Ultra-fast charging facility flexible regulation and control system for active balance of power distribution network load

Abstract

The invention provides a power distribution network load active balance-oriented ultrafast charging facility flexible regulation and control system, and the system comprises the steps: synchronously obtaining power distribution network operation data and ultrafast charging facility parameters through a multi-source data collection module, and transmitting the data and parameters to a dynamic load prediction module after preprocessing; the module adjusts a prediction period according to a load fluctuation coefficient, outputs a load prediction sequence in combination with an LSTM model, and provides a basis for the flexible regulation and control decision module. The flexible regulation and control decision module constructs a multi-objective optimization function, optimal total regulation and control power is solved according to power distribution network and facility constraints, and the pulse power distribution module distributes power according to SOC and generates a pulse parameter instruction according to the optimal total regulation and control power. The real-time execution control module receives an instruction, generates a PWM signal and controls a converter to output pulse power, and the closed-loop feedback optimization module collects actual data and prediction data, corrects model parameters or weight coefficients after calculating errors, and realizes load balance of the power distribution network and flexible regulation and control of the ultrafast charging facility.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power distribution network regulation, and specifically relates to a super-fast charging facility flexible regulation system for active load balance of a power distribution network. BACKGROUND

[0002] In the prior art, the power distribution network regulation system is mainly used for monitoring the operating state of the power distribution network node voltage, line power and the like, and maintaining the stable operation of the power distribution network through load distribution and power regulation; the super-fast charging facility related control technology focuses on improving the charging efficiency to meet the demand of electric vehicles for rapid energy replenishment, both of which jointly support the basic operation of the electric vehicle charging network and provide technical support for the electrification transformation of the transportation field and the efficient use of energy, thereby helping to reduce the environmental pressure caused by traditional fuel consumption.

[0003] Currently, the power distribution network regulation and the super-fast charging facility control technology have the problem of insufficient coordination. The power distribution network regulation mostly uses fixed cycle load prediction, which is difficult to adapt to the rapid load fluctuation after the super-fast charging facility is connected; the super-fast charging facility charging control mostly takes the charging efficiency of the facility itself as the primary target and does not fully consider the load balance demand of the power distribution network, which is easy to cause the overload of the power distribution network line power or the excessive fluctuation of the load; at the same time, the existing regulation system lacks a real-time closed-loop feedback mechanism, the prediction error cannot be corrected in time, and the regulation accuracy is difficult to meet the operation requirements of the power distribution network after the large-scale connection of the super-fast charging facility. SUMMARY

[0004] To solve the problems in the background art, the application provides a super-fast charging facility flexible regulation system for active load balance of a power distribution network to solve the problem of load imbalance of the power distribution network under the impact of the super-fast charging facility.

[0005] To achieve the above-mentioned purpose, the application provides the following technical scheme: a super-fast charging facility flexible regulation system for active load balance of a power distribution network, characterized in that it comprises a multi-source data acquisition module, a dynamic load prediction module, a flexible regulation decision module, a pulse power distribution module, a real-time execution control module and a closed-loop feedback optimization module, wherein:

[0006] The multi-source data acquisition module is used for synchronously acquiring the power distribution network operating state data and the super-fast charging facility parameter data, filtering and standardizing the collected data, generating a standardized data matrix and transmitting the standardized data matrix to the dynamic load prediction module;

[0007] The dynamic load prediction module generates a power distribution network load prediction sequence in a future specific period based on the standardized data matrix output by the multi-source data acquisition module, adopts an adaptive cycle prediction algorithm and a long short-term memory (LSTM) model, and transmits the power distribution network load prediction sequence to the flexible regulation decision module;

[0008] Flexible regulation decision module: receives the load prediction sequence of the dynamic load prediction module, constructs a multi-objective optimization objective function, combines the constraint conditions of the power distribution network and the super-fast charging facility, solves the optimal total regulation power through an optimization algorithm, and outputs to the pulse power distribution module;

