Charging pile multi-module dynamic scheduling method and system

Abstract

The invention discloses a charging pile multi-module dynamic scheduling method and system, and belongs to the technical field of electric vehicle charging facilities, and the method comprises the following steps: S1, collecting a vehicle battery state and a module state in real time; s2, predicting a vehicle demand power curve based on an LSTM model, and calculating derated power according to the module temperature; s3, constructing a multi-objective optimization function; s4, solving based on an improved genetic algorithm, wherein chromosome coding is an N * M-dimensional module-vehicle matching matrix, and a health degree loss penalty term is introduced into fitness calculation; s5, performing double-time-scale execution: generating a module-vehicle matching power base line in a long period; in a short period, the PI controller performs fine tuning output; and setting a load transfer factor to trigger load transfer. According to the invention, the problems of load balancing, life optimization, real-time response and user experience collaboration in multi-module dynamic allocation can be solved, and efficient and reliable operation of the charging pile is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle charging facilities, and in particular to a method and system for dynamic scheduling of multiple modules of a charging pile. Background Art

[0002] With the rapid growth of electric vehicle ownership, the development of charging piles, as a key supporting infrastructure, is urgent. In the early days, charging piles were scarce, scattered, and had inconsistent charging standards. Today, driven by policy support and market forces, the scale of charging piles has expanded significantly. Public charging piles are widely installed in urban public areas and highway service areas, and private charging piles are also increasing with the popularity of electric vehicles in households. However, the application of high-power charging piles still faces some challenges, such as:

[0003] Load imbalance: Traditional round-robin scheduling causes some modules to have low utilization efficiency, resulting in low average utilization.

[0004] Response lag: Some centralized products have long scheduling cycles (≥1 minute) and cannot adapt to sudden changes in vehicle SOC or module failures;

[0005] Serious life loss: Frequent power switching causes the average daily loss rate of internal power modules to reach 1.2%.

[0006] The following related patents exist in the prior art, such as CN118572715A, which addresses the aforementioned severe lifespan loss issue without considering hardware lifespan optimization. CN111682528A only optimizes grid economics while ignoring module-level efficiency balancing. CN119250413A utilizes the NSGA-II algorithm but lacks a real-time correction mechanism, resulting in significant power fluctuations. CN120088091A supports V2G scheduling but lacks a module health protection strategy. Summary of the Invention

[0007] The technical task of the present invention is to address the above shortcomings and provide a method and system for dynamic scheduling of multiple modules of a charging stack, which can solve the problems of load balancing, life optimization, real-time response and user experience coordination in dynamic allocation of multiple modules, and realize efficient and reliable operation of the charging stack.

[0008] The technical solution adopted by the present invention to solve its technical problem is:

[0009] A method for dynamic scheduling of multiple modules of a charging stack, the implementation of which includes the following steps:

[0010] S1: Real-time collection of vehicle battery status (SOC, temperature) and module status (health H i , temperature T i );

[0011] S2: Predict vehicle demand power curve P based on LSTM model i (t), and according to the module temperature T j Calculate the derating power:

[0012]

[0013] Among them, P rated is the module rated power (unit: kW), P i,max is the maximum allowable output power of module i (in kW);

[0014] S3: Construct a multi-objective optimization function:

[0015]

[0016] Among them, t avg is the average charging time of the current allocation scheme (in minutes); η avg is the average working efficiency of all modules (percentage %); ΔP grid is the total power change rate of the power grid (unit: kW / s), and the calculation formula is ||P t -P t-1 ||;P grid,max is the maximum input power allowed by the grid (unit: kW); L i is the instantaneous loss value of the module (dimensionless), N is the total number of available modules (integer); α, β, γ, λ are weight factors;

[0017] S4: Solved based on improved genetic algorithm: chromosome encoding is an N×M-dimensional module-vehicle matching matrix, and fitness calculation introduces a health loss penalty term;

[0018] S5: Dual time scale execution:

[0019] Long cycle (10 minutes), generating module-vehicle matching power baseline

[0020] Short cycle (10 seconds), PI controller fine-tuning P adj =K p e+K i ∫edt fine-tunes the output, where e = SOC target -SOC real , SOC is the battery state of charge (unit %), when the load transfer factor Load transfer is triggered when

[0021] Among them, T max The maximum allowable temperature of the module (unit: °C, default: 85 °C), t run,i is the continuous operation time of module i (unit: hour), t max,runThe maximum continuous operation time threshold of the module (unit: hour, default 8 hours), T i 、T opt They are respectively the real-time temperature of module i and the optimal operating temperature (such as the default 45°C).

