) in a ratio of 7:2:1.

[0120] The mean squared error (MSE) is used as the loss function, and the Adam optimizer is employed. The specific training process will not be described in detail here.

[0121] After the trained GRU-D model is deployed, λ(t) is predicted in real time, for example, following the procedure below.

[0122] Synchronized with sliding window feature extraction, model prediction is automatically triggered every time a new sliding window feature is generated (at 100ms interval). That is, after feature extraction is completed, the constructed input sequence (10×5 matrix) is input into the GRU-D model, and the model outputs the predicted value of λ(t).

[0123] Optionally, the model output is a normalized λ(t) value of 0-1, which can be converted into a degradation rate in a practical physical sense, as shown in the following formula: ,in The preset maximum degradation rate (based on historical data statistics, such as 0.1mΩ / s, which means that when the contact resistance increases by 0.1mΩ per second, it is considered extremely severe degradation). Output values ​​for the model. The degradation rate after transformation, such as the model output. =0.5, then =0.5×0.1=0.05mΩ / s.

[0124] Preferably, the predicted value of λ(t) is updated based on a preset time period, wherein the update frequency is adaptively adjusted based on the stability of the charging process, and the greater the fluctuation of the charging current, the higher the update frequency.

[0125] Specifically, to ensure the coordination between λ(t) prediction and subsequent virtual lifetime consumption L(t) accumulation and power control, a preset base period for λ(t) update is first set, which matches the sliding window interval of the nonlinear feature parameter extraction mentioned above (e.g., 100ms / feature extraction).

[0126] For example, the default preset time period is consistent with the sliding window feature extraction period, i.e., 100ms / time. This setting can ensure that the λ(t) prediction can be connected in real time with the latest feature parameters such as average deviation, fluctuation variance and rise rate, avoiding prediction lag due to period misalignment. For example, every time a new set of nonlinear features is generated, a λ(t) prediction update is immediately triggered, so that λ(t) is always calculated based on the latest deteriorated features.

[0127] The stability of the charging process is judged primarily by the amplitude of charging current fluctuations (current fluctuations directly affect the contact resistance deviation ΔR(t), which in turn is related to the degradation rate). Specific judgment methods include...

[0128] Synchronously collect charging current data within each λ(t) update cycle (from the same source as real-time voltage and current acquisition), and calculate the maximum value I_max and minimum value I_min of the current within that cycle;

[0129] The degree of fluctuation is calculated using the formula: Current fluctuation amplitude = |I_max - I_min| / I_avg × 100% (where I_avg is the average current value within the cycle). For example, if I_max = 105A, I_min = 95A, and I_avg = 100A within a certain cycle, then the fluctuation amplitude = |105 - 95| / 100 × 100% = 10%.

[0130] Three levels of stability standards are set, such as low fluctuation (fluctuation amplitude ≤ 5%, stable charging), medium fluctuation (5% < fluctuation amplitude ≤ 15%, relatively stable charging), and high fluctuation (fluctuation amplitude > 15%, unstable charging).

[0131] Based on the above charging stability determination results, the update frequency of λ(t) is dynamically adjusted. The greater the fluctuation of the charging current, the higher the update frequency.

[0132] Specifically, for low-fluctuation scenarios (stable charging), when the current fluctuation amplitude is ≤5%, the charging process is considered stable, and the changes in ΔR(t) and nonlinear characteristic parameters are gradual, eliminating the need for high-frequency updates to λ(t). In this case, the update frequency is extended from the basic period of 100ms to 200-300ms / time (e.g., extended to 200ms) to reduce computational power consumption while ensuring prediction accuracy. For example, one λ(t) prediction update is triggered every two sets of nonlinear features generated.

[0133] For medium fluctuation scenarios (relatively stable charging), when 5% < fluctuation amplitude ≤ 15%, there are slight fluctuations in the charging process, and ΔR(t) may show small instantaneous changes. Maintain a basic update frequency of 100ms / time to ensure that small fluctuations in the degradation rate can be captured in time and avoid misjudgment of trends due to reduced frequency.