[0009] Pulse power distribution module: according to the optimal total regulation power output by the flexible regulation decision module, combining the state of charge (SOC) of each super-fast charging facility, using the SOC-load collaborative pulse scheduling algorithm, generating the pulse parameter instruction set of each super-fast charging facility, and transmitting to the real-time execution control module;

[0010] Real-time execution control module: receives the pulse parameter instruction set of the pulse power distribution module, generates a pulse width modulation (PWM) signal, controls the power electronic converter to realize the accurate output of pulse current / voltage, and monitors the power deviation in real time and dynamically adjusts;

[0011] Closed-loop feedback optimization module: collects the actual operation data of the real-time execution control module and the prediction data of the dynamic load prediction module, calculates the load prediction error, and modifies the model parameters of the dynamic load prediction module or the weight coefficients of the flexible regulation decision module according to the error size, to improve the regulation accuracy of the system.

[0012] Optionally, the multi-source data acquisition module includes: a power distribution network data acquisition subunit and a facility parameter acquisition subunit;

[0013] The power distribution network data acquisition subunit: used for collecting the node voltage, line active power, reactive power and load fluctuation of the power distribution network, the sampling frequency is set to 50Hz-2kHz, and the collected time series measurement data is preliminarily screened to eliminate obvious outliers;

[0014] The facility parameter acquisition subunit: used for collecting the current state of charge (SOC), battery temperature, rated charging power and pulse charging period of each super-fast charging facility, and synchronously receiving the operation state feedback signal of each facility, to ensure the real-time and integrity of parameter acquisition.

[0015] Optionally, the dynamic load prediction module includes: a prediction period adjustment subunit and a load sequence prediction subunit;

[0016] The prediction period adjustment subunit: calculates the fluctuation coefficient through the power distribution network load fluctuation: ;

[0017] Wherein, is the load fluctuation at time t, is the load fluctuation at time t-1; and according to the prediction period is set to , ≤5% =15 minutes, 5%< ≤15% =5 minutes >15% =1 minute;

[0018] The load sequence prediction subunit employs a Long Short-Term Memory (LSTM) network model. It takes standardized historical load data from the multi-source data acquisition module and the current SOC matrix as input, and through model training and iterative calculation, outputs the future load sequence. Load forecast sequence within the time period to .

[0019] Optionally, the flexible control decision module further includes: an optimization target construction subunit and an optimal power solution subunit;

[0020] The optimization objective construction sub-unit: Constructing a multi-objective optimization function: ;

[0021] in, Costs associated with power distribution network load fluctuations (t=1 to ); (i is the ultra-fast charging facility number), and a dynamic weighting coefficient is set simultaneously: ;

[0022] ;

[0023] in, For the average SOC, and + =1;

[0024] The optimal power calculation subunit, under the constraints of the upper limit of the distribution network line capacity, the maximum power of the ultra-fast charging facility, and the battery temperature threshold, uses a particle swarm optimization algorithm to solve the objective function and iteratively calculates the optimal total control power. The output is sent to the pulse power distribution module.

[0025] Optionally, the pulse power distribution module further includes: a power distribution calculation subunit and a pulse parameter generation subunit;

[0026] The power allocation calculation subunit: based on the optimal total control power output by the flexible control decision module. Combined with the state of charge of various ultra-fast charging facilities Calculate the allocation coefficient:

[0027] ;

[0028] This allows us to obtain the allocated power for each facility:

[0029] ;

[0030] The pulse parameter generation subunit: based on the allocated power and Adjust pulse parameters in stages. When ≤30%, duty cycle D=0.8; when 30% < ≤80%, D=0.5+0.3×( -30%) / 50%, >80%, D=0.5 and amplitude I linearly decreases to 0.5I rated, generating an instruction set including pulse period, high-level duration ton, low-level duration toff, and amplitude adjustment curve, ensuring that the high-level pulse periods of each facility do not overlap.