[0022] This method establishes a module life quantification scheduling model, incorporating hardware life into the optimization target; designs a dual-time-scale closed-loop control mechanism, which reduces user waiting time, improves module utilization, reduces module health loss, and extends equipment service life through health-driven load balancing, behavior-aware priority scheduling, and dual-time-scale control.

[0023] Furthermore, the health loss model is:

[0024]

[0025] Among them, P i is the real-time output power of the module (unit: kW); ||ΔP i || is the absolute value of the module's power change (unit: kW), and the calculation formula is ||P i -P i-1 ||;T opt is the optimal operating temperature of the module (the default value is 45°C); k1, k2, k3 are model coefficients; health H i Modules with a power fluctuation of < 0.8 will only be allocated to loads with a power fluctuation of ≤ 5%.

[0026] Furthermore, health Where H0 is the initial health (new module = 1), and m is the aging coefficient (default 0.001, calibrated by accelerated life testing).

[0027] Furthermore, the coefficients of the health loss model are calibrated as follows:

[0028] k1 Power square term influence factor, fitted by fixed power aging experiment, such as 0.002 / kW 2 ;

[0029] k2 power fluctuation sensitivity coefficient, through step power change (±30kW) temperature rise test, such as the value of 0.05 / kW;

[0030] k3 temperature offset penalty coefficient, heating to T by constant power max Loss monitoring, such as taking the value as 0.001 / ℃ 2 .

[0031] Furthermore, the penalty function of the genetic algorithm is designed as:

[0032] When the number of start and stop times of a single module exceeds the limit (>50 times / day), the fitness is deducted by ΔF=0.1×(N start-stop-50) 2 .

[0033] Furthermore, fault tolerance includes:

[0034] After detecting module failure, ramp load transfer is started within 100ms, meeting the voltage fluctuation ≤ 2%.

[0035] Furthermore, the dynamic adjustment rules of weight factors α, β, γ, and λ are as follows:

[0036] Peak period: α = 0.7, β = 0.2, γ = 0.1, λ = 0.3;

[0037] During the power grid fluctuation period: α = 0.3, β = 0.2, γ = 0.8, λ = 0.2;

[0038] Valley period: β=0.4, β=0.3, γ=0.1, λ=0.5.

[0039] The present invention also claims protection for a charging stack multi-module dynamic scheduling system, comprising:

[0040] Data acquisition module, real-time collection of vehicle battery status (SOC, temperature) and module status (health, temperature);

[0041] The prediction layer predicts the vehicle's required power curve based on the LSTM model and calculates the derated power according to the module temperature;

[0042] The decision layer calculates the health degree based on the multi-objective optimization function and the improved genetic algorithm;

[0043] The execution layer implements dual-time-scale execution and sets the trigger conditions for load transfer;

[0044] The system realizes dynamic scheduling of multiple modules of the charging stack based on the above method.

[0045] The present invention also claims protection for a multi-module dynamic scheduling device for a charging stack, comprising: at least one memory and at least one processor;

[0046] The at least one memory is configured to store a machine-readable program;

[0047] The at least one processor is configured to call the machine-readable program to implement the above method.

[0048] The present invention also claims protection for a computer-readable medium having computer instructions stored thereon, which are capable of implementing the above method when executed by a processor.

[0049] Compared with the prior art, the method and system for dynamic scheduling of multiple modules of a charging stack of the present invention have the following beneficial effects:

[0050] The present invention reduces user waiting time, improves module utilization, reduces module health loss, and extends equipment service life through health-driven load balancing, behavior-aware priority scheduling, and dual-time scale control. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flowchart of a method for dynamic scheduling of multiple modules of a charging stack provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The present invention will be further described below with reference to specific embodiments.

