A charging pile system management method based on dynamic load balancing

Abstract

The application discloses a charging pile system management method based on dynamic load balancing, and particularly relates to the technical field of charging pile management. Time domain scheduling data of a charging pile side, a power grid side and a distributed renewable energy output device side is collected, a station level state vector is obtained by fusion filtering, a station level prediction sequence of the charging pile in a next control period is generated based on the station level state vector and a rolling prediction model; a station level state vector and a station level prediction sequence are read by an inner layer controller, a power prediction optimization model is established and solved, and an expected station level power trajectory and a station level power redundancy are output; a result output by the inner layer controller is received, a station level power envelope curve and energy storage charging and discharging set values are comprehensively solved, and transmission is performed; a slope upper limit of the station level power envelope curve is contracted based on an oscillation amplitude, the new slope upper limit is substituted into the next solving, a subsequent power change rate is limited, and an unstable power exchange problem between a charging station and a superior power grid is effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of charging pile management, and more specifically, to a charging pile system management method based on dynamic load balancing. Background Art

[0002] With the parallel deployment of high-power DC fast charging technology and distributed renewable energy output devices, charging infrastructure has gradually evolved into an "active load" node in the power system. Many charging stations are coupled to the public distribution network through three-phase low-voltage busbars and are equipped with distributed renewable energy output devices such as rooftop photovoltaics, energy storage batteries, and power converters. The instability of renewable energy output devices and grid dispatch instructions increases the difficulty of dynamic load balancing of charging pile systems. Specifically,

[0003] The externally given power boundary and the random fluctuations of renewable power cause the site load to exhibit rapid, bidirectional, and unpredictable jump characteristics. The time constant of its change is significantly mismatched with the vehicle battery charging time constant and the distribution network voltage regulating device action time constant, thereby introducing a multi-time scale coupled operation scenario; the exposed technical problem is: how to meet the instantaneous power demand of the vehicle while ensuring that the power exchange between the charging station and the upper-level power grid remains steady-state and safe, that is, to avoid rapid drop or surge of bus voltage, excessive frequency deviation and deterioration of power quality. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a charging pile system management method based on dynamic load balancing to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a charging pile system management method based on dynamic load balancing, wherein the charging pile system and the distributed renewable energy output device are deployed in parallel, comprising the following steps:

[0006] Step 1: Multi-time domain state acquisition and fusion prediction, including: collecting the first time domain operation data of the charging pile side, the second time domain scheduling data of the grid side, and the third time domain scheduling data of the distributed renewable energy output device side, fusing and filtering to obtain the station-level state vector, and generating the station-level prediction sequence of the charging pile in the next control cycle based on the station-level state vector and the rolling prediction model;

[0007] Step 2: Inner prediction and balancing control, including: the inner controller reads the station-level state vector and station-level prediction sequence, establishes and solves the power prediction optimization model, and outputs the expected station-level power trajectory and station-level power margin;

[0008] Step 3: Generate the outer power grid collaborative envelope, including: receiving the output of the inner controller, comprehensively solving the station-level power envelope curve and energy storage charging and discharging set values, and transmitting them;

[0009] Step 4: Update the control parameters, including adjusting the smoothing weight factor, power damping dead zone boundary, and slope upper limit, and synchronously write them into the status database; shrink the slope upper limit of the station-level power envelope curve based on the oscillation amplitude, and substitute the new slope upper limit into the next solution to limit the subsequent power change rate.

[0010] Preferably, the first time domain operation data is collected through sensors deployed on each charging pile, and includes at least instantaneous current, instantaneous voltage, vehicle battery power and temperature inside the pile; the second time domain scheduling data is received through a communication link with an external power grid management system, and includes at least power grid scheduling instructions, real-time electricity price signals, common bus voltage and frequency; the third time domain scheduling data includes environmental parameters and energy storage status that affect renewable energy output; the station-level state vector refers to a multidimensional vector constructed based on the fusion filtering results, including the total power of the charging pile, the power distribution of each charging pile, the bus voltage frequency, prediction error, renewable energy output status and energy storage status, which is used as the input of the station-level prediction sequence of the rolling prediction model.

[0011] Preferably, the rolling prediction model includes a renewable energy output prediction channel, a station-level power prediction channel, and a grid-side voltage prediction channel. The renewable energy output prediction channel is used to predict the renewable energy output status and energy storage status of the next control cycle (the next 1-5 minutes); the station-level power prediction channel is used to predict the total power of the charging pile in the next control cycle; the grid-side voltage prediction channel is used to predict the bus voltage frequency in the next control cycle. The operation process of the rolling prediction model includes:

[0012] State space modeling: using the observed initial station-level state vector as input, linear time-varying state space equations for renewable energy forecast output, station-level forecast power, and bus voltage and frequency are established.

[0013] Explanation: PV output forecasts are generated by fitting short-term trends using historical irradiance data, irradiance sensor data, and the weather API interface. The bus voltage change rate is linearly related to the difference between the grid dispatch command power and the total charging pile power, with the proportionality coefficient determined through regression analysis of historical data. Renewable energy is affected by the integral of charge and discharge power and is constrained by the upper and lower limits of physical capacity.

[0014] Use recursive least squares or extended Kalman filtering to update the state transfer matrix and input gain matrix;

[0015] Multi-step forward deduction, after each control step, expands the identification forward to the prediction time domain and outputs the station-level prediction sequence;

[0016] Confidence interval evaluation, calculates the noise covariance through Bayesian estimation, and generates the upper and lower confidence boundaries corresponding to the renewable energy output forecast channel, station-level power forecast channel, and grid-side voltage forecast channel.

[0017] Preferably, the inner controller is used to quickly control the power of charging piles in the charging pile group local area network, and has a millisecond-level refresh cycle; the power prediction optimization model adopts a rolling optimization method and takes into account vehicle charging requirements and hardware safety constraints;

[0018] Explanation: The power prediction optimization model uses charging pile power as the decision variable, vehicle charging demand and hardware safety constraints as boundaries, and adopts a rolling optimization method based on the station-level prediction sequence of the next control cycle to minimize power fluctuations and imbalances between piles, thereby achieving load balancing of the charging piles. It continuously learns and optimizes the power prediction optimization model parameters based on historical charging data and real-time feedback to improve the accuracy and adaptability of load distribution; the charging pile power sequence is the power setting vector generated by the inner controller for each charging pile within the prediction window, and is sent to the execution layer in a time sequence; the station-level power redundancy is the difference between the total available power upper limit allowed by the charging pile hardware and the total output power after inner layer optimization; the expected station-level power trajectory is the station-level power change curve over time obtained by the inner controller summarizing the power sequences of each charging pile.

[0019] Preferably, the prediction step of the inner prediction model is equal to the refresh period multiplied by the optimization step number (the millisecond control period of the inner controller, a typical value is 1-10 milliseconds); the hardware safety constraints include that the current of a single pile does not exceed the hardware limit and the temperature inside the pile is lower than the safety threshold; the optimization step number is determined based on the balance between the system computing power and the prediction accuracy requirements.