[0134] In high-fluctuation scenarios (unstable charging), when the fluctuation amplitude is greater than 15%, the charging process becomes unstable (e.g., the vehicle's BMS dynamically adjusts charging demand, or grid voltage fluctuations cause sudden current changes). ΔR(t) is prone to nonlinear fluctuations such as "sudden increase-sudden decrease," and the degradation rate λ(t) may change rapidly. In this case, the update frequency is shortened from the basic period of 100ms to 50-80ms / time (e.g., shortened to 50ms), increasing the λ(t) update frequency. For example, for each set of nonlinear features generated, two λ(t) prediction updates are triggered (with a 50ms interval) to ensure real-time tracking of the dynamic changes in the degradation rate.

[0135] In some embodiments, for S105, the virtual lifetime consumption L(t) is cumulatively calculated based on the degradation rate λ(t). The core of this method is to transform the dynamically changing degradation rate into a measurable lifetime consumption indicator. L(t) is essentially the integral of the degradation rate λ(t) over time, representing the cumulative consumption of the contact point's lifetime under the continuous effect of the degradation rate over a certain period. Since λ(t) changes in real time (e.g., λ is lower at the beginning of charging and increases after the contact point heats up), this embodiment calculates L(t) using a piecewise accumulation method, rather than a fixed-rate integral. The core formula is:

[0136]

[0137] Where L(t) is the virtual lifetime consumption at time t (dimensionless, normalized to the range of 0-100, where 0 = no consumption and 100 = lifetime exhausted).

[0138] , where is the degradation rate of the i-th time segment (converted to a normalized rate value, such as...). =0.1mΩ / s corresponds to normalization =1);

[0139] , where is the duration of the i-th time segment (in seconds, usually consistent with the prediction interval of λ(t), such as 100ms / prediction). );

[0140] n represents the total number of time segments from the start of charging to time t (e.g., if charging lasts 100 seconds and there is one segment every 0.1 seconds, then n = 1000).

[0141] The value of L(t) represents the proportion of the virtual lifetime that the contact point has consumed. For example, L(t)=30 means that 30% of the virtual lifetime of the contact point has been consumed, and L(t)=100 means that the virtual lifetime has been completely exhausted and a forced shutdown for maintenance is required. The upper limit of L(t) is set to 100 (corresponding to the exhaustion of lifetime) and the lower limit is 0.

[0142] Since the actual value of λ(t) (e.g., 0.02 mΩ / s) is not directly related to the physical meaning of the lifetime consumption of L(t), λ(t) needs to be converted into a lifetime consumption rate per unit time, as shown in the formula:

[0143]

[0144] in, Let be the actual degradation rate of the i-th segment (mΩ / s). The preset maximum degradation rate (e.g., 0.1 mΩ / s, based on historical data statistics of the contact point's ultimate degradation rate) is defined, where k is the lifetime consumption coefficient, with a value of 0.1. hour, =0.1, meaning that 0.01 lifespan value is consumed every 0.1s, and it can accumulate to 100 in 100s.

[0145] Based on the above formula, this embodiment combines the actual scenarios of the charging process (such as charging start-up, pause, and rate fluctuation) to formulate cumulative calculation rules to avoid L(t) calculation deviations due to scenario changes.

[0146] Specifically, in the charging start-up scenario, the initial value of L(t) is set to 0 each time charging starts (that is, the lifetime consumption of this charging starts from 0 and is unrelated to the L(t) of the previous charging. The historical accumulated consumption is recorded separately through the facility's total lifetime consumption file and is used to dynamically adjust the threshold based on the service life).

[0147] In charging pause scenarios, when a user pauses charging (e.g., unplugs the charging gun, or actively pauses the charging process), the accumulation of L(t) immediately stops, and the current L(t) value is saved (e.g., L(t) = 35.2). When charging resumes, the accumulation continues from the saved L(t) value and is not reset to zero.

[0148] In case of abnormal interruption scenarios (such as power grid failure or equipment failure), if the interruption time is ≤5 minutes, the L(t) value before the interruption will continue to accumulate after recovery; if the interruption time is >5 minutes, the current charging is determined to be terminated, the L(t) value before the interruption is archived to the history record, and the next charging will start from 0 again.