[0053] An embodiment of the present invention provides a method for dynamic scheduling of multiple modules of a charging stack, the implementation of which includes the following steps:

[0054] S1: Real-time collection of vehicle battery status (SOC, temperature) and module status (health H i , temperature T i );

[0055] S2: Predict vehicle demand power curve P based on LSTM model i (t), and according to the module temperature T j Calculate the derating power:

[0056]

[0057] Among them, P rated is the module rated power (unit: kW), P i,max is the maximum allowable output power of module i (in kW);

[0058] S3: Construct a multi-objective optimization function:

[0059]

[0060] Among them, t avg is the average charging time of the current allocation scheme (in minutes); η avg is the average working efficiency of all modules (percentage %); ΔP grid is the total power change rate of the power grid (unit: kW / s), and the calculation formula is ||P t -P t-1 ||;P grid,max is the maximum input power allowed by the grid (unit: kW); L i is the instantaneous loss value of the module (dimensionless), N is the total number of available modules (integer); α, β, γ, λ are weight factors;

[0061] S4: Solved based on improved genetic algorithm: chromosome encoding is an N×M-dimensional module-vehicle matching matrix, and fitness calculation introduces a health loss penalty term;

[0062] S5: Dual time scale execution:

[0063] Long cycle (10 minutes), generating module-vehicle matching power baseline

[0064] Short cycle (10 seconds), PI controller fine-tuning P adj =K p e+K i ∫edt fine-tunes the output, where e = SOC target -SOC real ,SOC is the battery state of charge (unit %), when the load transfer factor Load transfer is triggered when

[0065] Among them, T max The maximum allowable temperature of the module (unit: °C, default: 85 °C), t run,i is the continuous operation time of module i (unit: hour), t max,run The maximum continuous operation time threshold of the module (unit: hour, default 8 hours), T i 、T opt They are respectively the real-time temperature of module i and the optimal operating temperature (such as the default 45°C).

[0066] Among them, the health loss model is:

[0067]

[0068] Among them, P i is the real-time output power of the module (unit: kW); ||ΔP i || is the absolute value of the module's power change (unit: kW), and the calculation formula is ||P i -P i-1 ||;T opt is the optimal operating temperature of the module (the default value is 45°C); k1, k2, k3 are model coefficients; health H i Modules with a power fluctuation of < 0.8 will only be allocated to loads with a power fluctuation of ≤ 5%.

[0069] Health Where H0 is the initial health (new module = 1), and m is the aging coefficient (default 0.001, calibrated by accelerated life testing).

[0070] The health loss model coefficients are calibrated as follows:

[0071] k1 Power square term influence factor, fitted by fixed power aging experiment, such as 0.002 / kW 2 ;

[0072] k2 power fluctuation sensitivity coefficient, through step power change (±30kW) temperature rise test, such as the value of 0.05 / kW;

[0073] k3 temperature offset penalty coefficient, heating to T by constant power max Loss monitoring, such as taking the value as 0.001 / ℃ 2 .

[0074] The penalty function of the genetic algorithm is designed as:

[0075] When the number of start and stop times of a single module exceeds the limit (>50 times / day), the fitness is deducted by ΔF=0.1×(N start-stop -50) 2 .

[0076] Fault tolerance includes:

[0077] After detecting module failure, ramp load transfer is started within 100ms, meeting the voltage fluctuation ≤ 2%.

[0078] The dynamic adjustment rules of weight factors α, β, γ, and λ are as follows:

[0079] Peak hours (9-12 o'clock): α = 0.7, β = 0.2, γ = 0.1, λ = 0.3;

[0080] During the power grid fluctuation period: α = 0.3, β = 0.2, γ = 0.8, λ = 0.2;

[0081] Valley period (0-6 o'clock): α=0.4, β=0.3, γ=0.1, λ=0.5.