[0031] Optionally, the real-time execution control module further includes: a control signal generation subunit and a power output regulation subunit;

[0032] The control signal generation subunit receives the pulse parameter instruction set from the pulse power distribution module, performs signal conversion and processing through a digital signal processor (DSP), generates a pulse width modulation (PWM) control signal, and transmits it to the power electronic converter.

[0033] The power output regulation subunit controls the power electronic converter to achieve precise output of pulse current / voltage according to the PWM signal, while monitoring the deviation between the output power and the commanded power in real time, and dynamically correcting the PWM duty cycle through the PID controller to ensure power output accuracy.

[0034] Optionally, the closed-loop feedback optimization module includes: an error calculation and analysis subunit and a parameter correction execution subunit;

[0035] The error calculation and analysis subunit collects the actual operating power data of the real-time execution control module. Predicted power data from the dynamic load forecasting module Calculate the load forecasting error Determine whether the error exceeds the preset threshold;

[0036] The parameter correction execution subunit: when When the error is >10%, the error value is input into the dynamic load forecasting module (102) to correct the weight matrix of the LSTM model; when When ≤10%, the weighting coefficient of the flexible control decision module (103) Fine-tuning is performed on λ, with a correction amount of Δ. =0.1× ×sign( -0.5), continuously optimize the system regulation accuracy through parameter correction.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] Based on the load fluctuation coefficient, the present invention adjusts the prediction period. When the load fluctuation ≤ 5%, a 15-minute period is adopted to reduce the consumption of computing resources. When 5% < fluctuation ≤ 15%, a 5-minute period is adopted. When the fluctuation > 15%, it is shortened to a 1-minute period to improve the response speed. Combined with the LSTM model incorporating the SOC matrix of ultra-fast charging facilities, the generated load prediction sequence is more in line with the actual operation scenario, which can reduce the risk of distribution network load imbalance caused by prediction lag or deviation, and provide accurate load data support for subsequent regulation decisions.

[0039] In the present invention, the flexible regulation decision-making module constructs an optimization function that combines the load fluctuation cost and the charging completion time cost of the distribution network. The weight coefficient is dynamically adjusted according to the average SOC. When the average SOC of the ultra-fast charging facilities is relatively low, it focuses on shortening the charging time. When the SOC is relatively high, it focuses on the load balance of the distribution network. With the line capacity of the distribution network, the maximum power of the facilities, and the battery temperature as constraints, the optimal total regulation power is solved through the particle swarm optimization algorithm, so that the regulation scheme takes into account the user's energy replenishment needs and the stability of the distribution network within a safe range, and reduces the situation of distribution network overload or low charging efficiency under the single-objective orientation.

[0040] In the present invention, the pulse power distribution module calculates the distribution coefficient based on the SOC of each ultra-fast charging facility, and distributes the optimal total regulation power to each facility to avoid some facilities from overusing the resources of the distribution network. The pulse parameters are adjusted according to the SOC grading. When SOC ≤ 30%, the duty cycle D = 0.8 is set to ensure the charging efficiency. When 30% < SOC ≤ 80%, D changes adaptively according to 0.5 + 0.3×(SOC_i - 30%) / 50%. When SOC > 80%, D = 0.5 and the pulse amplitude linearly decreases to 0.5I rated to protect the battery. Through the peak-shifting design of the high-level period of the pulse, the impact of multi-facility simultaneous high-power charging on the distribution network is reduced, and the probability of line overload is lowered, realizing the flexible coordinated operation of ultra-fast charging facilities and the distribution network.

