Fuzzy control power dynamic adjustment method of charging pile node and related device

Abstract

The invention discloses a fuzzy control power dynamic adjustment method of a charging pile node and a related device. The method and the device are used for improving dynamic adaptability and operation stability when the charging pile node is controlled. The method comprises the following steps: carrying out preprocessing calculation on feedback data to obtain a characteristic quantity; performing nonlinear composite deviation calculation on the characteristic quantity of each charging pile node to obtain a comprehensive state deviation; performing fuzzification processing on the comprehensive state deviation and the characteristic quantity to obtain a membership degree value; calculating the trigger intensity of each fuzzy rule according to the membership value, updating the weight of each fuzzy rule based on the trigger intensity to obtain a target weight, carrying out weighted aggregation on the output of each fuzzy rule, and carrying out nonlinear output correction after defuzzification through a centroid method to obtain a target adjustment coefficient; calculating adjustment power corresponding to the charging pile node according to the target adjustment coefficient; and performing unification correction on the adjustment power of each charging pile node according to the total demand power to obtain target power.

Description

Technical Field

[0001] This application relates to the field of charging management, and in particular to a fuzzy control power dynamic adjustment method and related device for charging pile nodes. Background Technology

[0002] Current electric vehicle charging station systems mostly employ modular power regulation methods based on PID or conventional fuzzy control. These control methods typically adjust power output according to set rules by collecting signals such as module output voltage, current, and temperature. However, in actual charging scenarios with multiple modules connected in parallel, system operation is affected by a combination of factors, including grid fluctuations, changes in vehicle BMS demand, and the module's own thermal state, resulting in nonlinear and time-varying characteristics in the control system.

[0003] Traditional control algorithms often exhibit delayed responses to these complex coupling factors, making it difficult to maintain efficient power balance while ensuring safety. Specifically, the sensitivity of parameters such as power, voltage, and temperature remains unchanged under different rules, causing the controller to output the same control trend even during sudden voltage fluctuations or accelerated temperature rises, resulting in a decrease in response speed. Furthermore, fuzzy inputs are mostly linearly weighted structures, failing to reflect the nonlinear characteristics of voltage drops or sudden temperature rises. This causes some modules to be incorrectly amplified under high temperature or low voltage conditions, making it difficult for traditional fuzzy control structures to balance safety and dynamic response, leading to reduced energy distribution efficiency and uneven module lifespan. Summary of the Invention

[0004] To address the aforementioned technical issues, this application provides a fuzzy control power dynamic adjustment method and related apparatus for charging pile nodes, which enhances the dynamic adaptability and operational stability when controlling charging pile nodes.

[0005] The technical solution provided in this application is described below: The first aspect of this application provides a fuzzy control power dynamic adjustment method for charging pile nodes, including: Feedback data from all charging pile nodes is received according to a fixed sampling period; The feedback data is preprocessed and calculated to obtain feature quantities; Nonlinear composite deviation calculation is performed on the characteristic quantities of each charging pile node to obtain the comprehensive state deviation; The comprehensive state deviation and the feature quantity are fuzzified to obtain the membership value; The trigger strength of each fuzzy rule is calculated based on the membership value, and the weight of each fuzzy rule is updated based on the trigger strength to obtain the target weight; The outputs of each fuzzy rule are weighted and aggregated according to the target weight, and then defuzzified using the centroid method to obtain the power adjustment coefficient. The power adjustment coefficient is nonlinearly corrected to obtain the target adjustment coefficient. Calculate the adjustment power corresponding to the charging pile node based on the target adjustment coefficient; The power adjustment of each charging pile node is consistently corrected based on the total power demand to obtain the target power, which is used to control the output power of the corresponding charging pile node.

[0006] Optionally, the step of calculating the trigger strength of each fuzzy rule based on the membership value, and updating the weight of each fuzzy rule based on the trigger strength to obtain the target weight includes: Obtain a preset fuzzy rule set, which contains the initial weight of each fuzzy rule; Within the fixed sampling period, the trigger intensity of each fuzzy rule is calculated based on the membership value; Calculate the average trigger strength of all fuzzy rule trigger strengths; The initial weights of the fuzzy rules are dynamically adjusted based on the difference between the trigger strength of each fuzzy rule and the average trigger strength to obtain the target weights.

[0007] Optionally, the step of dynamically adjusting the initial weights of the fuzzy rules based on the difference between the trigger strength of each fuzzy rule and the average trigger strength to obtain the target weights includes: The initial weights of the fuzzy rules are dynamically adjusted according to the weight update formula to obtain the target weights; The weight update formula is:

[0008] Among them, w r (k) represents the weight of the r-th fuzzy rule in the current period, w r (k-1) represents the weight of the r-th fuzzy rule in the previous cycle, and ρ is the weight adjustment rate. r (k) represents the trigger strength of the current fuzzy rule. This is the average trigger strength of all fuzzy rules.

[0009] Optionally, the preprocessing and calculation of the feedback data to obtain feature quantities includes: The feedback data is cleaned and normalized to obtain the calculation parameters; The power deviation, voltage deviation, power change rate, and temperature change rate of the charging pile node corresponding to the feedback data are calculated by calculation parameters. The calculation results are combined and the characteristic quantities are output. The power deviation is the difference between the required power and the actual output power of the charging pile node. The voltage deviation is the difference between the reference voltage and the current bus voltage of the charging pile node. The power change rate is the change in power deviation between two adjacent sampling periods of the charging pile node divided by the sampling period. The temperature change rate is the temperature difference between two adjacent sampling periods of the charging pile node divided by the sampling period.

[0010] Optionally, the nonlinear composite deviation calculation of the characteristic quantities of each charging pile node to obtain the comprehensive state deviation includes: The formula for calculating the nonlinear composite deviation is:

[0011] Where, ΔE i Let ΔP be the overall state deviation of the i-th charging pile node, ΔP be the power deviation of the current node, and ΔV be the voltage deviation of the current node. The rate of change of power, Let be the rate of temperature change, and α, β, γ, and η be the weighting coefficients for power, voltage, rate of change, and temperature change, respectively.

[0012] Optionally, the nonlinear output correction of the power adjustment coefficient to obtain the target adjustment coefficient includes: The power adjustment coefficient is subjected to amplitude limiting and dead zone processing to obtain effective data; The effective data is calculated using a nonlinear output correction algorithm to obtain the target adjustment coefficient; The calculation formula for the nonlinear output correction algorithm is: λ′=k1·λ+k2·λ 3 Where λ′ is the target adjustment coefficient, λ is the effective data of the power adjustment coefficient, k1 is the linear gain coefficient, and k2 is the nonlinear correction coefficient.

[0013] Optionally, after the method performs a consistent adjustment of the power of each charging pile node based on the total power demand to obtain the target power, it further includes: The stability parameters of each charging pile node are calculated based on the temperature change rate and power change rate of each charging pile node. When the stability parameter is lower than the preset stability threshold, the power redistribution logic is triggered to recalculate the target power adjustment coefficient.

[0014] A second aspect of this application provides a fuzzy control power dynamic adjustment system for a charging pile node, the system comprising: The receiving unit is used to receive feedback data from all charging pile nodes according to a fixed sampling period; The first calculation unit is used to preprocess and calculate the feedback data to obtain feature quantities; The second calculation unit is used to perform nonlinear composite deviation calculation on the characteristic quantities of each charging pile node to obtain the comprehensive state deviation. A fuzzification unit is used to fuzzify the comprehensive state deviation and the feature quantity to obtain a membership value. The update unit is used to calculate the trigger strength of each fuzzy rule based on the membership value, and update the weight of each fuzzy rule based on the trigger strength to obtain the target weight; The weighting unit is used to weight and aggregate the outputs of each fuzzy rule according to the target weight, and then defuzzify them using the centroid method to obtain the power adjustment coefficient. The correction unit is used to perform nonlinear output correction on the power adjustment coefficient to obtain the target adjustment coefficient; The third calculation unit is used to calculate the adjustment power corresponding to the charging pile node based on the target adjustment coefficient. The correction unit is used to make consistent corrections to the adjustment power of each charging pile node according to the total demand power to obtain the target power, which is used to control the output power of the corresponding charging pile node.