[0082] This method establishes a module life quantification scheduling model, incorporating hardware life into the optimization target; designs a dual-time-scale closed-loop control mechanism, which reduces user waiting time, improves module utilization, reduces module health loss, and extends equipment service life through health-driven load balancing, behavior-aware priority scheduling, and dual-time-scale control.

[0083] An embodiment of the present invention further provides a charging stack multi-module dynamic scheduling system, comprising:

[0084] Data acquisition module, real-time collection of vehicle battery status (SOC, temperature) and module status (health, temperature);

[0085] The prediction layer predicts the vehicle's required power curve based on the LSTM model and calculates the derated power according to the module temperature;

[0086] The decision layer calculates the health degree based on the multi-objective optimization function and the improved genetic algorithm;

[0087] The execution layer implements dual-time-scale execution and sets the trigger conditions for load transfer;

[0088] The system implements dynamic scheduling of multiple modules of a charging stack based on the dynamic scheduling method of multiple modules of a charging stack described in the above embodiment.

[0089] 1. Prediction layer:

[0090] Predicting vehicle demand power curve P based on LSTM model i (t), (P j represents the power demand of vehicle j in kW) according to the module temperature T j (Real-time temperature of module i, unit: °C) Predicted derating power:

[0091]

[0092] Among them, P rated is the module rated power (unit: kW), P i,max is the maximum allowable output power of module i (in kW).

[0093] 2. Decision-making level:

[0094] Construct a multi-objective optimization function:

[0095]

[0096] Among them, t avg is the average charging time of the current allocation scheme (in minutes); η avg is the average working efficiency of all modules (percentage %); ΔP grid is the total power change rate of the power grid (unit: kW / s), and the calculation formula is ||P t -P t-1 ||;P grid,max is the maximum input power allowed by the grid (unit: kW); L i is the instantaneous loss value of the module (dimensionless), N is the total number of available modules (integer); α, β, γ, and λ are weight factors.

[0097] The health loss model is:

[0098]

[0099] Among them, P i is the real-time output power of the module (unit: kW); ||ΔP i || is the absolute value of the module's power change (unit: kW), and the calculation formula is ||P i -Pi-1 ||;T opt is the optimal operating temperature of the module (the default value is 45°C); k1, k2, k3 are model coefficients; health H i Modules with a power fluctuation of < 0.8 will only be allocated to loads with a power fluctuation of ≤ 5%.

[0100] Health Where H0 is the initial health (new module = 1), and m is the aging coefficient (default 0.001, calibrated by accelerated life testing).

[0101] 3. Execution layer:

[0102] Dual time scale control:

[0103] Long cycle (10 minutes), generating module-vehicle matching power baseline

[0104] Short cycle (10 seconds), PI controller fine-tuning P adj =K p e+K i ∫edt fine-tunes the output, where e = SOC target -SOC real , (SOC is the battery state of charge, unit is %).

[0105] Load transfer trigger condition: load transfer factor Load transfer is triggered when

[0106] Among them, T max The maximum allowable temperature of the module (unit: °C, default: 85 °C), t run,i is the continuous operation time of module i (unit: hour), t max,run The maximum continuous operation time threshold of the module (unit: hour, default 8 hours), T i 、T opt They are respectively the real-time temperature of module i and the optimal operating temperature (such as the default 45°C).

[0107] The health loss model coefficients are calibrated as follows:

[0108] k1 Power square term influence factor, fitted by fixed power aging experiment, such as 0.002 / kW 2 ;

[0109] k2 power fluctuation sensitivity coefficient, through step power change (±30kW) temperature rise test, such as the value of 0.05 / kW;

[0110] k3 temperature offset penalty coefficient, heating to T by constant power max Loss monitoring, such as taking the value as 0.001 / ℃ 2 .

[0111] The dynamic adjustment rules of weight factors α, β, γ, and λ are shown in Table 1 below:

[0112] Table 1 Dynamic adjustment rules of weight factors

[0113] Time α β γ λ Peak (9-12 o'clock) 0.7 0.2 0.1 0.3 Grid fluctuation period 0.3 0.2 0.8 0.2 Valley period (0-6 o'clock) 0.4 0.3 0.1 0.5 .