[0041] In the present invention, the closed-loop feedback optimization module collects the actual operation power data and the predicted power data, calculates the load prediction error. When ε > 10%, the weight matrix of the LSTM model is corrected to improve the subsequent prediction accuracy. When ε ≤ 10%, the weight coefficient of the flexible regulation decision-making module is finely adjusted to form a closed-loop mechanism of "prediction - decision - execution - correction". This dynamic correction method can continuously optimize the system regulation accuracy, reduce the accumulation of regulation deviation caused by parameter curing in the traditional non-feedback system, and enable the system to adapt to the changes in the access scale of ultra-fast charging facilities and the fluctuations in the operating state of the distribution network, and maintain a stable and reliable regulation effect in the long term. Description of the Drawings

[0042] Figure 1 This is a flowchart of the overall system in this invention;

[0043] Figure 2 This is a flowchart of the multi-source data acquisition module in this invention;

[0044] Figure 3 This is a flowchart of the dynamic load forecasting module in this invention;

[0045] Figure 4 This is a flowchart of the flexible control decision-making module in this invention;

[0046] Figure 5 This is a flowchart of the pulse power distribution module in this invention;

[0047] Figure 6 This is a flowchart of the real-time execution control module in this invention;

[0048] Figure 7 This is a flowchart of the closed-loop feedback optimization module in this invention;

[0049] In the picture:

[0050] 101. Multi-source data acquisition module; 102. Dynamic load prediction module; 103. Flexible control decision module; 104. Pulse power distribution module; 105. Real-time execution control module; 106. Closed-loop feedback optimization module. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] like Figures 1 to 7 As shown, this invention provides a flexible control system for ultra-fast charging facilities oriented towards active load balancing in distribution networks. It is characterized by comprising a multi-source data acquisition module 101, a dynamic load prediction module 102, a flexible control decision module 103, a pulse power allocation module 104, a real-time execution control module 105, and a closed-loop feedback optimization module 106, wherein:

[0053] Multi-source data acquisition module 101: used to synchronously acquire power distribution network operation status data and ultra-fast charging facility parameter data, perform noise filtering and standardization processing on the acquired data, generate a standardized data matrix and transmit it to dynamic load forecasting module 102;

[0054] Dynamic load forecasting module 102: Based on the standardized data matrix output by the multi-source data acquisition module 101, it uses an adaptive periodic forecasting algorithm and a long short-term memory network (LSTM) model to generate a distribution network load forecasting sequence for a specific future period, and transmits it to the flexible control decision module 103.

[0055] Flexible control decision module 103: Receives the load forecast sequence from dynamic load forecast module 102, constructs a multi-objective optimization objective function, combines the constraints of the distribution network and ultra-fast charging facilities, solves the optimal total control power through optimization algorithm, and outputs it to pulse power distribution module 104;

[0056] Pulse power distribution module 104: Based on the optimal total control power output by the flexible control decision module 103, and combined with the state of charge (SOC) of each ultra-fast charging facility, the SOC-load collaborative pulse scheduling algorithm is used to generate a pulse parameter instruction set for each ultra-fast charging facility and transmit it to the real-time execution control module 105.

[0057] The real-time execution control module 105 receives the pulse parameter instruction set from the pulse power distribution module 104, generates a pulse width modulation (PWM) signal, controls the power electronic converter to achieve accurate output of pulse current / voltage, and monitors the power deviation in real time and performs dynamic adjustment.

[0058] Closed-loop feedback optimization module 106: Collects the actual operating data of the real-time execution control module 105 and the prediction data of the dynamic load prediction module 102, calculates the load prediction error, and corrects the model parameters of the dynamic load prediction module 102 or the weight coefficients of the flexible control decision module 103 according to the error magnitude, thereby improving the system control accuracy.

[0059] The multi-source data acquisition module 101 includes: a power distribution network data acquisition subunit and a facility parameter acquisition subunit;

[0060] The distribution network data acquisition subunit is used to collect node voltages, line active power, reactive power and load fluctuations of the distribution network. The sampling frequency is set to 50Hz-2kHz. At the same time, it performs preliminary screening on the collected time-series measurement data to remove obvious outliers.

[0061] The facility parameter acquisition subunit is used to collect the current state of charge (SOC), battery temperature, rated charging power and pulse charging cycle of each ultra-fast charging facility, and synchronously receive the operating status feedback signals of each facility to ensure the real-time and completeness of parameter acquisition.