[0015] Optionally, the update unit is further configured to: Obtain a preset fuzzy rule set, which contains the initial weight of each fuzzy rule; Within the fixed sampling period, the trigger intensity of each fuzzy rule is calculated based on the membership value; Calculate the average trigger strength of all fuzzy rule trigger strengths; The initial weights of the fuzzy rules are dynamically adjusted based on the difference between the trigger strength of each fuzzy rule and the average trigger strength to obtain the target weights.

[0016] Optionally, the update unit is further configured to: The initial weights of the fuzzy rules are dynamically adjusted according to the weight update formula to obtain the target weights; The weight update formula is:

[0017] Among them, w r (k) represents the weight of the r-th fuzzy rule in the current period, w r (k-1) represents the weight of the r-th fuzzy rule in the previous cycle, and ρ is the weight adjustment rate. r (k) represents the trigger strength of the current fuzzy rule. This is the average trigger strength of all fuzzy rules.

[0018] Optionally, the first computing unit is further configured to: The feedback data is cleaned and normalized to obtain the calculation parameters; The power deviation, voltage deviation, power change rate, and temperature change rate of the charging pile node corresponding to the feedback data are calculated by calculation parameters. The calculation results are combined and the characteristic quantities are output. The power deviation is the difference between the required power and the actual output power of the charging pile node. The voltage deviation is the difference between the reference voltage and the current bus voltage of the charging pile node. The power change rate is the change in power deviation between two adjacent sampling periods of the charging pile node divided by the sampling period. The temperature change rate is the temperature difference between two adjacent sampling periods of the charging pile node divided by the sampling period.

[0019] Optionally, the second computing unit is further configured to: The formula for calculating the nonlinear composite deviation is:

[0020] Where, ΔE i Let ΔP be the overall state deviation of the i-th charging pile node, ΔP be the power deviation of the current node, and ΔV be the voltage deviation of the current node. The rate of change of power, Let be the rate of temperature change, and α, β, γ, and η be the weighting coefficients for power, voltage, rate of change, and temperature change, respectively.

[0021] Optionally, the correction unit is further configured to: The power adjustment coefficient is subjected to amplitude limiting and dead zone processing to obtain effective data; The effective data is calculated using a nonlinear output correction algorithm to obtain the target adjustment coefficient; The calculation formula for the nonlinear output correction algorithm is: λ′=k1·λ+k2·λ 3 Where λ′ is the target adjustment coefficient, λ is the effective data of the power adjustment coefficient, k1 is the linear gain coefficient, and k2 is the nonlinear correction coefficient.

[0022] Optionally, the system further includes: The fourth calculation unit is used to calculate the stability parameters of the corresponding charging pile node based on the temperature change rate and power change rate of each charging pile node. The redistribution unit is used to trigger the power redistribution logic to recalculate the target power adjustment coefficient when the stability parameter is lower than the preset stability threshold.

[0023] A third aspect of this application provides a fuzzy control power dynamic adjustment device for a charging pile node, the device comprising: Processor, memory, input / output units, and bus; The processor is connected to the memory, the input / output unit, and the bus; The memory stores a program, which the processor invokes to execute the first aspect and any one of the optional methods in the first aspect.

[0024] A fourth aspect of this application provides a computer-readable storage medium on which a program is stored, which, when executed on a computer, performs the methods of the first aspect and any one of the first aspects.

[0025] As can be seen from the above technical solutions, this application has the following advantages: This invention receives feedback data from charging pile nodes at a fixed sampling period and preprocesses the feedback data to obtain characteristic quantities. Based on this, nonlinear composite deviation calculations are performed on the characteristic quantities to characterize the node operating status. An adaptive inference model is established through fuzzification processing and a dynamic weight update mechanism to achieve real-time correction of the power adjustment coefficient. After nonlinear output correction and consistency correction, the output target power can simultaneously meet the total power demand and the operating constraints of each node. Therefore, this scheme can achieve rapid response and stable control of power adjustment during the operation of a charging pile group, effectively improving the dynamic adaptability of multi-node collaboration and the overall system's operational stability. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a schematic flowchart of an embodiment of the fuzzy control power dynamic adjustment method for charging pile nodes in this application; Figure 2a This is a schematic flowchart of an embodiment of the fuzzy control power dynamic adjustment method for charging pile nodes in this application. Figure 2b This is a schematic flowchart of an embodiment of the fuzzy control power dynamic adjustment method for charging pile nodes in this application. Figure 3 This is a schematic diagram of an embodiment of the fuzzy control power dynamic adjustment system for charging pile nodes in this application; Figure 4 This is a schematic diagram of another embodiment of the fuzzy control power dynamic adjustment system for charging pile nodes in this application; Figure 5 This is a schematic diagram of an embodiment of the fuzzy control power dynamic adjustment device for charging pile nodes in this application. Detailed Implementation

[0028] It should be noted that the fuzzy control power dynamic adjustment method for charging pile nodes provided in this application can be applied to terminals, systems, and servers. For example, a terminal can be a smartphone, computer, tablet, smart TV, smartwatch, portable computer, or a desktop computer, etc. For ease of explanation, this application uses a terminal as the implementing entity for illustration.

[0029] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0030] Please see Figure 1 This application first provides an embodiment of a fuzzy control power dynamic adjustment method for charging pile nodes, which includes: S101. Receive feedback data from all charging pile nodes according to a fixed sampling period; A charging pile node refers to a power unit within the same charging system that is subject to unified scheduling by the terminal and has independent power conversion and monitoring capabilities. It can be a parallel power module within a single pile or multiple independent pile modules operating in parallel within a station, but they are coordinated and managed by the same control domain.

[0031] The feedback data consists of voltage, current, temperature, and operating status flags reported by each charging pile node within the sampling period, along with timestamps for alignment. The BMS (Battery Management System) provides command information such as power demand.

[0032] In this embodiment, the sampling period is set in milliseconds. Taking the parallel operation of multiple nodes within the station as an example, when several nodes provide charging for different vehicles or are in standby mode, the terminal simultaneously receives the voltage, current, temperature and power demand information sent by the BMS from each node within the same sampling period, thereby forming a state set with a unified time base and avoiding judgment deviations caused by asynchronous sampling.

[0033] When the sampling window is open, the terminal completes timestamp alignment, duplicate frame removal, and packet loss detection; late data is registered in the current cycle's supplementary queue but does not participate in the current cycle's decision-making, to ensure that the inputs involved in the calculation share a consistent time base and to avoid control deviations caused by cross-cycle mixed calculations.

[0034] To improve the reliability of input data, the terminal performs range and rationality checks on the receiving side, including physical constraints such as voltage upper limit, temperature upper limit and bus drop threshold. Abnormal samples are first marked as isolated and handed over to the preprocessing process. At the same time, the terminal records the communication link quality indicators and sets a degradation preparation flag when continuous packet loss reaches the threshold to ensure that subsequent control calculations are performed under stable input conditions.

[0035] S102. Preprocess and calculate the feedback data to obtain the feature quantity; Feature quantities refer to the set of parameters that reflect the node's operating status after data processing, and are used as input variables in the subsequent fuzzy control stage. The terminal's preprocessing process includes timestamp alignment, outlier identification and removal, missing sample imputation, dimensional normalization and range verification. Outlier identification is performed in conjunction with equipment physical constraints and historical window statistical thresholds. For single-point missing values, the most recent reliable value is used for imputation, and for continuous missing values ​​in short windows, the valid value of the previous period is maintained and recorded in the metadata.

[0036] In this embodiment, the terminal first performs data cleaning on the collected feedback data to remove data points with abnormal fluctuations, noise interference, or signal loss. Subsequently, the valid data is normalized to ensure that the numerical dimensions of different nodes remain consistent, guaranteeing the comparability of subsequent calculations. Normalization is used to eliminate differences in measurement ranges between different nodes, ensuring horizontal comparability; the trend of change is expressed using the direction and amplitude levels of the difference between adjacent sampling periods to reflect an upward, flat, or downward trend and provide stronger decision-making information than instantaneous values.