[0114] The penalty function of the genetic algorithm is designed as:

[0115] When the number of start and stop times of a single module exceeds the limit (>50 times / day), the fitness is deducted by ΔF=0.1×(N start-stop -50) 2 .

[0116] Fault tolerance includes:

[0117] After detecting module failure, ramp load transfer is started within 100ms, meeting the voltage fluctuation ≤ 2%.

[0118] A specific application example of this system is as follows:

[0119] The application is a high-speed service area charging station. Hardware configuration: 8 modules, single module rated power 60kW, total grid power limit 400kW, temperature sensor accuracy ±1°C.

[0120] The scheduling process is as follows:

[0121] (1) Prediction layer:

[0122] Vehicle A (initial SOC = 30%) requires power for the first 30 minutes as predicted by LSTM: P j (t) = 120kW; module 3 temperature is T3 = 65℃> 60℃, calculate the derating power: P 3,max =0.8×60kW=48kW;

[0123] (2) Decision-making level:

[0124] Calculate the health loss value of module 4:

[0125] L4=0.002×(55) 2 +0.05×|60-55|+0.001×(50-45) 2 =6.8

[0126] (3) Execution layer:

[0127] Long cycle: Modules 1+2 provide 120kW for vehicle A, and module 4 provides 60kW for vehicle B.

[0128] Short-cycle fine-tuning: When the vehicle SOC reaches 80%, the PI controller calculates the adjustment amount:

[0129] P adj =0.3×(90-80)+0.05×∫(90-80)dt=4.5kW

[0130] Reduce module 1 output from 60kW to 55.5kW;

[0131] Fault response: When module 2 overcurrent fault occurs, the load transfer factor Trigger a 10-second load transfer to module 5.

[0132] An embodiment of the present invention further provides a charging stack multi-module dynamic scheduling device, comprising: at least one memory and at least one processor;

[0133] The at least one memory is configured to store a machine-readable program;

[0134] The at least one processor is configured to call the machine-readable program to implement the method for dynamic scheduling of multiple modules of a charging stack as described in the above embodiment.

[0135] An embodiment of the present invention further provides a computer-readable medium having computer instructions stored thereon. When executed by a processor, the computer instructions cause the processor to execute the method for dynamic scheduling of multiple modules of a charging stack as described in the above-mentioned embodiment. Specifically, a system or device equipped with a storage medium can be provided, on which software program code implementing the functions of any of the above-mentioned embodiments is stored, and a computer (or CPU or MPU) of the system or device can be caused to read and execute the program code stored in the storage medium.

[0136] In this case, the program code itself read from the storage medium can realize the function of any one of the above-mentioned embodiments, and thus the program code and the storage medium storing the program code constitute part of the present invention.

[0137] Examples of storage media for providing program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, the program code can be downloaded from a server computer via a communication network.

[0138] In addition, it should be clear that the functions of any of the above embodiments can be achieved not only by executing the program code read by the computer, but also by enabling the operating system operating on the computer to complete part or all of the actual operations based on the instructions of the program code.

[0139] In addition, it can be understood that the program code read from the storage medium is written into the memory provided in the expansion board inserted into the computer or into the memory provided in the expansion unit connected to the computer, and then based on the instructions of the program code, the CPU installed on the expansion board or expansion unit is enabled to perform part or all of the actual operations, thereby realizing the functions of any of the above embodiments.

[0140] The present invention has been shown and described in detail above through the accompanying drawings and preferred embodiments. However, the present invention is not limited to these disclosed embodiments. Based on the above multiple embodiments, those skilled in the art can know that the code review methods in the above different embodiments can be combined to obtain more embodiments of the present invention, and these embodiments are also within the scope of protection of the present invention.