[0020] Preferably, the outer controller is used to perform slow scheduling of the charging pile power, and the refresh period is in seconds (1-10 seconds); when the error between the actual power and the envelope curve alternates between positive and negative for three consecutive cycles and the absolute value increases, it triggers a warning state that stabilizes to an oscillation warning state, and adjusts the control strategy according to the situation.

[0021] Explanation: The slow scheduling of the outer controller includes the following state transition rules:

[0022] When the error between the actual power and the envelope curve alternates between positive and negative for three consecutive cycles and the absolute value increases, the warning state from the stable state to the oscillation warning state is triggered;

[0023] If the warning state lasts for more than two outer control cycles, the smooth weight factor increment strategy is activated, and the outer envelope slope update is frozen, triggering the intervention state from the oscillation warning state to the damping intervention state;

[0024] When the absolute value of the error continuously decreases to a preset range, the slope constraint is gradually released and the weight factor is reset, triggering the recovery state in which the control parameters are restored.

[0025] Preferably, the inner controller feeds back the imbalance degree between charging piles after power distribution (the standard deviation of the power of each charging pile) to the outer controller. The imbalance degree between charging piles is the standard deviation of the power distribution of each charging pile. The outer controller dynamically adjusts the distribution strategy of the station-level power redundancy based on the imbalance degree between charging piles. If the imbalance degree between charging piles increases, the outer controller increases the station-level power redundancy and relaxes the upper limit of the slope; if the imbalance degree between charging piles decreases, the outer controller reduces the station-level power redundancy and tightens the upper limit of the slope.

[0026] Preferably, in an off-grid microgrid scenario, the outer controller establishes a closed-loop association between power ramp rate limit and inertial support capability through equivalent inertia perception, dynamic adjustment of the power envelope slope upper limit, and energy storage priority compensation;

[0027] The outer controller performs the following coordinated control steps when the charging station operates in the off-grid microgrid mode:

[0028] Calculate the equivalent moment of inertia: Let n be the number of virtual synchronous generators, and t be the index number; let the output power of the t-th virtual synchronous generator be P gen,t ; The mechanical time constant of the t-th virtual synchronous generator is recorded as T m,t , reflecting the inertial response characteristics, the real-time charging and discharging power of the energy storage device is recorded as P ess ; The virtual inertia time constant of the energy storage device is recorded as T ess , preset through the configuration interface; the rated angular frequency is recorded as ω0; the maximum allowable frequency deviation is recorded as Δf max ; Through the formula The equivalent moment of inertia of the microgrid is calculated and used to characterize the ability of the charging pile system to resist frequency fluctuations;

[0029] Adjust the upper limit of the power envelope slope: according to the equivalent moment of inertia J eq Adaptively adjust the slope upper limit S of the station-level power envelope curve max , to ensure that the power change rate matches the system inertia, the adjustment formula is: Among them S base represents the preset slope upper limit, l represents the slope gain coefficient, which is obtained through historical data training; J min , J max They represent the minimum and maximum allowable equivalent moments of inertia respectively.

[0030] Preferably, the specific implementation process of step 4 includes the following steps:

[0031] Real-time collection of the actual station-level power after the inner controller executes each control cycle, and synchronous reception of the station-level power envelope curve set by the outer controller;

[0032] For each control cycle, the error sequence trend is determined based on the error between the actual station-level power and the power envelope curve. If the error reverses between two consecutive control cycles and the amplitude increase ratio exceeds the preset value, it is determined to be an oscillation trend.

[0033] When an oscillation trend is detected, the smoothing weight factor of the inner controller is increased to enhance the temporal continuity of power scheduling. A power damping dead zone boundary centered on the current power mean is set, and the slope upper limit of the station-level power envelope curve is dynamically reduced. This ensures that the error is within the power damping dead zone boundary, suppressing high-frequency corrections to the inner and outer control commands and preventing high-frequency loop coupling, which refers to phase resonance caused by the bandwidth overlap of the inner and outer control loops.

[0034] When the inner controller and the outer controller simultaneously and rapidly correct the power command and lack damping, the error sequence undergoes high-frequency reverse flipping and gradually amplifies, which is considered to be a high-frequency coupled oscillation. The smoothing weight factor is the continuity penalty coefficient in the objective function of the inner controller.

[0035] The updated control parameters include a smoothing weight factor, a power damping dead zone boundary, and an upper limit of an outer envelope slope.

[0036] Preferably, the power prediction optimization model includes a capacity drift self-calibration compensation step: continuously collecting energy storage current, terminal voltage, temperature and internal resistance within a sliding time window, suppressing noise on energy storage current and terminal voltage data through ampere-hour integration combined with Kalman filtering, and obtaining cumulative charge and discharge capacity to ensure the accuracy of capacity estimation; inputting the cumulative charge and discharge capacity and the nominal capacity into the recursive least squares algorithm to estimate the effective capacity online, evaluating the battery health status based on the increase in internal resistance and cycle life loss, and predicting the future capacity drift and health degradation trend of the battery based on historical data; using the difference between the nominal capacity and the effective capacity to represent the capacity drift, and according to the capacity drift The updated battery state and power mapping function are used to update the state; the updated power mapping function is used to regenerate the power difference correction value, and the charging and discharging power of the energy storage energy conversion device is adjusted according to the power difference correction value to complete the compensation; according to the assessed battery health state and the predicted future capacity drift and health degradation trend, the control strategy is dynamically adjusted, such as limiting the maximum charging current to slow down aging, or redistributing energy storage tasks among battery modules to balance the health state; the battery health state is integrated into the dynamic load balancing algorithm, high-load tasks are assigned to charging piles with better health status first, and the charging and discharging power of the energy storage energy conversion device is adjusted according to the real-time capacity and health state of the battery.

[0037] Technical effects and advantages of the present invention:

[0038] (1) The present invention effectively solves the time constant mismatch problem between charging stations and distribution network voltage regulators through the hierarchical coordination mechanism of inner and outer layer controllers (inner layer millisecond-level fast power distribution and outer layer second-level grid coordinated scheduling); the inner layer predictive balancing control quickly responds to vehicle demand based on the rolling optimized power trajectory, and the outer layer grid coordinated envelope ensures the steady-state safety of power exchange on the grid side through slope limitation, so that the minute-level battery charging process and the second-level grid voltage regulation demand are dynamically decoupled; by adopting the power damping dead zone boundary and the dynamic shrinkage strategy of the slope upper limit, when it is detected that the error sequence alternates between positive and negative and the amplitude increases, a phase compensation barrier is constructed by increasing the smoothing weight factor and shrinking the slope upper limit; the problem of low-frequency oscillation generated by the voltage regulator or inverter of the station and the distribution network is solved.