[0037] After preprocessing, the terminal calculates the deviations and trends in power, voltage, and temperature based on the normalization results, generating the main characteristic quantities of node operation. Taking a typical scenario as an example, when the output power of a node in the previous cycle is lower than the terminal's required power, and the bus voltage simultaneously shows a downward trend, the power and voltage deviations exhibited by that node in the characteristic quantities will become important inputs for subsequent fuzzy inference. To improve noise resistance, the terminal employs a strategy combining amplitude limiting and window smoothing in the characteristic quantity calculation path, and attaches a source identifier and timestamp to each characteristic quantity to facilitate problem backtracking in subsequent modules.

[0038] Based on this, the terminal forms a set of characteristic quantities for each charging pile node, which includes at least power deviation, voltage deviation, power change trend and temperature change trend, and maintains a one-to-one correspondence with the original sampled data, in order to support subsequent nonlinear composite and fuzzy inference.

[0039] S103. Perform nonlinear composite deviation calculation on the characteristic quantities of each charging pile node to obtain the comprehensive state deviation; The terminal has a pre-set nonlinear mapping and weight configuration in the parameter area. The weights are derived from the prototype calibration and trial operation statistics and are managed in a controlled manner by version number. Parameter loading and rollback records audit information to ensure mass production consistency.

[0040] The overall state deviation refers to a composite evaluation index obtained by fusing multiple characteristic quantities of node operation (such as power, voltage, temperature, etc.) according to a nonlinear weight relationship, which is used to reflect the overall operating status of the node.

[0041] Unlike traditional linear weighted calculations, this embodiment uses a nonlinear approach to fuse multidimensional data, enabling more sensitive responses to state changes under complex operating conditions. A compressed mapping is used for deviations with excessive amplitude to prevent extreme values ​​from dominating the output; an enhanced mapping is used for safety-sensitive quantities such as temperature changes to improve early warning capabilities; boundaries are set for each mapping interval based on equipment safety specifications and calibration results, and on-site loading is performed according to the equipment type profile.

[0042] For example, when node power output is stable but temperature rises rapidly, nonlinear calculations will assign higher weight to temperature parameters, thus reflecting potential overheating risks in advance in the calculation results. Through this composite deviation calculation method, the terminal can refine the identification and response of single-node states without changing the overall power output structure, providing more representative input signals for subsequent fuzzy inference. To ensure engineering feasibility, the terminal performs consistency checks on input features (including multi-source temperature channel cross-validation and missing rate detection) before composite calculation, and sets a confidence interval and limit-crossing flag for the comprehensive state deviation after composite calculation, facilitating boundary processing and protection strategies in the subsequent fuzzification stage.

[0043] S104. The comprehensive state deviation and the feature quantity are fuzzed to obtain the membership value; Membership value is a quantitative indicator used in fuzzy control to describe the relationship between input variables and fuzzy sets, mapping continuous numerical values ​​to linguistic variables. The purpose of fuzzification is to transform complex numerical states into fuzzy linguistic forms for subsequent fuzzy inference calculations.

[0044] In this embodiment, the membership function is used to map continuous variables to language level values; language levels include low, moderate, high or slow, average, fast, etc. The terminal has preset membership function forms and level boundaries in the device model range. The boundary values ​​are derived from factory calibration and security specifications, loaded with the model and managed by the version number.

[0045] Specifically, when the overall state deviation of a node is within a small range, the terminal can fuzzify it to a smaller deviation and assign a corresponding membership value. When the temperature rises rapidly, the temperature change rate can be mapped to a high-rising fuzzy set and obtain a higher membership degree. To ensure numerical coverage and robustness, the terminal configures membership functions for the overall state deviation, power deviation, bus voltage deviation, and temperature change trend. Narrower curves are used for boundary-sensitive quantities (such as temperature change trends) to enhance boundary identification, while wider curves are used for quantities that need to suppress jitter (such as power deviation) to improve stability. The functions overlap in adjacent level intervals to absorb measurement noise and avoid critical jumps.

[0046] During runtime, the terminal performs a consistency check on the membership calculation results. If any individual input is missing or out of bounds, the boundary value of the neighboring interval is used to substitute and the result is marked as a downgraded state to avoid fuzzy inference failure due to single point of failure.

[0047] S105. Calculate the trigger strength of each fuzzy rule based on the membership value, and update the weight of each fuzzy rule based on the trigger strength to obtain the target weight; In this embodiment, the rule base consists of several rules that output a trend level if the input language condition is met. For example, if the overall state deviation is moderate and the temperature change trend is slow, the output power increases slightly. When multiple rules are satisfied simultaneously and the output directions are opposite, the terminal retains all trigger results for subsequent weighted aggregation, and the synthesis trend is determined by the weights and coverage intervals, without prior elimination.

[0048] Trigger strength reflects the degree to which each fuzzy rule is activated in the current state, while weight update refers to dynamically adjusting the importance of each rule based on these trigger strengths.

[0049] In this embodiment, the rule weights are automatically updated in each sampling period. The terminal evaluates the trigger strength of each rule based on its membership degree and performs convergence evaluation using the consistency criterion with the conditions, or performs expansion evaluation using the coverage criterion with the conditions. At the same time, a near-end time window is introduced to statistically analyze the matching degree between each rule trigger and the actual output, which is used as the basis for subsequent weight adjustment, enabling the terminal to have self-learning capabilities.

[0050] To ensure numerical stability, the terminal sets upper and lower limits for rule weights and performs normalization processing in each sampling period to keep the sum of all weights equal to one. To avoid long-term monopoly of a single rule, a slow decay factor is introduced to gradually reduce the weight of rules that have not been triggered for a long time or whose contribution has decreased. In the initial stage, the weights of each rule are loaded according to equal weights or the default weight of the pattern to ensure controllability during the cold start period.

[0051] For example, when a rule is frequently triggered over multiple cycles and its control effect is stable, the weight of that rule will gradually increase so that it will have a greater influence in subsequent reasoning; conversely, if a rule has a poor effect after being triggered for a long time, its weight will automatically decrease.

[0052] The terminal implements versioned management of the rule base and weight configuration, recording the loading time, version number and operator identifier; when the field environment or equipment type changes, the terminal completes consistency verification and rollback protection according to the process to ensure that the weight update process is traceable.

[0053] S106. The outputs of each fuzzy rule are weighted and aggregated according to the target weight, and then defuzzified using the centroid method to obtain the power adjustment coefficient. In this embodiment, the terminal sets the output universe of discourse and unit range for defuzzification, so that the power adjustment coefficient represents the increase or decrease in the form of continuous real numbers; the upper and lower bounds of the output universe of discourse are consistent with the power change steps allowed by the device, so as to avoid generating unexecutable control quantities.

[0054] Centroid method is a common defuzzification algorithm. The terminal first weights and aggregates the output trends of each rule according to the target weight, and then performs centroid calculation on the output universe to obtain a single-value result. To reduce boundary jitter, the terminal applies lightweight smoothing and rate limiting constraints to the defuzzification results of adjacent sampling periods, ensuring continuous output variation under small perturbations and avoiding overshoot under large perturbations. The core idea of ​​centroid method is to average the outputs of each fuzzy rule according to their weights, thereby obtaining a clear control output value. In this embodiment, the terminal weights the output trends of each rule according to the target weights obtained in the previous stage, synthesizing a continuous power adjustment signal. The power adjustment coefficient is used to characterize whether the current terminal power should be increased or decreased to achieve consistency with the target power requirement.

[0055] For example, when the overall power of multiple nodes is low and the temperature rise is stable, the power adjustment coefficient of the terminal output is positive, indicating that the power should be increased; if the node temperature is generally high, the power adjustment coefficient tends to be negative, indicating that the output should be reduced to prevent overload. When the aggregation result exceeds the output domain or has a significant conflict with the current device state (e.g., it is recommended to increase power even when the temperature is close to the upper limit), the terminal performs a consistency check according to the safety priority principle and adjusts to the nearest executable value, while recording the event for traceability.