[0062] The dynamic load forecasting module 102 includes: a forecasting cycle adjustment subunit and a load sequence forecasting subunit;

[0063] The predicted cycle adjustment subunit calculates the fluctuation coefficient based on the distribution network load shift.

[0064] ;

[0065] in, Let be the load fluctuation at time t. The load fluctuation at time t-1; and according to Value setting prediction period , ≤5% =15 minutes, 5%< ≤15% =5 minutes >15% =1 minute;

[0066] The load sequence prediction subunit employs a Long Short-Term Memory (LSTM) network model. It inputs standardized historical load data and the current SOC matrix from the multi-source data acquisition module (101), and through model training and iterative calculation, outputs future load data. Load forecast sequence within the time period to .

[0067] The flexible control decision module 103 further includes: an optimization target construction subunit and an optimal power solution subunit;

[0068] The optimization objective construction sub-unit: Constructing a multi-objective optimization function:

[0069] ;

[0070] in, Costs associated with power distribution network load fluctuations (t=1 to ); (i is the ultra-fast charging facility number), and a dynamic weighting coefficient is set simultaneously:

[0071] ;

[0072] ;

[0073] in, For the average SOC, and + =1;

[0074] The optimal power calculation subunit, under the constraints of the upper limit of the distribution network line capacity, the maximum power of the ultra-fast charging facility, and the battery temperature threshold, uses a particle swarm optimization algorithm to solve the objective function and iteratively calculates the optimal total control power. The output is sent to the pulse power distribution module 104.

[0075] The pulse power distribution module 104 further includes: a power distribution calculation subunit and a pulse parameter generation subunit;

[0076] The power allocation calculation subunit: based on the optimal total control power output by the flexible control decision module 103. Combined with the state of charge of various ultra-fast charging facilities Calculate the allocation coefficient:

[0077] ;

[0078] This allows us to obtain the allocated power for each facility:

[0079] ;

[0080] The pulse parameter generation subunit: based on the allocated power and Adjust pulse parameters in stages. When ≤30%, duty cycle D=0.8; when 30% < ≤80%, D=0.5+0.3×( -30%) / 50%, >80%, D=0.5 and amplitude I linearly decreases to 0.5I rated, generating an instruction set including pulse period, high-level duration ton, low-level duration toff, and amplitude adjustment curve, ensuring that the high-level pulse periods of each facility do not overlap.

[0081] The real-time execution control module 105 further includes: a control signal generation subunit and a power output adjustment subunit;

[0082] The control signal generation subunit receives the pulse parameter instruction set from the pulse power distribution module 104, performs signal conversion and processing through a digital signal processor (DSP), generates a pulse width modulation (PWM) control signal, and transmits it to the power electronic converter.

[0083] The power output regulation subunit controls the power electronic converter to achieve precise output of pulse current / voltage according to the PWM signal, while monitoring the deviation between the output power and the commanded power in real time, and dynamically correcting the PWM duty cycle through the PID controller to ensure power output accuracy.

[0084] The closed-loop feedback optimization module 106 includes: an error calculation and analysis subunit and a parameter correction execution subunit;

[0085] The error calculation and analysis subunit collects the actual operating power data of the real-time execution control module 105. Predicted power data from dynamic load forecasting module 102 Calculate the load forecasting error Determine whether the error exceeds the preset threshold;

[0086] The parameter correction execution subunit: when When the error is greater than 10%, the error value is input into the dynamic load prediction module 102 to correct the weight matrix of the LSTM model; when When ≤10%, the weighting coefficient of the flexible control decision module 103 Fine-tuning is performed on λ, with a correction amount of Δ. =0.1× ×sign( -0.5), the system control accuracy is continuously optimized through parameter correction.