[0039] (2) This invention uses a three-channel fusion mechanism based on a rolling forecast model to spatially and temporally align irradiance sensor data with short-term forecast data from the weather API using the Kalman gain matrix. This reduces the forecast error for renewable energy output compared to traditional single-channel forecasting. The power envelope generated through confidence interval assessment provides a ±5% dynamic safety margin for outer control, avoiding power command overshoot due to forecast deviations. By unifying the timing alignment and forecast fusion mechanism, the problem of dynamic balancing algorithms making decisions based on lagged information is resolved. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a flowchart of the charging pile system management method based on dynamic load balancing of the present invention.

[0041] Figure 2 This is a flow chart of the control parameter update of the present invention. DETAILED DESCRIPTION

[0042] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0043] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0044] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present disclosure, its application, or uses.

[0045] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0046] Example 1, see Figure 1 The present invention provides a flowchart of a charging pile system management method based on dynamic load balancing. Figure 1 A charging pile system management method based on dynamic load balancing is shown, where the charging pile system is deployed in parallel with a distributed renewable energy output device, and includes the following steps:

[0047] Step 1: Multi-time domain state acquisition and fusion prediction, including: collecting the first time domain operation data of the charging pile side, the second time domain scheduling data of the grid side, and the third time domain scheduling data of the distributed renewable energy output device side, fusing and filtering to obtain the station-level state vector, and generating the station-level prediction sequence of the charging pile in the next control cycle based on the station-level state vector and the rolling prediction model;

[0048] Step 2: Inner prediction and balancing control, including: the inner controller reads the station-level state vector and station-level prediction sequence, establishes and solves the power prediction optimization model, and outputs the expected station-level power trajectory and station-level power margin;

[0049] The power forecast optimization model is an optimal scheduling model based on MPC or rolling horizon optimization. It continuously iterates to find the optimal power allocation through station-level state vectors and station-level prediction sequences. The objective function of the power forecast optimization model includes vehicle charging demand constraints and power smoothing penalties. The vehicle charging demand constraint dynamically allocates power based on the vehicle's remaining battery capacity and user priority weights. The power smoothing penalty is used to suppress power jumps between adjacent control steps.

[0050] Step 3: Generate the outer power grid collaborative envelope, including: receiving the output of the inner controller, comprehensively solving the station-level power envelope curve and energy storage charging and discharging set values, and transmitting them;

[0051] Explanation: The updated control parameters are synchronously written into the state database for the next round of multi-time domain state acquisition and fusion prediction to achieve stable operation of cross-time domain collaborative control. The updated control parameters include new parameters obtained through adaptive adjustment, such as the smoothing weight factor, power damping dead zone boundary, and slope upper limit.

[0052] Explanation: The smoothing weight factor refers to: the continuity penalty coefficient of the power smoothing penalty term in the objective function of the power prediction optimization model, which is used to suppress the power command jump between adjacent control steps. The larger the value, the smoother the output power; the power damping dead zone boundary refers to the power tolerance band set with the current total power mean as the center. Within the tolerance band, the command difference between the inner controller and the outer controller is temporarily ignored to prevent high-frequency micro-amplitude commands from superimposing each other; the slope upper limit of the station-level power envelope curve refers to: the maximum allowable slope of the station-level power envelope curve on the time axis, which is used to limit the total power change rate to prevent the outer scheduling from generating too fast power steps;

[0053] Step 4: Update the control parameters including adjusting the smoothing weight factor, power damping dead zone boundary, and slope upper limit, and write them into the state database synchronously;

[0054] Step 4 is explained. Based on the oscillation amplitude, the slope upper limit of the station-level power envelope curve (the upper limit of the power change rate) is shrunk. The new slope upper limit is substituted into the next solution to limit the subsequent power change rate. According to the detected power oscillation trend, the power change rate constraint is dynamically adjusted to prioritize suppressing conflicts between internal and external control instructions to ensure the stability of power exchange between the charging station and the upper-level power grid.

[0055] What needs to be explained in the embodiments of the present invention is that the first time domain operation data is collected by sensors deployed on each charging pile, and at least includes instantaneous current, instantaneous voltage, vehicle battery power and temperature inside the pile; the second time domain scheduling data is received through a communication link with the external power grid management system, and at least includes power grid scheduling instructions, real-time electricity price signals, common bus voltage and frequency; the third time domain scheduling data includes environmental parameters affecting renewable energy output (such as irradiance, wind speed, temperature) and energy storage status; the station-level state vector refers to a multidimensional vector constructed based on the fusion filtering results, including the total power of the charging pile, the power distribution of each charging pile, the bus voltage frequency, prediction error, renewable energy output status and energy storage status, which is used as the input of the station-level prediction sequence of the rolling prediction model.

[0056] Explanation: PV output forecasts are generated by fitting short-term trends using historical irradiance data, irradiance sensor data, environmental parameters such as temperature and cloud cover, and the weather API. When API data is unavailable, historical data interpolation or a default model is used. These environmental parameters are spatiotemporally aligned and processed using weighted fusion or machine learning models (such as support vector machines or neural networks) to improve forecast accuracy. For example, temperature data is collected in real time by sensors, and cloud cover data is obtained from the weather API. After fusion, a short-term output forecast of 1-5 minutes is generated.

[0057] It should be explained in the embodiment of the present invention that the rolling prediction model includes a renewable energy output prediction channel, a station-level power prediction channel, and a grid-side voltage prediction channel. The renewable energy output prediction channel is used to predict the renewable energy output status and energy storage status of the next control cycle (the next 1-5 minutes); the station-level power prediction channel is used to predict the total power of the charging pile in the next control cycle; the grid-side voltage prediction channel is used to predict the bus voltage frequency in the next control cycle. The operation process of the rolling prediction model includes:

[0058] State space modeling: using the observed initial station-level state vector as input, linear time-varying state space equations for renewable energy forecast output, station-level forecast power, and bus voltage and frequency are established.

[0059] Explanation: PV output forecasts are generated by fitting short-term trends using historical irradiance data, irradiance sensor data, and the weather API interface. The bus voltage change rate is linearly related to the difference between the grid dispatch command power and the total charging pile power, with the proportionality coefficient determined through regression analysis of historical data. Renewable energy is affected by the integral of charge and discharge power and is constrained by the upper and lower limits of physical capacity.

[0060] Use recursive least squares or extended Kalman filtering to update the state transfer matrix and input gain matrix;

[0061] Multi-step forward deduction, after each control step, expands the identification forward to the prediction time domain and outputs the station-level prediction sequence;

[0062] Confidence interval evaluation, calculates the noise covariance through Bayesian estimation, and generates the upper and lower confidence boundaries corresponding to the renewable energy output forecast channel, station-level power forecast channel, and grid-side voltage forecast channel.