[0056] S107. Perform nonlinear output correction on the power adjustment coefficient to obtain the target adjustment coefficient; In this embodiment, the terminal sets the limiting boundary based on the maximum power step allowed by the device, the allowable fluctuation range of the bus, and the node thermal load capacity. The dead zone threshold is set based on the minimum perceptible change under steady state to filter out minor disturbances. The nonlinear correction curve remains sensitive in the small amplitude range and gradually converges in the large amplitude range to avoid overshoot and oscillation.

[0057] Specifically, the terminal reduces the frequency of frequent adjustments caused by small fluctuations by setting output limiting, dead zone, and nonlinear correction curves. The corrected target adjustment coefficient maintains responsiveness while possessing anti-interference capabilities. To prevent excessively rapid output changes between adjacent sampling periods, the terminal applies an upper limit to the rate of change of the target adjustment coefficient and sets upper and lower hysteresis intervals near the threshold to prevent power jitter caused by frequent back-and-forth switching, thus providing a reliable signal for the safe adjustment of charging pile node power. When input data is missing or exceeds limits, the terminal activates a conservative correction strategy and records the event identifier and timestamp for subsequent traceability and parameter tuning verification.

[0058] S108. Calculate the adjustment power corresponding to the charging pile node based on the target adjustment coefficient; After the terminal adjusts the power of the computing node, it first performs boundary verification, including the node's maximum allowable power, temperature limit and bus constraints. If the verification fails, the target adjustment coefficient is cut to the nearest executable range, and the adjustment behavior is marked as protection trigger by the flag bit.

[0059] The adjusted power corresponding to each charging pile node is calculated based on the target adjustment coefficient. The terminal multiplies the target adjustment coefficient of each node by the current output power to obtain the adjusted power value of the node. This adjustment result is fed back to the power conversion unit of each node. To ensure execution reliability, the terminal issues commands using a reliable message mechanism with confirmation. If no confirmation is received from the node within a set time window, the terminal records the event and retryes according to the safety policy. If the retry fails, the previous valid command is maintained and a degraded operation flag is set until the link is restored, so as to correct the output in real time. For example, when the temperature of a node rises rapidly, its target adjustment coefficient is relatively low, thereby automatically reducing the output power to prevent the temperature from rising further. The terminal performs a consistency check between the issued result and the actual power read back. If the deviation exceeds the threshold, local protection is triggered and the adjustment range of subsequent single cycles is limited to ensure continuous and safe output.

[0060] S109. The adjustment power of each charging pile node is consistently corrected according to the total demand power to obtain the target power, which is used to control the output power of the corresponding charging pile node.

[0061] In this embodiment, the uniformity correction is based on meeting the total power demand, which is the power required by the power supply to support the operation of all charging piles under the premise that the charging piles are in a stable charging state. Following the principles of node electrical and thermal constraints, priority of available margin, and smoothness of change, the preliminary adjustment results are proportionally and constrainedly redistributed to avoid single node exceeding limits or sudden changes.

[0062] The terminal performs a consistency adjustment of the power of each charging pile node based on the total power demand to obtain the target power and control the node output. Before writing the target power, the terminal performs a safety interlock check: if the temperature is detected to be close to the upper limit or the bus fluctuation exceeds the limit, which conflicts with the power increase command, the terminal will prioritize the reduction or maintenance according to the safety strategy and record the event for traceability. The purpose is to keep the terminal's total power consistent with the demand power, while ensuring a coordinated and balanced load distribution among the nodes. Through consistency adjustment, the terminal can dynamically balance node differences while ensuring overall output, achieving optimized energy allocation for group operation. The terminal writes the consistent target power to the target channel of each node, and after confirmation of the write, it enters the data acquisition and calculation of the next sampling cycle, thereby maintaining closed-loop control of acquisition, decision-making, execution, and verification.

[0063] This embodiment receives feedback data from charging pile nodes through a fixed sampling period and preprocesses the feedback data to obtain characteristic quantities. Based on this, nonlinear composite deviation calculations are performed on the characteristic quantities to characterize the node operating status. An adaptive inference model is established through fuzzification processing and a dynamic weight update mechanism to achieve real-time correction of the power adjustment coefficient. After nonlinear output correction and consistency correction, the output target power can simultaneously meet the total power demand and the operating constraints of each node. Therefore, this scheme can achieve rapid response and stable control of power adjustment during the operation of a charging pile group, effectively improving the dynamic adaptability of multi-node collaboration and the overall operational stability.

[0064] Please see Figure 2a and Figure 2b This application provides another embodiment of the fuzzy control power dynamic adjustment method for charging pile nodes, which includes: S201. Receive feedback data from all charging pile nodes according to a fixed sampling period; Step S201 in this embodiment is similar to step S101 in the previous embodiment, and will not be described in detail here.

[0065] S202. The feedback data is cleaned and normalized to obtain the calculation parameters; The data cleaning process requires removing communication jitter or abnormal measurement points. For example, when a temperature sensor at a node experiences a sudden value jump, the jump value is considered invalid data and interpolated to complete the dataset. Normalization maps values ​​of different dimensions such as power, voltage, and temperature to a unified range, making subsequent calculations comparable. The processed data is labeled as calculation parameters and used as input for subsequent feature extraction.

[0066] S203. Calculate the power deviation, voltage deviation, power change rate, and temperature change rate of the charging pile node corresponding to the feedback data through calculation parameters, integrate the calculation results, and output the characteristic quantity. The required power is issued by the vehicle's battery management system; sampling period T s This is a constant for the terminal control cycle; a positive power deviation indicates that the terminal needs to increase power, and a negative value indicates that the power needs to be reduced; the voltage deviation is positively oriented with the reference voltage.

[0067] The terminal calculates the power deviation, voltage deviation, power change rate, and temperature change rate of each node based on the calculation parameters, thereby obtaining the node characteristic quantities. For example, in a typical case, if a node output power of 18 kW in the previous sampling period, and the terminal's power demand in the current period is 20 kW, the power deviation of that node is 2 kW. Simultaneously, if the bus voltage drops by 2% from the rated voltage, the voltage deviation is recorded as negative. If the temperature rises by 3℃ in an adjacent sampling period, the temperature change rate is 3℃ / T. s These features together constitute a four-dimensional input vector of the node's operating state, providing a basis for nonlinear deviation calculation.

[0068] The calculation methods for characteristic quantities are defined as follows: power deviation equals demanded power minus actual node power; voltage deviation equals reference voltage minus current bus voltage; power change rate equals the difference in power deviation between adjacent periods divided by the sampling period; temperature change rate equals the difference in temperature between adjacent periods divided by the sampling period.

[0069] Specifically, the characteristic quantities include power deviation, voltage deviation, power change rate, and temperature change rate. These parameters are calculated as follows: The formula for calculating the power deviation is:

[0070] Among them, P dem P is the power demand for the current cycle. i This represents the actual output power of the i-th charging pile node; The formula for calculating the voltage deviation is:

[0071] Among them, V ref The rated voltage, V busThis is the current bus voltage; The formula for calculating the power change rate is:

[0072] Where ΔP[k] and ΔP[k-1] are the power deviations between the current and previous sampling periods, respectively, and T s The sampling period; The formula for calculating the rate of temperature change is:

[0073] Among them, T i [k] and T i [k-1] represents the temperature of the i-th charging pile node in the current and previous sampling periods, respectively, T s The sampling period.

[0074] When any original channel is missing or goes out of bounds, the terminal replaces it with the most recent reliable value and marks the period as a degraded state. It does not participate in subsequent self-learning evaluation statistics, but still generates feature quantities to ensure continuous control.

[0075] S204. Perform nonlinear composite deviation calculation on the characteristic quantities of each charging pile node to obtain the comprehensive state deviation; The terminal has pre-configured nonlinear mapping and weights in the parameter area. The weights are derived from prototype calibration and trial operation statistics. Version numbers are used for management and loading time and operator identification are recorded. Consistency verification and safe rollback are supported.