[0149] In the charging completion scenario, when charging is complete (e.g., the BMS sends a termination command), the accumulation of L(t) stops, and the final value of L(t) (e.g., L(t) = 42.8) is archived to the contact point historical life consumption database for subsequent facility maintenance decisions (e.g., after accumulating 3 charging cycles with L(t) ≥ 80, prompt for contact point replacement).

[0150] Preferably, the contact resistance increases with increasing temperature, and the degradation rate λ(t) increases exponentially with increasing temperature. In this preferred embodiment, a temperature correction factor is introduced. Adjust the cumulative rate of L(t).

[0151] Specifically, the ambient temperature T (unit: °C) near the contact point is collected by the temperature sensor of the charging facility, and the sampling interval is consistent with λ(t) (100ms).

[0152] Based on the Arrhenius equation (which describes the effect of temperature on chemical reaction rate and is adapted to the mechanism of contact point oxidation and degradation).

[0153]

[0154] in, A is the reference temperature (e.g., 25°C, the normal operating temperature of the contact point); A is the temperature coefficient (determined based on the material of the contact point, e.g., A=0.02 for copper contacts, A=0.03 for aluminum contacts).

[0155] Based on the temperature correction factor, single-segment consumption Revised to For example, T=55℃ (30℃ above the reference temperature), copper contacts ≈1.822, original =0.005, after correction =0.005×1.822≈0.0091, which means that the accumulation rate of L(t) increases by 82.2% under extreme temperatures.

[0156] In some embodiments, for S106, during the charging process, based on the real-time L(t) value and the set threshold, combined with the degradation rate λ(t) (auxiliary judgment), the current control stage is determined in real time to ensure accurate and delay-free stage switching.

[0157] In this embodiment, the control phase is divided into a normal phase, a transition phase, and a critical phase. The core difference between each phase lies in the numerical range of L(t). The L(t) threshold (Llow, Lhigh) is set in combination with the contact point life characteristics and charging scenario requirements to ensure that the threshold not only conforms to the degradation mechanism but also adapts to the actual control requirements.

[0158] Specifically, refer to the design life of the contact points (e.g., copper contacts can withstand 1000 charging cycles under normal use, and the maximum allowable virtual life consumption per charge is 80, i.e., Lhigh≤80), to avoid the lifespan being exhausted prematurely due to excessively high L(t) in a single charge.

[0159] Based on the balance between user experience and safety, Low is set in the early stage of degradation and in the range where there is no safety risk (e.g., Low=30) to ensure that high-power charging can be maintained during normal stages; Lhigh is set in the range where degradation accumulates to the point where mandatory intervention is required (e.g., Lhigh=70) to avoid taking measures only when L(t) approaches 100 (lifespan is exhausted), thus reserving a safety buffer.

[0160] Based on scenario adaptability considerations, the thresholds for different charging types (DC fast charging and AC slow charging) can be set differently. For example, DC fast charging has a large current and degrades quickly, so Low can be set to 25 and Lhigh to 65; AC slow charging has a small current and degrades slowly, so Low can be set to 35 and Lhigh to 75.

[0161] Preferably, the low consumption criterion in the combination relationship is dynamically adjusted based on the service life of the charging facility. The longer the service life, the lower the L(t) threshold corresponding to low consumption.

[0162] Specifically, to adapt to dynamic factors such as contact point aging and ambient temperature, the threshold can be adjusted periodically to avoid control lag or over-control caused by a fixed threshold. The specific adjustment logic includes...

[0163] Adjustments are made based on the service life of the contact points. For example, if the service life is less than 1 year, the initial threshold is maintained (e.g., Llow=30, Lhigh=70); if the service life is less than 3 years, Llow is reduced by 10% (27) and Lhigh is reduced by 5% (66.5); if the service life is greater than or equal to 3 years, Llow is reduced by 20% (24) and Lhigh is reduced by 10% (63) to avoid accelerated deterioration of aging contacts due to excessively high thresholds.