[0063] The explanation is that the rolling prediction model contains three channels of renewable energy / power / voltage, and adopts recursive least squares or extended Kalman filtering to update the matrix; the inner and outer layer controllers dynamically adjust the redundancy and slope upper limit based on the imbalance between piles; when an oscillation trend is detected, the smoothing weight factor is increased and the power damping dead zone boundary is set; the parts of the three channels of renewable energy / power / voltage in the rolling prediction model that are not explained in detail are existing technologies and can be implemented based on conventional technical means. The present invention does not make specific limitations on this.

[0064] It should be explained in the embodiment of the present invention that the inner controller is used to quickly control the power of charging piles in the charging pile group local area network, with a millisecond-level refresh cycle; the power prediction optimization model adopts a rolling optimization method and takes into account vehicle charging requirements and hardware safety constraints;

[0065] Explanation: The power prediction optimization model uses charging pile power as the decision variable, vehicle charging demand and hardware safety constraints as boundaries, and adopts a rolling optimization method based on the station-level prediction sequence of the next control cycle to minimize power fluctuations and imbalances between piles, thereby achieving load balancing of the charging piles. It continuously learns and optimizes the power prediction optimization model parameters based on historical charging data and real-time feedback to improve the accuracy and adaptability of load distribution; the charging pile power sequence is the power setting vector generated by the inner controller for each charging pile within the prediction window, and is sent to the execution layer in a time sequence; the station-level power redundancy is the difference between the total available power upper limit allowed by the charging pile hardware and the total output power after inner layer optimization; the expected station-level power trajectory is the station-level power change curve over time obtained by the inner controller summarizing the power sequences of each charging pile.

[0066] What needs to be explained in the embodiments of the present invention is that the prediction step size of the inner prediction model is equal to the refresh period multiplied by the number of optimization steps, and the number of optimization steps is determined based on the balance between the system computing power and the prediction accuracy requirements; for example, the number of optimization steps is 3-10 steps, which is adjusted according to the system computing power; the hardware safety constraints include that the current of a single pile does not exceed the hardware limit and the temperature inside the pile is lower than the safety threshold.

[0067] In a possible embodiment, the number of optimization steps is obtained as follows:

[0068] When starting a new control cycle, the inner controller collects the total power demand of the charging piles, grid dispatch instructions, renewable energy output, and energy storage status. It synchronizes the data using timestamps and applies a decision tree classifier to classify the operating scenario into high-load, medium-load, or low-load modes. It also identifies frequency deviation anomalies, generates scenario status labels and anomaly feature records, and stores them in a shared state database to support scenario-based computing resource allocation.

[0069] After receiving the scene status label and abnormal feature record, the inner controller measures the CPU usage, available memory, and communication delay, uses the resource allocation model to calculate the maximum number of optimization iterations per second, generates computing power indicators and resource constraint alarms, and stores them in the shared state database to constrain the range of optimization steps;

[0070] After obtaining scene state labels, abnormal feature records, and computing power indicators, the prediction model analyzes historical prediction error data, estimates prediction accuracy under different step numbers through a Bayesian inference model, selects a set of candidate step numbers that meet the accuracy threshold, and stores the candidate step number set and accuracy anomaly flags in a shared state database to support final step number optimization.

[0071] After obtaining the candidate step set and the accuracy anomaly flag, the inner controller tests each step in the set and optimizes the final step number by using an online learning algorithm to minimize the power fluctuation error between the actual power trajectory and the predicted power trajectory. The step number with the smallest error is selected, and the stability is verified by monitoring the convergence of the power trajectory. The final optimized step number and verification results are stored in a shared state database to ensure the stability of power allocation.

[0072] Explanation: The prediction accuracy formula calculates the prediction accuracy under the step number in the following way: After taking the absolute value of the historical prediction error data divided by the total system power input (obtained by adding the total power demand of the charging pile and the output of renewable energy), it is processed by an exponential decay function and multiplied by the scenario error amplification factor determined by historical data regression analysis based on the scenario state label. The result is integrated on the time axis, and the obtained integral value is divided by the integral of the exponential decay function that only contains the scenario error amplification factor on the time axis to generate a prediction accuracy value with a value range of 0 to 1. It is stored in the shared state database to quantify the accuracy of the prediction model and support the selection of candidate step number sets.

[0073] In one possible embodiment, the prediction accuracy is calculated using the following formula:

[0074]

[0075] Among them, E history (t, N) represents the historical prediction error, which is obtained by extracting the difference between the predicted power and the actual power in the past control cycle through the shared state database. The unit is kilowatt and is calculated by the inner controller based on the sensor data and the prediction model output; P total (t) represents the total power input of the system, which is obtained by adding the total power demand of the charging pile and the output of renewable energy after aligning the timestamps. The unit is kilowatt and is collected in real time from the charging pile sensor and the renewable energy sensor; F scenario (L scenario ) represents the scenario error amplification factor, which is obtained by historical data regression analysis based on the scenario state label (high load, medium load, low load or off-grid), has no unit and is stored in the shared state database. N represents the number of optimization steps, which is obtained by iteratively testing the candidate step set through the prediction model to determine the number of steps that meet the accuracy requirements. It has no unit and is dynamically generated by the prediction model. predict (N) represents the prediction accuracy under the number of steps, which is based on the ratio of the historical prediction error to the total system power input after exponential decay and scenario adjustment. It has no unit and the value range is [0, 1]. It is stored in the shared state database.

[0076] For ease of understanding, in an embodiment of the present invention, the scenario error amplification coefficient is generated in the following manner: extracting the operating data of the past 30 days from the shared status database, including scenario status labels (high load, medium load, low load or off-grid), total power demand of charging piles, renewable energy output and frequency deviation records, and constructing a multidimensional data set containing scenario characteristics and prediction errors; using a support vector regression model to train the data set, with the scenario status label as input and the mean of the ratio of prediction error to total power input as output, to generate scenario error amplification coefficients with a value of 1.5 in high load scenarios, a value of 1.0 in medium load scenarios, a value of 0.8 in low load scenarios, and a value of 2.0 in off-grid scenarios, which are stored in the shared status database and updated every 24 hours. The abnormal fluctuation amplification factor is generated in the following way: abnormal feature records including frequency deviation, power mutation and communication delay are extracted from the shared state database to construct an abnormal data set; a random forest regression model is used with the abnormal feature records as input and the peak value of the power fluctuation error as output to generate an abnormal fluctuation amplification factor with a value of 2.5 when the frequency deviation is greater than 0.5 Hz, a value of 2.0 when the power mutation is greater than 10% of the rated power, and a value of 1.5 when the communication delay is greater than 50 milliseconds. The amplification factor is stored in the shared state database and updated every 12 hours.

[0077] Explanation: The power fluctuation error formula calculates the power fluctuation error under the step number in the following way: take the absolute value of the difference between the actual power trajectory and the predicted power trajectory, divide it by the system constraint power (obtained by adding the grid dispatch instruction and the energy storage state multiplied by the energy storage capacity coefficient), take the square and multiply it by the abnormal fluctuation amplification coefficient determined by historical data analysis based on the abnormal feature record, integrate the result on the time axis to generate a power fluctuation error value in kilowatt-square-seconds, and store it in the shared state database to evaluate the stability of power distribution and optimize the final step selection.