[0076] The formula for calculating the nonlinear composite deviation is:

[0077] Where, ΔE i Let ΔP be the overall state deviation of the i-th charging pile node, ΔP be the power deviation of the current node, and ΔV be the voltage deviation of the current node. The rate of change of power, Let be the rate of temperature change, and α, β, γ, and η be the weighting coefficients for power, voltage, rate of change, and temperature change, respectively.

[0078] Nonlinear composite processing follows two mapping strategies: compression mapping is used for large deviations to prevent extreme values ​​from dominating the output; enhancement mapping is used for safety-sensitive quantities such as temperature changes to improve early warning capabilities. The working and saturation ranges of each mapping curve are set with boundaries according to equipment safety specifications, and the load is applied on-site according to the model specifications.

[0079] The terminal will calculate the comprehensive status index of each node based on the above. Taking a high-temperature node as an example, even if the power deviation is small, when the temperature rises rapidly, the contribution of the exponential term will increase rapidly, causing the comprehensive deviation of the node to exceed the limit in advance, thereby triggering the terminal to reduce its power in subsequent steps.

[0080] Before compositing, the terminal performs a consistency check on the input features (multi-source temperature cross-validation, missing rate detection). After compositing, it sets a reliable range and an out-of-limit flag for the overall state deviation, which facilitates boundary processing and protection strategy triggering in the subsequent fuzzing stage.

[0081] S205. The comprehensive state deviation and the feature quantity are fuzzed to obtain the membership value; The terminal configures membership functions and language levels for comprehensive state deviation, power deviation, voltage deviation, and temperature change trends, respectively. Narrower curves are used to enhance boundary recognition for boundary-sensitive quantities, while wider curves are used to enhance stability for quantities requiring jitter suppression. Language levels and boundary values ​​are derived from factory calibration and safety specifications, loaded with the device model range, and managed under version control.

[0082] The terminal inputs the comprehensive state deviation and corresponding feature quantity into the fuzzification module, mapping continuous numerical values ​​into fuzzy linguistic terms, such as high power, low voltage, and accelerated temperature rise. Through a set membership function, each input variable corresponds to multiple fuzzy sets and their membership values. For example, when ΔP = 0.15, this value has membership degrees of 0.4 and 0.6 in the normal and high sets, respectively, thus providing a linguistic and quantitative basis for subsequent reasoning.

[0083] The system maintains reasonable overlap between adjacent language levels to absorb measurement noise and avoid jumps near the threshold. When the input is missing or out of bounds, the terminal substitutes the nearest boundary value and marks the degraded state to ensure continuous fuzzification operation.

[0084] S206. Obtain a preset fuzzy rule set, wherein the fuzzy rule set contains the initial weight of each fuzzy rule; The preset fuzzy rule set and initial weights are stored in the terminal's non-volatile memory and are loaded according to the model when the device is powered on; during the cold start phase, equal weights or model default weights are used to ensure controllability before historical statistics are formed.

[0085] The preset fuzzy rule set is stored in memory. This set is generated from historical operating data and statistical analysis of simulation and historical operating data, and includes the correspondence between input conditions such as power, voltage, and temperature and adjustment strategies. For example, the rule 'If the power is too high and the temperature rises rapidly, then reduce the output power' has an initial weight reflecting its importance. The terminal loads this rule set at the beginning of operation and assigns an initial weight to each rule, which serves as the basis for dynamic weight updates.

[0086] The rules are organized as follows: "If the input language condition is met, the output trend level will be determined." When multiple rules output directions contradict each other, the terminal does not perform prior elimination but retains them for subsequent weighted aggregation. The combined trend is determined by the weight and the coverage range. The terminal implements version management and signature verification for the rule set and initial weights, and records the version number, loading time and operator identifier to ensure traceability.

[0087] S207. Within the fixed sampling period, calculate the trigger intensity of each fuzzy rule based on the membership value; In this embodiment, the terminal measures each rule based on its membership value, with the trigger strength ranging from [0,1]. When a rule contains multiple input language conditions, the terminal uses a consistency criterion for AND relations and a coverage criterion for OR relations to complete the synthetic measurement. For a single input that is not reported or exceeds the limit, the terminal substitutes the nearest boundary value and marks the period as a degraded state. The trigger strength of the degraded period participates in the control calculation but is not included in the self-learning evaluation statistics.

[0088] Trigger strength indicates the degree to which a rule is activated under the current operating conditions. For example, in a scenario with high power and rapid temperature rise, the trigger strength of the corresponding rule might reach 0.85, while the rule maintaining power might only have 0.2. The trigger strength of all rules is recorded in real time for subsequent weight adjustments. The terminal saves a timestamp sequence of the trigger strength for each rule and maintains near-end statistics according to a preset sliding window for stability determination in subsequent weight adjustments, avoiding misadjustments caused by single-cycle noise. When multiple rules are activated simultaneously with opposite output directions, the terminal does not perform prior rejection, retaining all trigger strengths for subsequent aggregation, where the synthesis trend is jointly determined by the subsequent weights and coverage range.

[0089] S208. Calculate the average trigger strength of all fuzzy rule trigger strengths; The average trigger strength is calculated based on the set of valid rules for the current period. When the number of valid rules is less than a preset threshold, the terminal uses the average value from the previous valid period to ensure numerical continuity. The terminal calculates the average trigger strength of all fuzzy rules and compares the deviation of each rule's trigger strength accordingly. To obtain robust judgment, the terminal performs sliding statistics on the average trigger strength and the trigger strength of each individual rule based on a preset window length, and uses amplitude limiting to handle abnormal changes, avoiding extreme values ​​from dragging down the window average, and providing a basis for dynamic weight adjustment. When a rule's trigger strength is consistently higher than the average, the terminal determines that the rule is too strong in the current load mode, and its weight will be reduced in subsequent periods; conversely, if a rule's trigger strength is consistently lower than the average but the associated control effect is good, its weight will gradually increase. Through this average trigger strength calculation and dynamic comparison mechanism, fuzzy control enables the terminal to achieve adaptive learning capabilities, allowing the control strategy to continuously adapt to the operating characteristics of the charging pile group. The terminal only performs weight adjustment when the window statistics meet the continuity condition; when in degraded operation or when the data quality does not meet the requirements, the terminal freezes weight updates and records the reason for freezing and the timestamp.

[0090] S209. The initial weights of the fuzzy rules are dynamically adjusted according to the weight update formula to obtain the target weights; In this embodiment, the terminal sets upper and lower bounds for the weight adjustment rate, sets minimum and maximum constraints for the weight of a single rule, and performs normalization processing on all weights in each sampling period to keep the total weights equal to one, thus preventing numerical drift or long-term monopoly of individual rules.

[0091] The weight update formula is:

[0092] Among them, w r (k) represents the weight of the r-th fuzzy rule in the current period, w r (k-1) represents the weight of the r-th fuzzy rule in the previous cycle, and ρ is the weight adjustment rate. r (k) represents the trigger strength of the current fuzzy rule. This is the average trigger strength of all fuzzy rules.

[0093] In the initial stage, a cold start is performed using equal weights or the default weights of the device type. When parameter inconsistencies are detected or the device type is switched on-site, the terminal completes a consistency check before initiating weight updates and supports rollback to the most recent valid weight version to ensure operational continuity. This process enables the terminal to have real-time self-learning capabilities. For example, when the vehicle's BMS suddenly increases charging demand, causing a continuous increase in power deviation, the terminal automatically increases the weight of rules related to rapid power increase; while during continuous high-temperature operation, the terminal reduces the weight of power-increasing rules and strengthens power-reducing rules to match the overall output response with the actual thermal state. Dynamic weight adjustment ensures a balance between stability and sensitivity of the terminal under different load, temperature, and voltage conditions.

[0094] When the terminal is in degraded operation (such as when communication packet loss exceeds the threshold or sensor malfunctions and fails to recover), the terminal suspends weight updates and records audit information, including timestamps, window statistics, and reasons for freezing, for easy traceability.