[0164] Based on ambient temperature adjustment, when the temperature near the contact point is >45℃ (extreme high temperature), Low is reduced by 5% (28.5) and Lhigh is reduced by 8% (64.4), shortening the normal phase duration and entering the transition phase earlier, reducing the accumulation of degradation under high temperature; when the temperature is <5℃ (extreme low temperature), Low is increased by 5% (31.5) and Lhigh is increased by 5% (73.5), taking advantage of the characteristics of low temperature environment to delay degradation and extend the high power charging time.

[0165] In this embodiment, the preliminary determination stage, based on the numerical relationship between L(t) and Llow and Lhigh, exemplarily includes,

[0166] The normal phase is determined as follows: when L(t) < Llow and λ(t) is at a low rate (e.g., λ(t) < 0.02mΩ / s), it is determined to be a normal phase; if L(t) < Llow but λ(t) is at a high rate (λ(t) ≥ 0.05mΩ / s), an early warning is triggered, but the phase is not switched for the time being, only the monitoring frequency of L(t) is increased (from 100ms / time to 50ms / time).

[0167] Transition phase determination: When Low ≤ L(t) < Lhigh, regardless of whether λ(t) is low or medium rate, it is determined to be the transition phase; if L(t) is close to Lhigh (e.g., L(t) ≥ Lhigh × 90%) and λ(t) is high rate, it is determined to be the late stage of the transition phase, and the power adjustment speed is accelerated.

[0168] Critical phase determination: When L(t)≥Lhigh, regardless of the λ(t) rate, it is directly determined to be a critical phase; if L(t)≥90 (close to the end of life), it is determined to be a critical phase emergency state, and a forced shutdown warning needs to be triggered.

[0169] Preferably, to avoid misjudgment of stages due to abnormal calculation of L(t) (such as sudden increase or stagnation), this embodiment combines λ(t) for auxiliary verification.

[0170] Specifically, the normal phase auxiliary verification, for example, if it is initially determined to be a normal phase, but λ(t) is high for 5 consecutive time segments (500ms), it is necessary to re-check whether the L(t) calculation has been omitted (such as whether the temperature factor has not been applied). If the L(t) calculation is correct, it is still maintained as a normal phase, but the output deterioration rate is abnormally indicated.

[0171] Transition phase verification can be used as an example. If it is initially determined to be a transition phase, but λ(t) remains at a low rate (for 10 consecutive time segments), the duration of the transition phase can be extended (e.g., the original plan to reduce power by 5% per minute can be adjusted to reduce power by 5% every 2 minutes) to balance safety and user experience.

[0172] For critical phase verification, for example, if it is initially determined to be a critical phase, but λ(t) drops sharply to a low rate and the temperature returns to normal, the forced power limit can be temporarily suspended, and the current power can be maintained for observation for 1 minute. If L(t) does not increase significantly, it can be reverted to the transition phase to avoid over-control.

[0173] In this embodiment, differentiated power control strategies are implemented based on the different stages determined above. The core objectives are to ensure efficiency during the normal stage, balance safety and user experience during the transition stage, and prioritize safety during the critical stage.

[0174] When L(t) is low power consumption and λ(t) is low rate, normal charging is performed; when L(t) is high power consumption and λ(t) is low rate, conservative power reduction is performed; when L(t) is low power consumption and λ(t) is high rate, fast power limiting is performed; when L(t) is high power consumption and λ(t) is high rate, forced shutdown or alarm is performed. Low power consumption and high power consumption are determined based on preset thresholds.

[0175] Specifically, during normal operation, maintain high-power charging. When there is no significant risk of contact point degradation, maintain the set high-power charging (usually 80%-100% of the facility's rated power) to maximize charging efficiency.

[0176] For the graded power reduction control during the transition phase, the current power at the time of entering the transition phase is used as the benchmark (e.g., the power at the time of entry is 100kW); the default is to reduce the rated power by 5% per minute (e.g., if the rated power is 120kW, it will be reduced by 6kW per minute), and the magnitude can be dynamically adjusted according to λ(t) (5% / minute when λ(t) is at medium speed, and increased to 8% / minute when it is at high speed); the power during the transition phase shall not be lower than 40% of the rated power (e.g., 48kW) to avoid the user experience being affected by excessively low power.