[0078] In a possible embodiment, the power fluctuation error is calculated using the following formula:

[0079]

[0080] Among them, P actual (t) represents the actual power trajectory, which is collected in real time by the charging pile sensor and the total power output of the station in kilowatts. It is stored in the shared state database for use by the inner controller; P predict (t, N) represents the predicted power trajectory, which is calculated by the inner prediction model based on the number of steps N and the station-level state vector, in kilowatts, extracted from the shared state database; P constraint (t) represents the system constraint power. After the timestamp synchronization is aligned, the grid dispatch instruction is added to the energy storage state multiplied by the energy storage capacity coefficient. The unit is kilowatt and is collected from the grid communication interface and energy storage monitoring equipment. F anomaly(A anomaly ) represents the abnormal fluctuation amplification factor, which is determined by analyzing historical data based on abnormal feature records (such as frequency deviation). It has no unit and is stored in the shared state database; E fluctuation (N) represents the power fluctuation error, which is normalized based on the square of the difference between the actual and predicted power trajectories, multiplied by the abnormal fluctuation amplification factor, and integrated. The unit is kilowatt-squared second and is stored in the shared state database.

[0081] In order to increase the robustness of the optimization step acquisition, during the optimization step acquisition process, in order to deal with sensor failures and communication interruption anomalies, the inner controller executes the following logic: when it is detected that the difference between the charging pile sensor reading (such as instantaneous current or voltage) and the previous cycle exceeds 20% of the rated value, or the reading is missing for three consecutive cycles, it is determined to be a sensor failure, triggering the backup historical data mode, using the average power demand and renewable energy output of the last 24 hours in the shared state database instead of real-time data, generating a temporary state label and abnormal feature record; when the communication delay exceeds 100 milliseconds or the grid dispatch instruction is not received for two consecutive cycles, it is determined to be a communication interruption, and the inner controller switches to the local prediction mode, based on the most recent valid station-level state vector and prediction sequence, fixes the optimization step number to 3, and suspends the online learning optimization until communication is restored; the above-mentioned abnormal processing results are stored in the shared state database, and after normal operation is restored, an abnormal event report is generated and pushed to the outer controller to adjust the subsequent power envelope curve.

[0082] What needs to be explained in the embodiment of the present invention is that the outer controller is used to perform slow scheduling of the charging pile power, and the refresh period is in seconds (1-10 seconds); when the error between the actual power and the envelope curve alternates between positive and negative for three consecutive cycles and the absolute value increases, it triggers a warning state that stabilizes to an oscillation warning state, and adjusts the control strategy according to the situation.

[0083] Explanation: The slow scheduling of the outer controller includes the following state transition rules:

[0084] When the error between the actual power and the envelope curve alternates between positive and negative for three consecutive cycles and the absolute value increases, the warning state from the stable state to the oscillation warning state is triggered;

[0085] If the warning state lasts for more than two outer control cycles, the smooth weight factor increment strategy is activated, and the outer envelope slope update is frozen, triggering the intervention state from the oscillation warning state to the damping intervention state;

[0086] When the absolute value of the error continuously decreases to a preset range, the slope constraint is gradually released and the weight factor is reset, triggering the recovery state in which the control parameters are restored.

[0087] What needs to be explained in the embodiment of the present invention is that the inner controller feeds back the imbalance between the charging piles after power distribution (the standard deviation of the power of each charging pile) to the outer controller. The imbalance between the charging piles is the standard deviation of the power distribution of each charging pile. The outer controller dynamically adjusts the distribution strategy of the station-level power redundancy based on the imbalance between the charging piles. If the imbalance between the charging piles increases, the outer controller increases the station-level power redundancy and relaxes the upper limit of the slope; if the imbalance between the charging piles decreases, the outer controller reduces the station-level power redundancy and tightens the upper limit of the slope.

[0088] Background: In off-grid microgrid scenarios, charging stations, deprived of the inertia and power support of the main grid, face the severe challenge of frequency instability when charging loads suddenly change. Traditional control strategies fail to dynamically match equivalent inertia (such as the rotational inertia simulated by a virtual synchronous machine) with the power change rate. As a result, when high-power charging demand surges, the microgrid is unable to buffer the power impact due to insufficient inertia, causing frequency drops beyond the limit or even protective tripping. At the same time, the fixed-threshold power ramp limit and the decoupled energy storage response mechanism make it difficult to balance charging efficiency and system stability. To this end, the embodiments of the present invention establish a closed-loop association between power ramp rate limit and system inertia support capability through a collaborative mechanism of equivalent inertia perception, dynamic power envelope slope constraint, and energy storage priority compensation. The equivalent inertia is calculated in real time and the upper limit of the power envelope slope is dynamically adjusted to ensure that the power change rate strictly adapts to the microgrid inertia level. When the frequency deviation exceeds the limit, the energy storage millisecond-level power compensation is preferentially triggered and the slope adjustment is frozen, forming a stable control loop across time scales. Ultimately, while ensuring charging efficiency, the risk of frequency collapse caused by power mutations is avoided.

[0089] It should be explained in the embodiment of the present invention that in the off-grid microgrid scenario, the outer controller establishes a closed-loop association between power ramp rate limit and inertial support capability through equivalent inertia perception, dynamic adjustment of the power envelope slope upper limit and energy storage priority compensation;

[0090] The outer controller performs the following coordinated control steps when the charging station operates in the off-grid microgrid mode:

[0091] Calculate the equivalent moment of inertia: Let n be the number of virtual synchronous generators, and t be the index number; let the output power of the t-th virtual synchronous generator be P gen,t ; The mechanical time constant of the t-th virtual synchronous generator is recorded as T m,t , reflecting the inertial response characteristics, the real-time charging and discharging power of the energy storage device is recorded as P ess ; The virtual inertia time constant of the energy storage device is recorded as T ess , preset through the configuration interface; the rated angular frequency is recorded as ω0; the maximum allowable frequency deviation is recorded as Δf max ; Through the formula The equivalent moment of inertia of the microgrid is calculated and used to characterize the ability of the charging pile system to resist frequency fluctuations;

[0092] Adjust the upper limit of the power envelope slope: according to the equivalent moment of inertia J eq Adaptively adjust the slope upper limit S of the station-level power envelope curve max , to ensure that the power change rate matches the system inertia, the adjustment formula is: Among them S base represents the preset slope upper limit, k represents the slope gain coefficient, which is obtained through historical data training; J min , J max They represent the minimum and maximum allowable equivalent moments of inertia respectively.

[0093] Furthermore, when the dynamic offset of the microgrid frequency exceeds the preset stability threshold and continues to exceed the limit, the transient power compensation mechanism of the energy storage device is triggered and the slope constraint parameter of the power envelope curve is locked. A topology optimization alarm of insufficient inertial support capacity is simultaneously generated, and a virtual synchronous unit expansion configuration recommendation is pushed.