[0095] S210. The outputs of each fuzzy rule are weighted and aggregated according to the target weight, and then defuzzified using the centroid method to obtain the power adjustment coefficient. The terminal sets the output universe of discourse and unit for deblurring, ensuring that the power adjustment coefficient represents the increase or decrease as a continuous real number. The upper and lower bounds of the output are consistent with the single-cycle power step size allowed by the device, ensuring that the generated output can be directly executed. The terminal first calculates the weighted contribution of each fuzzy rule based on its output trend (e.g., power increase, power decrease, or hold) and the target weight, and then aggregates and sums all the results. Afterward, the control signal is obtained using the centroid method to obtain the continuous power adjustment coefficient. To reduce boundary jitter, the terminal applies a light smoothing and a rate-of-change cap to the deblurring results of adjacent cycles, ensuring that the output changes continuously under small perturbations and avoids overshoot under large perturbations.

[0096] Taking a group charging scenario as an example, when the output power of some modules is low while the temperature is within the safe range, the terminal aggregation result tends to be positive, indicating that the output of that module should be increased. However, if the voltage fluctuation is large or the temperature rise of a certain module is significant, the aggregation result tends to be negative, indicating that the output should be reduced to maintain terminal balance. The defuzzified power adjustment coefficient has smoothness and real-time performance, making it easy for the execution unit to control directly. When the aggregation result conflicts with the current safety boundary (such as approaching the upper temperature limit but still showing an increasing power trend), the terminal performs a consistency check according to the safety priority principle and trims it to the nearest executable value, while recording the event for traceability.

[0097] S211. The power adjustment coefficient is subjected to amplitude limiting and dead zone processing to obtain effective data; In this embodiment, the terminal sets the limiting boundary based on the maximum power step allowed by the device, the allowable fluctuation range of the bus, and the node thermal load capacity; the dead zone threshold is set based on the minimum perceptible change under steady state, and is used to filter out disturbances.

[0098] The terminal performs amplitude limiting and dead-zone processing on the power adjustment coefficient to eliminate ineffective adjustments caused by minor disturbances. Amplitude limiting ensures that the adjustment coefficient does not exceed a predetermined range, such as ±10%, preventing over-correction that could lead to power oscillations. Dead-zone processing maintains a constant output when the power adjustment coefficient is close to zero to avoid instability in the module controller's pulse width modulation adjustment caused by frequent switching. Through this process, the terminal filters out high-frequency fluctuations caused by noise, improving the smoothness of the overall output. To avoid threshold triggering repeatedly, the terminal sets upper and lower hysteresis intervals near the zero point; simultaneously, it imposes an upper limit on the single-cycle change rate of the power adjustment coefficient to limit the amplification of high-frequency jitter to the execution end.

[0099] When input data is missing or out of bounds, the terminal uses the nearest boundary value and conservative dead zone processing, and records the timestamp and channel identifier to facilitate subsequent traceability and parameter tuning verification.

[0100] S212. The effective data is calculated using a nonlinear output correction algorithm to obtain the target adjustment coefficient; The calculation formula for the nonlinear output correction algorithm is: λ′=k1·λ+k2·λ 3 Where λ′ is the target adjustment coefficient, λ is the effective data of the power adjustment coefficient, k1 is the linear gain coefficient, and k2 is the nonlinear correction coefficient.

[0101] k1 and k2 are derived from prototype calibration and trial operation statistics, and are managed in a versioned manner in the terminal parameter area; the selection of coefficients follows the principles of monotonicity and sign preservation to ensure sensitive response in the small amplitude range and gradual convergence in the large amplitude range, and to avoid overshoot.

[0102] By introducing a cubic term, the terminal maintains a linear response within a small range, while automatically reducing the rate of output change during large adjustments, thereby suppressing overshoot and improving control stability. For example, when it is determined that power should be significantly reduced, the nonlinear correction slows down the rate of decrease, preventing a sudden rise in bus voltage from triggering inverter protection. Through this process, the control signal is optimized between dynamic response and steady-state balance.

[0103] The terminal applies a limit to λ′ consistent with the device's permitted step size and activates an anti-saturation strategy when approaching the saturation region to prevent unexecutable targets from appearing in the subsequent scaling-up phase. When in degraded operation or when data quality does not meet requirements, the terminal freezes the k1 and k2 adjustments and uses the most recent valid coefficient combination to ensure continuous output.

[0104] S213. Calculate the adjustment power corresponding to the charging pile node based on the target adjustment coefficient; After the terminal adjusts the power of the computing node, it first performs boundary verification: including the maximum / minimum allowable power of a single node, the upper limit of temperature, and bus constraints; if the conditions are not met, the value is trimmed to the nearest executable range and marked as protection trigger.

[0105] The terminal adjusts the coefficient λ according to the target. i Calculate the adjusted power for the corresponding charging pile node. The adjusted power for each node is calculated by multiplying its current power by (1 + λ). i λ is obtained from λ'). For example, if the current power of a node is 20 kW, λ' is... i If the value is -0.05, the new adjusted power is 19 kW. The terminal uses this adjusted value as a power control reference and sends it in real time to the power conversion units (DC converters or inverter modules) of each module. The power output correction is then performed by the current closed-loop regulation loop within the module. The terminal issues commands using a reliable message mechanism with confirmation. If no confirmation is received within the set time window, the event is recorded and the system retryes according to the safety policy. If the retry fails, the previous valid command is maintained and a degradation flag is set until the link is restored.

[0106] When a conflict arises between the triggering of stable parameters and the instantaneous power increase requirement of the BMS, the terminal will first execute power reduction or maintenance according to the safety priority policy, and record the event and current window statistics for traceability and parameter tuning.

[0107] S214. The adjustment power of each charging pile node is consistently corrected according to the total demand power to obtain the target power, which is used to control the output power of the corresponding charging pile node.

[0108] The terminal performs a consistent power adjustment on all nodes based on the total power demand to obtain the final target power, ensuring that the overall output meets BMS requirements. The adjustment method is as follows: P i (ref) = P i (adj) × (P_demand / ΣP j (adj)); Where, ΣP j (adj) represents the sum of adjusted power across all nodes, and P_demand represents the total power demand of the terminal. i (ref) represents the target power of the i-th charging pile node (or power module).

[0109] When ΣP jWhen (adj) approaches zero, the terminal uses the sum of the previous effective cycle or sets a lower limit to replace it to avoid division by zero errors; if the target power of any node exceeds its electrical or thermal constraints after proportional correction, the terminal cuts the target of that node to the nearest feasible value according to the 'safety first' strategy, and redistributes the remaining power among other nodes with margins until the total required power is met or the global feasible boundary is reached.

[0110] Through this proportional correction, if some nodes experience a power reduction due to temperature rise, other nodes will automatically share the remaining load, thus maintaining a constant total output power. For example, when the total demand power is 120 kW, but the total power before correction is only 114 kW, the terminal automatically increases the output of other nodes proportionally, bringing the total output back to the demand value and achieving energy self-balancing among multiple nodes. To avoid sudden power changes within the group, the terminal adjusts the P... i (ref) Set the upper limit of the single-cycle change rate and use a soft slope before writing the target to make the power of each node smoothly transition to the new target according to the set slope, reducing bus fluctuations and thermal shock.

[0111] After the terminal completes writing the target power to each node, it requests the node to send back confirmation. If no confirmation is received within the set time window, the terminal records the event and retryes according to the security policy. If the retry fails, the previous valid target is maintained and a degradation flag is set. After confirmation, the next sampling cycle of acquisition and processing begins.

[0112] S215. Calculate the stability parameters of each charging pile node based on the temperature change rate and power change rate of each charging pile node. Stability parameters are represented by a weighted metric consisting of temperature change rate and power change rate, with values ​​normalized to [0,1]. The larger the value, the more stable the system. Thresholds are determined based on equipment safety specifications and trial operation statistics, and are loaded into the terminal parameter area for version management according to equipment type.