Claims

1. A method for dynamic scheduling of multiple modules of a charging stack, characterized in that: The implementation of this method includes the following steps: S1: Real-time collection of vehicle battery status and module status; S2: Predict vehicle demand power curve P based on LSTM model i (t), and according to the module temperature T j Calculate the derating power: Among them, P rated is the module rated power, P i,max is the maximum allowed output power of module i; S3: Construct a multi-objective optimization function: Among them, t avg is the average charging time of the current allocation scheme; η avg is the average working efficiency of all modules; ΔP grid is the total power change rate of the power grid, and the calculation formula is ||P t -P t-1 ||;P grid,max is the maximum input power allowed by the grid; L i is the instantaneous loss value of the module, N is the total number of available modules; α, β, γ, λ are weight factors; S4: Solved based on improved genetic algorithm: chromosome encoding is an N×M-dimensional module-vehicle matching matrix, and fitness calculation introduces a health loss penalty term; S5: Dual time scale execution: Long cycle, generating module-vehicle matching power baseline Short cycle, PI controller fine-tunes P adj =K p e+K i ∫edt fine-tunes the output, where e = SOC target -SOC real , SOC is the battery state of charge, when the load transfer factor Load transfer is triggered when T max is the maximum allowable temperature of the module, is the continuous running time of module i, t max,run is the maximum continuous operation time threshold of the module, T i 、T opt They are the real-time temperature and optimal working temperature of module i respectively.

2. A charging stack multi-module dynamic scheduling method according to claim 1, characterized in that: The health loss model is: Among them, P i is the real-time output power of the module; ||ΔP i || is the absolute value of the module's power change, and the calculation formula is ||P i -P i-1 ||;T opt is the optimal operating temperature of the module; k1, k2, k3 are model coefficients; health H i Modules with a power fluctuation of < 0.8 will only be allocated to loads with a power fluctuation of ≤ 5%.

3. A charging stack multi-module dynamic scheduling method according to claim 2, characterized in that: Health Where H0 is the initial health and m is the aging coefficient.

4. A method for dynamic scheduling of multiple modules of a charging stack according to claim 2, characterized in that: The health loss model coefficients are calibrated as follows: k1 power square term influence factor, fitted by fixed power aging experiment; k2 power fluctuation sensitivity coefficient, through step power change temperature rise test; k3 temperature offset penalty coefficient, heating to T by constant power max Loss monitoring.

5. A method for dynamic scheduling of multiple modules of a charging stack according to claim 1 or 2, characterized in that: The penalty function of the genetic algorithm is designed as: When the number of start and stop times of a single module exceeds the limit, the fitness is deducted by ΔF=0.1×(N start-stop -50) 2 .

6. A method for dynamic scheduling of multiple modules of a charging stack according to claim 1, characterized in that: Fault tolerance includes: After detecting module failure, ramp load transfer is started within 100ms, meeting the voltage fluctuation ≤ 2%.

7. A method for dynamic scheduling of multiple modules of a charging stack according to claim 1, characterized in that: The dynamic adjustment rules of weight factors α, β, γ, and λ are as follows: Peak period: α = 0.7, β = 0.2, γ = 0.1, λ = 0.3; During the power grid fluctuation period: α = 0.3, β = 0.2, γ = 0.8, λ = 0.2; Valley period: α=0.4, β=0.3, γ=0.1, λ=0.

5.

8. A multi-module dynamic scheduling system for charging piles, characterized in that: include: Data acquisition module, real-time collection of vehicle battery status and module status; The prediction layer predicts the vehicle's required power curve based on the LSTM model and calculates the derated power according to the module temperature; The decision layer calculates the health degree based on the multi-objective optimization function and the improved genetic algorithm; The execution layer implements dual-time-scale execution and sets the trigger conditions for load transfer; The system realizes dynamic scheduling of multiple modules of a charging stack based on the methods described in claims 1-7.

9. A multi-module dynamic scheduling device for a charging stack, characterized in that: include: at least one memory and at least one processor; The at least one memory is configured to store a machine-readable program; The at least one processor is configured to call the machine-readable program to implement the method according to any one of claims 1 to 7.

10. A computer-readable medium, characterized in that The computer readable medium stores computer instructions, which, when executed by a processor, can implement the method according to any one of claims 1 to 7.

Images