[0094] It is necessary to explain in the embodiments of the present invention that Figure 2 The control parameter update flow chart of FIG. 1 includes the following steps:

[0095] Real-time collection of the actual station-level power after the inner controller executes each control cycle, and synchronous reception of the station-level power envelope curve set by the outer controller;

[0096] For each control cycle, the error sequence trend is determined based on the error between the actual station-level power and the power envelope curve. If the error reverses between two consecutive control cycles and the amplitude increase ratio exceeds the preset value, it is determined to be an oscillation trend.

[0097] When an oscillation trend is detected, the smoothing weight factor of the inner controller is increased to enhance the temporal continuity of power scheduling, a power damping dead zone boundary centered on the current power mean is set, and the slope upper limit of the station-level power envelope curve is dynamically reduced;

[0098] The error is kept within the power damping dead zone boundary, suppressing high-frequency corrections to the inner and outer control commands and preventing high-frequency coupling of the loops; high-frequency coupling refers to the phase resonance phenomenon caused by the bandwidth overlap of the inner and outer control loops.

[0099] Explanation: When the inner controller and the outer controller simultaneously and rapidly correct the power instructions and lack damping, the error sequence undergoes high-frequency reverse flipping and gradually amplifies, which is considered to be a high-frequency coupled oscillation.

[0100] In a possible embodiment, for each control cycle, the trend of the error sequence is determined according to the error between the actual station-level power and the power envelope curve; the station-level power error sequence is processed through fast Fourier transform (FFT) to extract the main frequency component in the range of 0.1 Hz to 1 Hz; if the amplitude of the main frequency component exceeds a preset threshold (for example, 5% of the rated power) and remains consistent in the last two control cycles, it is determined that there is an oscillation trend; according to the detected oscillation frequency, a proportional-integral (PI) controller is used to dynamically adjust the smoothing weight factor to enhance the time continuity of power scheduling; the specific implementation process is as follows:

[0101] In each control cycle, the Fourier transform is performed on the station-level power error sequence recorded in the current cycle to obtain frequency spectrum information;

[0102] If there is a frequency component in the frequency spectrum that is located in a preset frequency interval and has an amplitude exceeding a set threshold, and the condition is met in the last two cycles, it is determined that there is a periodic oscillation; if there is no periodic oscillation, a band-pass filter is used to process the error sequence;

[0103] If high-amplitude oscillation exceeding an energy threshold continuously occurs in a set frequency range, and the duration exceeds a preset time threshold, it is determined that there is an oscillation trend;

[0104] If any of the above conditions is met, it is determined that the oscillation trend is established.

[0105] In a possible embodiment, the inner controller and the outer controller are connected through a local area network to realize data exchange and instruction issuance; the output data of the inner controller and the outer controller are received through a communication network, and the updated control parameters are written into a shared state database; the state database is a shared storage device, which is accessed by the inner controller and the outer controller, and is used to store and update control parameters and prediction sequences.

[0106] Summary: The embodiment 1 of the present application provides a charging pile system management method based on dynamic load balancing, which solves the power jump and steady-state safety problems caused by the multi-time scale coupling between the charging station and the power grid through a hierarchical cooperative control architecture (inner millisecond-level fast power distribution + outer second-level power grid cooperative scheduling) and a multi-time domain prediction fusion mechanism; the specific implementation includes:

[0107] The real-time data of the charging pile (current / voltage / temperature), the power grid scheduling instruction (price / bus voltage), and the renewable energy output parameter (irradiance / energy storage SOC) are integrated to construct a multi-dimensional state vector (such as total power distribution, prediction error, etc.), and the sensor noise is eliminated through Kalman filtering;

[0108] A three-channel prediction model (renewable energy output, charging pile power, and bus voltage) is used to generate a power / voltage forecast series for the next 1-5 minutes based on a linear time-varying state-space equation (coefficients are updated via recursive least squares), and a Bayesian confidence bound (±5% error band) is calculated.

[0109] Based on the oscillation trend detection (error alternating between positive and negative with increasing amplitude), the slope upper limit contraction and smoothing weight factor increase are triggered, and the power damping dead zone is used to suppress the conflict between the inner and outer control commands;

[0110] In microgrid mode, the upper limit of the power envelope slope is dynamically adjusted based on the equivalent moment of inertia formula to ensure that the power ramp rate matches the system inertia.

[0111] Compared with the traditional fixed-step method, the optimized step acquisition scheme of this method significantly improved the prediction accuracy and power distribution stability in actual tests. In a 30-day charging pile system operation test, the average prediction error of the fixed-step method (the step number is set to 5) was 8.2%, and the average power fluctuation error was 12.5 kilowatt-seconds squared. However, through dynamic scene classification and online learning optimization, this method reduced the average prediction error to 4.7% and the average power fluctuation error to 7.3 kilowatt-seconds squared, improving the performance by 42.7% and 41.6% respectively. In the off-grid microgrid scenario, under abnormal conditions where the frequency deviation is greater than 0.5 Hz, the power distribution stabilization time of this method is shortened from 1.2 seconds of the fixed-step method to 0.6 seconds, improving the response speed by 50%.

[0112] Example 2. The difference between the embodiment of the present invention and Example 1 is that the power prediction optimization model includes a capacity drift self-calibration compensation step: continuously collecting energy storage current, terminal voltage, temperature and internal resistance within a sliding time window, and suppressing noise on the energy storage current and terminal voltage data through ampere-hour integration combined with Kalman filtering to obtain the cumulative charge and discharge amount, thereby ensuring the accuracy of capacity estimation; the cumulative charge and discharge amount and the nominal capacity are input into the recursive least squares algorithm to estimate the effective capacity online, the battery health status is evaluated based on the increase in internal resistance and cycle life loss, and the future capacity drift and health degradation trend of the battery are predicted based on historical data; the difference between the nominal capacity and the effective capacity is used to represent the capacity drift The system uses the assessed battery health status and the predicted future capacity drift and health degradation trends to dynamically adjust the control strategy, such as limiting the maximum charging current to slow down aging or redistributing energy storage tasks among battery modules to balance health status. The system also incorporates the battery health status into the dynamic load balancing algorithm, preferentially assigning high-load tasks to charging piles with better health status, and adjusting the charge and discharge power of the energy storage and energy conversion device according to the real-time capacity and health status of the battery.