[0113] The terminal calculates stability parameters based on the temperature and power change rates of each node to assess the dynamic stability of the node's operating state. These stability parameters can be obtained using a weighted function of the temperature rise rate and power change rate. For example, when both the temperature and power change rates are small, the stability parameter is close to 1, indicating stable node operation. To enhance noise immunity and robustness, the terminal employs a sliding window and extremum removal statistics for the stability parameters. A node is considered unstable only when the stability parameter is below a threshold for N consecutive sampling periods and is not in a degraded operating state; N is specified by the model specification, typically taking 3–5 sampling periods.

[0114] If any parameter rises rapidly, the stability parameter decreases significantly, indicating a potential risk of thermal or electrical instability. This parameter serves as a trigger condition for redistribution in subsequent steps. When the stability parameter trigger conflicts with the BMS's instantaneous power increase demand, the terminal performs power reduction or maintenance on the relevant nodes according to the safety priority principle, and records the timestamp, window statistics, and cause of the conflict as a basis for subsequent traceability and parameter tuning.

[0115] S216. When the stability parameter is lower than the preset stability threshold, the power redistribution logic is triggered to recalculate the target power adjustment coefficient.

[0116] To avoid oscillations caused by frequent entry and exit from the redistribution process, the terminal sets an upper and lower hysteresis range near the stability threshold and sets a minimum recalculation interval; if no more serious signs of instability appear during the minimum recalculation interval, the terminal will not trigger redistribution again.

[0117] When the stability parameter falls below the preset stability threshold, the terminal triggers the power redistribution logic to recalculate the target power adjustment coefficient. At this point, the terminal no longer uses the fuzzy inference result from the previous cycle, but re-executes the calculation process from S204 to S212 to generate a new λ. i 'distributed.

[0118] The recalculation scope includes: the latest valid set of features, comprehensive state bias, fuzzification, rule triggering strength and weight update, aggregation and defuzzification, and nonlinear output correction, forming a new λ. i The terminal freezes weight updates and uses the most recent valid weights if communication degradation or critical sensor anomalies persist. This avoids learning from abnormal data. For example, if a module's temperature rise rate exceeds the limit or its output fluctuation exceeds a threshold, the terminal immediately reduces its power target and correspondingly increases the output of other modules to maintain the overall total power. This mechanism allows the terminal to self-correct early in the abnormal operating trend, thus avoiding heat accumulation and voltage oscillations. In redistribution scenarios, the terminal prioritizes power reduction for unstable nodes and synchronously increases power proportionally for nodes with available capacity. Subsequently, it re-executes the consistency correction in S214 and completes the write confirmation. If the total power demand cannot be met even after reaching the global feasible boundary, the terminal reports a load reduction status and maintains a safe and feasible solution until conditions are restored.

[0119] The terminal records the triggering reason, timestamp, key parameters (stability parameters, threshold, hysteresis boundary, minimum recalculation interval), power vector before and after adjustment, and write confirmation results for this redistribution. When the stability parameters continuously recover to above the threshold and maintain a continuous recovery period of M sampling periods, the terminal exits the redistribution mode and removes the degradation flag, where M is specified by the model spectrum.

[0120] This embodiment performs feedback acquisition, feature extraction, nonlinear composite deviation calculation, fuzzy inference, dynamic weight update, and nonlinear output correction on charging pile nodes through a fixed sampling period. Combined with a consistency correction and stability redistribution mechanism, it enables the terminal to achieve rapid, stable, and safe coordinated control of power under complex operating conditions.

[0121] The above describes in detail the fuzzy control power dynamic adjustment method for charging pile nodes in the embodiments of this application. The following will describe in detail the fuzzy control power dynamic adjustment system and device for charging pile nodes.

[0122] Please see Figure 3 This application provides an embodiment of a fuzzy control power dynamic adjustment system for charging pile nodes, which includes: The receiving unit 301 is used to receive feedback data from all charging pile nodes according to a fixed sampling period; The first calculation unit 302 is used to preprocess and calculate the feedback data to obtain feature quantities; The second calculation unit 303 is used to perform nonlinear composite deviation calculation on the characteristic quantities of each charging pile node to obtain the comprehensive state deviation. The fuzzification unit 304 is used to fuzzify the comprehensive state deviation and the feature quantity to obtain the membership value; The update unit 305 is used to calculate the trigger strength of each fuzzy rule based on the membership value, and update the weight of each fuzzy rule based on the trigger strength to obtain the target weight; Weighting unit 306 is used to perform weighted aggregation of each fuzzy rule output according to the target weight, and to defuzzify it by centroid method to obtain power adjustment coefficient; The correction unit 307 is used to perform nonlinear output correction on the power adjustment coefficient to obtain the target adjustment coefficient; The third calculation unit 308 is used to calculate the adjustment power corresponding to the charging pile node based on the target adjustment coefficient. The correction unit 309 is used to make consistent corrections to the adjustment power of each charging pile node according to the total demand power to obtain the target power, and the target power is used to control the output power of the corresponding charging pile node.

[0123] In this embodiment, the functions of each unit are the same as those described above. Figure 1 The steps in the illustrated embodiments are the same and will not be repeated here.

[0124] Please see Figure 4 This application provides an embodiment of a fuzzy control power dynamic adjustment system for charging pile nodes, which includes: The receiving unit 401 is used to receive feedback data from all charging pile nodes according to a fixed sampling period; The first calculation unit 402 is used to preprocess and calculate the feedback data to obtain feature quantities; The second calculation unit 403 is used to perform nonlinear composite deviation calculation on the characteristic quantities of each charging pile node to obtain the comprehensive state deviation. The fuzzification unit 404 is used to fuzzify the comprehensive state deviation and the feature quantity to obtain the membership value; The update unit 405 is used to calculate the trigger strength of each fuzzy rule based on the membership value, and update the weight of each fuzzy rule based on the trigger strength to obtain the target weight; Weighting unit 406 is used to perform weighted aggregation of each fuzzy rule output according to the target weight, and to defuzzify it by centroid method to obtain power adjustment coefficient; The correction unit 407 is used to perform nonlinear output correction on the power adjustment coefficient to obtain the target adjustment coefficient; The third calculation unit 408 is used to calculate the adjustment power corresponding to the charging pile node based on the target adjustment coefficient. The correction unit 409 is used to make consistent corrections to the adjustment power of each charging pile node according to the total demand power to obtain the target power, and the target power is used to control the output power of the corresponding charging pile node. The fourth calculation unit 410 is used to calculate the stability parameters of the corresponding charging pile node based on the temperature change rate and power change rate of each charging pile node. The redistribution unit 411 is used to trigger the power redistribution logic to recalculate the target power adjustment coefficient when the stability parameter is lower than the preset stability threshold.

[0125] In this embodiment, the update unit 405 is further configured to: Obtain a preset fuzzy rule set, which contains the initial weight of each fuzzy rule; Within the fixed sampling period, the trigger intensity of each fuzzy rule is calculated based on the membership value; Calculate the average trigger strength of all fuzzy rule trigger strengths; The initial weights of the fuzzy rules are dynamically adjusted based on the difference between the trigger strength of each fuzzy rule and the average trigger strength to obtain the target weights.

[0126] In this embodiment, the update unit 405 is further configured to: The initial weights of the fuzzy rules are dynamically adjusted according to the weight update formula to obtain the target weights; The weight update formula is:

[0127] Among them, w r (k) represents the weight of the r-th fuzzy rule in the current period, w r (k-1) represents the weight of the r-th fuzzy rule in the previous cycle, and ρ is the weight adjustment rate. r (k) represents the trigger strength of the current fuzzy rule. This is the average trigger strength of all fuzzy rules.

[0128] In this embodiment, the first computing unit 402 is further configured to: The feedback data is cleaned and normalized to obtain the calculation parameters; The power deviation, voltage deviation, power change rate, and temperature change rate of the charging pile node corresponding to the feedback data are calculated by calculation parameters. The calculation results are combined and the characteristic quantities are output. The power deviation is the difference between the required power and the actual output power of the charging pile node. The voltage deviation is the difference between the reference voltage and the current bus voltage of the charging pile node. The power change rate is the change in power deviation between two adjacent sampling periods of the charging pile node divided by the sampling period. The temperature change rate is the temperature difference between two adjacent sampling periods of the charging pile node divided by the sampling period.