[0177] During critical phases, power is forcibly limited or charging is prompted to pause, forcibly limiting the power to within 30% of the rated power (e.g., rated 120kW, power limit ≤36kW). This standard is not adjustable, ensuring that the load at the contact point is reduced to a safe level. After entering the critical phase, the power is reduced from the current value to the limit value (e.g., from 48kW to 36kW) within a predetermined time (e.g., within 1 second) to avoid delays that cause L(t) to continue to grow rapidly.

[0178] Therefore, charging power control stages can be divided based on L(t), and differentiated control strategies can be implemented. During the normal stage, charging efficiency is guaranteed; during the transition stage, safety and user experience are balanced; and during the critical stage, safety is prioritized. At the same time, the strategy is optimized through dynamic adjustment and feedback mechanisms, ultimately achieving the core goal of protecting the lifespan of the contact points while improving the user experience, thus providing a proactive and flexible lifespan protection solution for charging facilities.

[0179] Figure 3 A control system 300 for electric vehicle charging facilities is shown. Embodiments of this system are... Figure 1 Corresponding to the illustrated method embodiments, this system can be specifically applied to various electronic devices. Specifically, it includes:

[0180] The data acquisition module 301 is used to acquire real-time voltage and real-time current during the charging process;

[0181] The contact resistance acquisition module 302 is used to calculate the contact resistance based on the difference between the real-time voltage and the real-time current, and compare the contact resistance with the preset nominal value under normal conditions to obtain the contact resistance deviation ΔR(t).

[0182] The nonlinear feature extraction module 303 is used to extract nonlinear feature parameters by performing sliding window processing on the contact resistance deviation ΔR(t). The nonlinear feature parameters include average deviation, fluctuation variance, and rise rate.

[0183] The degradation rate prediction module 304 is used to predict the degradation rate λ(t) of the charging contact point based at least on the nonlinear characteristic parameters and through a time series prediction model.

[0184] The virtual lifetime consumption calculation module 305 is used to cumulatively calculate the virtual lifetime consumption L(t) of the charging contact point based on the degradation rate λ(t).

[0185] The power control module 306 is used to determine the charging power control stage based on the virtual lifetime consumption L(t), and to perform control of maintaining high power, graded power reduction and forced power limiting based on the control stage, wherein the control stage includes a normal stage, a transition stage and a critical stage.

[0186] Based on the same inventive concept, this application also provides an electronic device. The method corresponding to the electronic device can be the method in the foregoing embodiments, and its problem-solving principle is similar to that method. The electronic device provided in this application includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the methods and / or technical solutions of the foregoing embodiments of this application.

[0187] Figure 4 The diagram illustrates the structure of an apparatus suitable for implementing the methods and / or technical solutions in the embodiments of this application. The apparatus 400 includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes based on a program stored in a read-only memory (ROM) 402 or a program loaded from a storage portion 408 into a random access memory (RAM) 403. The RAM 403 also stores various programs and data required for system operation. The CPU 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0188] The following components are connected to I / O interface 405: input section 406 including keyboard, mouse, touch screen, microphone, infrared sensor, etc.; output section 407 including cathode ray tube (CRT), liquid crystal display (LCD), LED display, OLED display, etc., and speakers, etc.; storage section 408 including one or more computer-readable media such as hard disk, optical disk, magnetic disk, semiconductor memory, etc.; and communication section 409 including network interface card such as LAN (local area network) card, modem, etc. Communication section 409 performs communication processing via a network such as the Internet.

[0189] In particular, the methods and / or embodiments in this application can be implemented as computer software programs. For example, the embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. When the computer program is executed by the central processing unit (CPU) 401, it performs the functions defined in the methods of this application.

[0190] Another embodiment of this application provides a computer-readable storage medium having computer program instructions stored thereon, which can be executed by a processor to implement the methods and / or technical solutions of any one or more embodiments of this application described above.