[0113] In a possible embodiment, after the oscillation event has been recorded, when similar oscillations are subsequently detected, the parameters are adjusted according to the historical data. The specific implementation process is as follows:

[0114] After each oscillation is detected, the characteristic information corresponding to the oscillation is recorded, including the oscillation frequency at that time, the number of cycles, the smoothing weight factor used, and the power damping dead zone boundary value;

[0115] Constructing the above oscillation events into historical records to form an oscillation event database;

[0116] When a new oscillation is detected in a subsequent control cycle, a similarity comparison is performed with the records in the historical database based on the current oscillation frequency and duration;

[0117] Select the parameter value corresponding to the record in the historical records that is most similar to the current oscillation characteristics as the control parameter setting for the current cycle;

[0118] The smoothing weight factor and the power damping dead zone boundary value are respectively updated to the corresponding optimal setting values ​​in the historical record.

[0119] In one possible embodiment, before detecting the oscillation trend, an abnormality identification and classification response process is performed to avoid the adaptive parameter adjustment operation. The specific implementation process is as follows:

[0120] During the current control cycle, the residual values ​​between the model prediction results and the actual power measurement values, the deviation values ​​between the bus voltage and the target voltage, and the difference between the consecutive cycle readings of key sensors are obtained respectively. If the total residual value exceeds the preset residual judgment threshold, it is considered that the model prediction is abnormal.

[0121] If the bus voltage deviation exceeds the preset voltage stability threshold, it is considered that there is an external disturbance; if the difference between the measurement value of the key sensor and the value of the previous cycle exceeds the set mutation amplitude threshold, it is considered that the sensor is faulty;

[0122] When any of the above three items meets the abnormal judgment conditions, the adaptive parameter adjustment operation of the current cycle is skipped, an alarm signal is issued, and the backup control strategy is switched to ensure stable operation;

[0123] When none of the above three criteria are triggered, the parameter adaptive adjustment process is executed.

[0124] In one possible embodiment, the slope upper limit is dynamically adjusted based on the voltage harmonic distortion rate monitored in real time. When the harmonic distortion rate is detected to exceed the standard limit, the slope upper limit is automatically tightened. The power error spectrum is analyzed. If the main frequency exceeds the bandwidth of the outer controller, it is determined that there is an oscillation trend caused by the overlap of the inner and outer control bandwidths. An emergency power envelope curve is generated, and the energy storage compensation instruction is preferentially called. The operation process of the energy storage compensation instruction includes:

[0125] Compare the predicted power at the station level with the actual power at the station level in each control cycle; if the absolute value of the power gap exceeds the confidence interval, the compensation process is triggered;

[0126] The difference between the predicted station-level power and the actual station-level power represents the power gap. If the power gap is positive, the energy storage conversion device is controlled to discharge immediately to fill the gap. If the power gap is negative, the energy storage conversion device is instructed to absorb the excess power.

[0127] Real-time analysis predicts the pattern and frequency of power shortfalls and uses fault diagnosis models to determine whether there are sensor failures or abnormal events (such as several charging piles suddenly going offline). If an anomaly is detected, the compensation strategy is automatically adjusted to prioritize critical loads or take other pre-set measures. The predicted power shortfall and the real-time power shortfall of the charging piles are input, and the output is the fault diagnosis result and the adjusted compensation strategy. A support vector machine is used to analyze the data to identify sensor or charging pile anomalies and adjust instructions according to the pre-set strategy.

[0128] Increase the weight of the latest measurement data in the state space model or switch to an alternative prediction model to quickly correct subsequent prediction results;

[0129] The error correction amount and the updated model parameters are written into the state database for the next round of power prediction optimization model call.

[0130] In one possible embodiment, dynamic mode switching logic is used to address the problem of a single control strategy being unable to adapt to multiple scenarios. Based on the urgency of grid dispatch instructions (such as peak shaving requirements and harmonic suppression instructions) and the real-time load rate of the charging pile group, the collaborative mode of the inner and outer controllers is dynamically selected, including the following modes:

[0131] Mode A (steady-state coordination): The outer controller takes the lead in generating the station-level power envelope curve, and the inner controller performs conventional rolling optimization allocation;

[0132] Mode B (Emergency Response): When it is detected that the grid instruction contains harmonic suppression requirements, it switches to the inner and outer layer joint calculation mode. The inner layer controller feeds back the power allocation results to the outer layer envelope slope constraint in real time.

[0133] Mode C (failure fallback): When it is detected that the communication delay exceeds the threshold, the inner layer local cache data is enabled to generate a power sequence, and the outer layer controller only monitors without intervening.

[0134] Summary: Based on Example 1, this embodiment adds a capacity drift self-calibration compensation and multi-mode oscillation suppression mechanism, focusing on solving the power distribution deviation and control loop resonance problems caused by battery aging.

[0135] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A charging pile system management method based on dynamic load balancing, wherein the charging pile system is deployed in parallel with a distributed renewable energy output device, characterized in that: The following steps are involved: Step 1: Multi-time domain state acquisition and fusion prediction, including: collecting the first time domain operation data of the charging pile side, the second time domain scheduling data of the grid side, and the third time domain scheduling data of the distributed renewable energy output device side, fusing and filtering to obtain the station-level state vector, and generating the station-level prediction sequence of the charging pile in the next control cycle based on the station-level state vector and the rolling prediction model; Step 2: Inner prediction and balancing control, including: the inner controller reads the station-level state vector and station-level prediction sequence, establishes and solves the power prediction optimization model, and outputs the expected station-level power trajectory and station-level power margin; Step 3: Generate the outer grid collaborative envelope, including: receiving the output of the inner controller, comprehensively solving the station-level power envelope curve and the energy storage charge and discharge setpoints, and transmitting them; the slope upper limit of the station-level power envelope curve is used to prevent the outer-layer scheduling from generating excessively fast power steps; Step 4: Real-time collection of the actual station-level power after the inner controller is executed in each control cycle, and synchronous reception of the station-level power envelope curve set by the outer controller; update control parameters including adjustment of the smoothing weight factor, power damping dead zone boundary, and slope upper limit, and synchronously write them into the state database; shrink the slope upper limit of the station-level power envelope curve based on the oscillation amplitude, and substitute the new slope upper limit for the next solution; the smoothing weight factor refers to: the continuity penalty coefficient of the power smoothing penalty term in the objective function of the power prediction optimization model, which is used to suppress the power command jump of adjacent control steps, and the power damping dead zone boundary refers to the power tolerance band set with the current total power mean as the center to prevent high-frequency micro-amplitude commands from superimposing each other.

2. A charging pile system management method based on dynamic load balancing according to claim 1, characterized in that: The first time-domain operating data is collected by sensors deployed at each charging pile, and includes at least instantaneous current, instantaneous voltage, vehicle battery charge, and temperature inside the pile; the second time-domain scheduling data is received via a communication link with an external power grid management system, and includes at least power grid scheduling instructions, real-time electricity price signals, and common bus voltage and frequency; the third time-domain scheduling data includes environmental parameters and energy storage status that affect renewable energy output; The station-level state vector refers to a multidimensional vector constructed based on the fusion filtering results, including the total power of the charging pile, the power distribution of each charging pile, the bus voltage frequency, the prediction error, the renewable energy output status and the energy storage status, which is used as the input of the station-level prediction sequence of the rolling prediction model.