[0129] In this embodiment, the second computing unit 403 is further configured to: The formula for calculating the nonlinear composite deviation is:

[0130] Where, ΔE i Let ΔP be the overall state deviation of the i-th charging pile node, ΔP be the power deviation, and ΔV be the voltage deviation. The rate of change of power, Let be the rate of temperature change, and α, β, γ, and η be the weighting coefficients for power, voltage, rate of change, and temperature change, respectively.

[0131] In this embodiment, the correction unit 407 is further configured to: The power adjustment coefficient is subjected to amplitude limiting and dead zone processing to obtain effective data; The effective data is calculated using a nonlinear output correction algorithm to obtain the target adjustment coefficient; The calculation formula for the nonlinear output correction algorithm is: λ′=k1·λ+k2·λ 3 Where λ′ is the target adjustment coefficient, λ is the effective data of the power adjustment coefficient, k1 is the linear gain coefficient, and k2 is the nonlinear correction coefficient.

[0132] In this embodiment, the functions of each unit are the same as those described above. Figure 2a and Figure 2b The steps in the illustrated embodiments are the same and will not be repeated here.

[0133] Please see Figure 5 This application provides an embodiment of a fuzzy control power dynamic adjustment device for charging pile nodes, comprising: Processor 501, memory 502, input / output unit 503, bus 504; The processor 501 is connected to the memory 502, the input / output unit 503 and the bus 504; The processor 501 specifically executes... Figures 1 to 2b The specific operations corresponding to the steps in the method will not be elaborated here.

[0134] This application also relates to a computer-readable storage medium on which a program is stored, which, when run on a computer, causes the computer to perform any of the methods described above.

[0135] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0136] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0137] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0138] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0139] If the integrated unit is implemented as 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 this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A fuzzy control power dynamic adjustment method for charging pile nodes, characterized in that, The method includes: Feedback data from all charging pile nodes is received according to a fixed sampling period; The feedback data is preprocessed and calculated to obtain feature quantities; Nonlinear composite deviation calculation is performed on the characteristic quantities of each charging pile node to obtain the comprehensive state deviation; The comprehensive state deviation and the feature quantity are fuzzified to obtain the membership value, which is a quantitative indicator used in fuzzy control to describe the relationship between input variables and fuzzy sets. The trigger strength of each fuzzy rule is calculated based on the membership value, and the weight of each fuzzy rule is updated based on the trigger strength to obtain the target weight; The outputs of each fuzzy rule are weighted and aggregated according to the target weight, and then defuzzified using the centroid method to obtain the power adjustment coefficient. The power adjustment coefficient is nonlinearly corrected to obtain the target adjustment coefficient. Calculate the adjustment power corresponding to the charging pile node based on the target adjustment coefficient; The power adjustment of each charging pile node is consistently corrected based on the total power demand to obtain the target power, which is used to control the output power of the corresponding charging pile node.

2. The method according to claim 1, characterized in that, The step of calculating the trigger strength of each fuzzy rule based on the membership value, and updating the weight of each fuzzy rule based on the trigger strength to obtain the target weight includes: Obtain a preset fuzzy rule set, which contains the initial weight of each fuzzy rule; Within the fixed sampling period, the trigger intensity of each fuzzy rule is calculated based on the membership value; Calculate the average trigger strength of all fuzzy rule trigger strengths; The initial weights of the fuzzy rules are dynamically adjusted based on the difference between the trigger strength of each fuzzy rule and the average trigger strength to obtain the target weights.

3. The method according to claim 2, characterized in that, The step of dynamically adjusting the initial weights of the fuzzy rules based on the difference between the trigger strength of each fuzzy rule and the average trigger strength to obtain the target weights includes: The initial weights of the fuzzy rules are dynamically adjusted according to the weight update formula to obtain the target weights; The weight update formula is: Among them, w r (k) represents the weight of the r-th fuzzy rule in the current period, w r (k-1) represents the weight of the r-th fuzzy rule in the previous cycle, and ρ is the weight adjustment rate. r (k) represents the trigger strength of the current fuzzy rule. This is the average trigger strength of all fuzzy rules.

4. The method according to claim 1, characterized in that, The preprocessing and calculation of the feedback data to obtain feature quantities includes: The feedback data is cleaned and normalized to obtain the calculation parameters; The power deviation, voltage deviation, power change rate, and temperature change rate of the charging pile node corresponding to the feedback data are calculated by calculation parameters. The calculation results are combined and the characteristic quantities are output. The power deviation is the difference between the required power and the actual output power of the charging pile node. The voltage deviation is the difference between the reference voltage and the current bus voltage of the charging pile node. The power change rate is the change in power deviation between two adjacent sampling periods of the charging pile node divided by the sampling period. The temperature change rate is the temperature difference between two adjacent sampling periods of the charging pile node divided by the sampling period.

5. The method according to claim 4, characterized in that, The nonlinear composite deviation calculation of the characteristic quantities of each charging pile node yields the comprehensive state deviation, including: The formula for calculating the nonlinear composite deviation is: Where, ΔE i Let ΔP be the overall state deviation of the i-th charging pile node, ΔP be the power deviation of the current node, and ΔV be the voltage deviation of the current node. The rate of change of power, Let be the rate of temperature change, and α, β, γ, and η be the weighting coefficients for power, voltage, rate of change, and temperature change, respectively.

6. The method according to any one of claims 1 to 5, characterized in that, The nonlinear output correction of the power adjustment coefficient to obtain the target adjustment coefficient includes: The power adjustment coefficient is subjected to amplitude limiting and dead zone processing to obtain effective data; The effective data is calculated using a nonlinear output correction algorithm to obtain the target adjustment coefficient; The calculation formula for the nonlinear output correction algorithm is: λ′=k1·λ+k2·λ 3 Where λ′ is the target adjustment coefficient, λ is the effective data of the power adjustment coefficient, k1 is the linear gain coefficient, and k2 is the nonlinear correction coefficient.

7. The method according to any one of claims 1 to 5, characterized in that, After the method involves consistently adjusting the power of each charging pile node based on the total power demand to obtain the target power, it further includes: The stability parameters of each charging pile node are calculated based on the temperature change rate and power change rate of each charging pile node. When the stability parameter is lower than the preset stability threshold, the power redistribution logic is triggered to recalculate the target power adjustment coefficient.

8. A fuzzy control power dynamic adjustment system for a charging pile node, characterized in that, The system includes: The receiving unit is used to receive feedback data from all charging pile nodes according to a fixed sampling period; The first calculation unit is used to preprocess and calculate the feedback data to obtain feature quantities; The second calculation unit is used to perform nonlinear composite deviation calculation on the characteristic quantities of each charging pile node to obtain the comprehensive state deviation. A fuzzification unit is used to fuzzify the comprehensive state deviation and the feature quantity to obtain a membership value. The update unit is used to calculate the trigger strength of each fuzzy rule based on the membership value, and update the weight of each fuzzy rule based on the trigger strength to obtain the target weight; The weighting unit is used to weight and aggregate the outputs of each fuzzy rule according to the target weight, and then defuzzify them using the centroid method to obtain the power adjustment coefficient. The correction unit is used to perform nonlinear output correction on the power adjustment coefficient to obtain the target adjustment coefficient; The third calculation unit is used to calculate the adjustment power corresponding to the charging pile node based on the target adjustment coefficient. The correction unit is used to make consistent corrections to the adjustment power of each charging pile node according to the total demand power to obtain the target power, which is used to control the output power of the corresponding charging pile node.

9. A fuzzy control power dynamic adjustment device for a charging pile node, characterized in that, The device includes: Processor, memory, input / output units, and bus; The processor is connected to the memory, the input / output unit, and the bus; The memory stores a program, which the processor invokes to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains a program that, when executed on a computer, performs the method as described in any one of claims 1 to 7.

Images