[0191] The flowcharts or block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of devices, methods, and computer program products according to various embodiments of this application.

[0192] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a device claim may also be implemented by a single unit or device through software or hardware. The terms "first," "second," etc., are used to indicate names and do not indicate any specific order.

Claims

1. A control method for electric vehicle charging facilities, characterized in that, Includes the following steps: Collect real-time voltage and current during the charging process; The contact resistance is calculated based on the difference between the real-time voltage and the real-time current. The contact resistance is then compared with the preset nominal value under normal conditions to obtain the contact resistance deviation ΔR(t). The contact resistance deviation ΔR(t) is processed by a sliding window to extract nonlinear characteristic parameters, including average deviation, fluctuation variance and rise rate. Based at least on the aforementioned nonlinear characteristic parameters, the degradation rate λ(t) of the charging contact point is predicted using a time series prediction model. The virtual lifetime consumption L(t) of the charging contact points is cumulatively calculated based on the degradation rate λ(t); Based on the virtual lifetime consumption L(t), the charging power control stage is determined, and based on the control stage, control is performed to maintain high power, reduce power in stages, and limit power in a forced manner. The control stage includes a normal stage, a transition stage, and a critical stage. When the control stage is the normal stage, high power charging is maintained; when the control stage is the transition stage, graded power reduction charging is performed; and when the control stage is the critical stage, forced power limiting charging is performed.

2. The control method for electric vehicle charging facilities according to claim 1, characterized in that, The transition phase adopts a graded power reduction, with each power reduction being a preset percentage of the rated power, and is gradually adjusted in minutes. The power reduction is dynamically adjusted based on the degradation rate λ(t).

3. The control method for electric vehicle charging facilities according to claim 1, characterized in that, Also includes: The degradation rate λ(t) of the charging contact point is predicted by the time series prediction model. The predicted value of degradation rate λ(t) is updated based on a preset time period. The update frequency is adaptively adjusted based on the stability of the charging process. The greater the fluctuation of the charging current, the higher the update frequency.

4. The control method for electric vehicle charging facilities according to claim 1, characterized in that, The step of cumulatively calculating the virtual lifetime consumption L(t) of the charging contact point based on the degradation rate λ(t) further includes introducing a temperature correction factor to adjust the accumulation rate of the virtual lifetime consumption L(t), wherein the temperature correction factor is related to the reference temperature and the temperature coefficient of the contact point material.

5. A control system for electric vehicle charging facilities, characterized in that, include: The data acquisition module is used to collect real-time voltage and real-time current during the charging process; The contact resistance acquisition module is used to calculate the contact resistance based on the difference between the real-time voltage and the real-time current, and compare the contact resistance with the preset nominal value under normal conditions to obtain the contact resistance deviation ΔR(t). The nonlinear feature extraction module is used to extract nonlinear feature parameters by performing sliding window processing on the contact resistance deviation ΔR(t). The nonlinear feature parameters include average deviation, fluctuation variance, and rise rate. The degradation rate prediction module is used to predict the degradation rate λ(t) of the charging contact point based at least on the nonlinear characteristic parameters and through a time series prediction model. The virtual lifetime consumption calculation module is used to cumulatively calculate the virtual lifetime consumption L(t) of the charging contact point based on the degradation rate λ(t); The power control module is used to determine the charging power control stage based on the virtual lifetime consumption L(t), and to perform control of maintaining high power, graded power reduction, and forced power limiting based on the control stage. The control stage includes a normal stage, a transition stage, and a critical stage. When the control stage is the normal stage, high power charging is maintained; when the control stage is the transition stage, graded power reduction charging is performed; and when the control stage is the critical stage, forced power limiting charging is performed.

6. An electronic device, characterized in that, include: At least one processor; and a memory communicatively connected to the processor; wherein, The memory stores instructions that can be executed by the processor to enable the processor to perform the method of any one of claims 1-4.

7. A computer-readable medium having computer program instructions stored thereon, characterized in that, The computer program instructions can be executed by a processor to implement the method as described in any one of claims 1-4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-5.

Images