3. A charging pile system management method based on dynamic load balancing according to claim 2, characterized in that: The rolling prediction model includes a renewable energy output prediction channel, a station-level power prediction channel, and a grid-side voltage prediction channel. The renewable energy output prediction channel is used to predict the renewable energy output status and energy storage status of the next control cycle; the station-level power prediction channel is used to predict the total power of the charging piles in the next control cycle; The grid-side voltage prediction channel is used to predict the bus voltage and frequency of the next control cycle. The operation process of the rolling prediction model includes: State-space modeling uses the observed initial station-level state vector as input to establish linear time-varying state-space equations for renewable energy output forecast, station-level power forecast, and bus voltage frequency. PV output forecasts are generated by fitting short-term trends using historical irradiance data, irradiance sensor data, and the weather API interface. The bus voltage change rate is linearly related to the difference between the grid dispatch command power and the total charging pile power, with the proportionality coefficient determined through historical data regression analysis. Renewable energy is affected by the integral of charge and discharge power and is constrained by the upper and lower limits of physical capacity. Use recursive least squares or extended Kalman filtering to update the state transfer matrix and input gain matrix; Multi-step forward deduction, after each control step, expands the identification forward to the prediction time domain and outputs the station-level prediction sequence; Confidence interval evaluation, calculates the noise covariance through Bayesian estimation, and generates the upper and lower confidence boundaries corresponding to the renewable energy output forecast channel, station-level power forecast channel, and grid-side voltage forecast channel.

4. A charging pile system management method based on dynamic load balancing according to claim 1, characterized in that: The inner controller is used to quickly control the power of charging piles in the local area network of the charging pile group. The power prediction optimization model uses the charging pile power as the decision variable and the vehicle charging demand and hardware safety constraints as the boundaries. It adopts a rolling optimization method based on the station-level prediction sequence of the next control cycle to minimize power fluctuations and imbalances between piles, thereby achieving load balancing of the charging piles. It continuously learns and optimizes the power prediction optimization model parameters based on historical charging data and real-time feedback; the charging pile power sequence is the power setting vector generated by the inner controller for each charging pile within the prediction window, which is sent to the execution layer in a time sequence; the station-level power redundancy is the difference between the total available power upper limit allowed by the charging pile hardware and the total output power after inner-layer optimization; the expected station-level power trajectory is the station-level power change curve over time obtained by the inner controller summarizing the power sequences of each charging pile.

5. A charging pile system management method based on dynamic load balancing according to claim 4, characterized in that: The prediction step length of the inner controller is equal to the refresh period multiplied by the number of optimization steps; the hardware safety constraints include that the current of a single pile does not exceed the hardware limit and the temperature inside the pile is lower than the safety threshold.

6. A charging pile system management method based on dynamic load balancing according to claim 4, characterized in that: The outer controller is used to perform slow scheduling of charging pile power, with a refresh cycle of seconds; when the error between the actual power and the envelope curve alternates between positive and negative for three consecutive cycles and the absolute value increases, an oscillation warning state is triggered.

7. A charging pile system management method based on dynamic load balancing according to claim 1, characterized in that: The inner controller feeds back the imbalance degree between charging piles after power distribution to the outer controller. The imbalance degree between charging piles is the standard deviation of the power distribution of each charging pile. The outer controller dynamically adjusts the allocation strategy of the station-level power redundancy based on the imbalance degree between charging piles. If the imbalance degree between charging piles increases, the station-level power redundancy is increased and the upper limit of the slope is relaxed. If the imbalance between piles decreases, reduce the station-level power redundancy and tighten the upper limit of the slope.

8. The charging pile system management method based on dynamic load balancing according to claim 1, characterized in that: The outer controller performs the following coordinated control steps when the charging station operates in the off-grid microgrid mode: Calculate the equivalent moment of inertia: Let n be the number of virtual synchronous generators, and t be the index number; let the output power of the t-th virtual synchronous generator be ; The mechanical time constant of the t-th virtual synchronous generator is recorded as , reflecting the inertial response characteristics, the real-time charging and discharging power of the energy storage device is recorded as ; The virtual inertia time constant of the energy storage device is recorded as , preset through the configuration interface; record the rated angular frequency as ; The maximum frequency deviation allowed is recorded as ; Through the formula The equivalent moment of inertia of the microgrid is calculated and used to characterize the ability of the charging pile system to resist frequency fluctuations; Adjust the upper limit of the power envelope slope: according to the equivalent moment of inertia Adaptively adjust the slope upper limit of the station-level power envelope curve , to ensure that the power change rate matches the system inertia, the adjustment formula is: ;in Indicates the preset slope upper limit, Represents the slope gain coefficient, which is obtained through historical data training; , They represent the minimum and maximum allowable equivalent moments of inertia respectively.

9. A charging pile system management method based on dynamic load balancing according to claim 8, characterized in that: The specific implementation process of step 4 includes the following steps: For each control cycle, the error sequence trend is determined based on the error between the actual station-level power and the power envelope curve. If the error reverses between two consecutive control cycles and the amplitude increase ratio exceeds the preset value, it is determined to be an oscillation trend. When an oscillation trend is detected, the smoothing weight factor of the inner controller is increased to enhance the temporal continuity of power scheduling, a power damping dead zone boundary centered on the current power mean is set, and the slope upper limit of the station-level power envelope curve is dynamically reduced; The error is kept within the power damping dead zone boundary, suppressing high-frequency corrections to the inner and outer control commands and preventing high-frequency coupling of the loops; high-frequency coupling refers to the phase resonance phenomenon caused by the bandwidth overlap of the inner and outer control loops.

10. A charging pile system management method based on dynamic load balancing according to claim 9, characterized in that: The power prediction optimization model includes a capacity drift self-calibration compensation step: continuously collecting energy storage current, terminal voltage, temperature and internal resistance within a sliding time window, and suppressing noise on energy storage current and terminal voltage data through ampere-hour integration combined with Kalman filtering to obtain cumulative charge and discharge capacity; The cumulative charge and discharge volume and the nominal capacity are input into the recursive least squares algorithm to estimate the effective capacity online. The battery health status is evaluated based on the increase in internal resistance and cycle life loss. The battery's future capacity drift and health degradation trend are predicted based on historical data. The capacity drift is represented by the difference between the nominal capacity and the effective capacity, and the state and power mapping function is updated according to the capacity drift; The updated power mapping function is used to regenerate the power difference correction value, and the charging and discharging power of the energy storage energy conversion device is adjusted according to the power difference correction value to complete the compensation; Dynamically adjust control strategies based on the assessed battery health status and predicted future capacity drift and health degradation trends; The battery health status is integrated into the dynamic load balancing algorithm, high-load tasks are assigned preferentially to charging piles with better health status, and the charging and discharging power of the energy storage and energy conversion device is adjusted according to the real-time capacity and health status of the battery.

Images