[0087] Specifically, this embodiment uses a 10kV distribution network in an urban community as the application scenario. This distribution network connects to a charging station (numbered 1-5) containing five ultra-fast charging devices. The maximum capacity of the distribution network lines is 800kW, each ultra-fast charging device has a rated power of 180kW, and the battery temperature safety threshold is 55℃. The various modules of the system interact via industrial Ethernet. The multi-source data acquisition module has a sampling frequency set to 1kHz, and the closed-loop feedback optimization module performs error analysis and parameter correction every 2 minutes. The specific embodiment is illustrated in the following data table:

[0088] Multi-source data acquisition module 101 collects data.

[0089]

[0090] Calculation results of dynamic load forecasting module 102

[0091]

[0092] Flexible control decision module 103 calculates data

[0093]

[0094] Pulse power distribution module 104 distribution results

[0095]

[0096] Real-time execution control module 105 operation data

[0097]

[0098] Closed-loop feedback optimization module 106 corrects data

[0099]

[0100] Implementation effect verification data table (comparing the system before and after implementation)

[0101]

[0102] The working principle and usage process of this invention are as follows: The multi-source data acquisition module 101 synchronously acquires distribution network operation data and ultra-fast charging facility parameters, and transmits them to the dynamic load prediction module 102 after preprocessing. This module adjusts the prediction period according to the load fluctuation coefficient and outputs a load prediction sequence based on the LSTM model, providing a basis for the flexible control decision module. The flexible control decision module 103 constructs a multi-objective optimization function to solve for the optimal total control power under the constraints of the distribution network and facilities. The pulse power allocation module 104 allocates power according to the SOC and generates pulse parameter instructions accordingly. The real-time execution control module 105 receives the instructions and generates a PWM signal to control the converter to output pulse power. The closed-loop feedback optimization module 106 collects actual data and predicted data, calculates errors, and corrects model parameters or weight coefficients, forming a closed-loop process of "acquisition-prediction-decision-allocation-execution-correction" to achieve distribution network load balance and flexible control of ultra-fast charging facilities.

[0103] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0104] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A super-fast charging facility flexible regulation system for power distribution network load active balancing, characterized in that, The system comprises a multi-source data acquisition module (101), a dynamic load prediction module (102), a flexible regulation and decision module (103), a pulse power distribution module (104), a real-time execution control module (105) and a closed-loop feedback optimization module (106), wherein: The multi-source data acquisition module (101) is used for synchronously acquiring power distribution network operation state data and super-fast charging facility parameter data, filtering and standardizing the collected data, generating a standardized data matrix and transmitting the standardized data matrix to the dynamic load prediction module (102); The dynamic load prediction module (102) is based on the standardized data matrix output by the multi-source data acquisition module (101), adopts an adaptive cycle prediction algorithm and a long short-term memory network (LSTM) model, generates a power distribution network load prediction sequence in a future specific period, and transmits the load prediction sequence to the flexible regulation and decision module (103); The flexible regulation and decision module (103) receives the load prediction sequence of the dynamic load prediction module (102), constructs a multi-objective optimization objective function, combines the constraint conditions of the power distribution network and the super-fast charging facility, solves the optimal total regulation and control power through an optimization algorithm, and outputs the optimal total regulation and control power to the pulse power distribution module (104); The pulse power distribution module (104) generates a pulse parameter instruction set of each super-fast charging facility according to the optimal total regulation and control power output by the flexible regulation and decision module (103), combines the state of charge (SOC) of each super-fast charging facility, adopts an SOC-load collaborative pulse scheduling algorithm, and transmits the pulse parameter instruction set to the real-time execution control module (105); The real-time execution control module (105) receives the pulse parameter instruction set of the pulse power distribution module (104), generates a pulse width modulation (PWM) signal, controls a power electronic converter to realize accurate output of a pulse current / voltage, and monitors a power deviation in real time and dynamically adjusts the power deviation; The closed-loop feedback optimization module (106) collects actual operation data of the real-time execution control module (105) and prediction data of the dynamic load prediction module (102), calculates a load prediction error, modifies model parameters of the dynamic load prediction module (102) or weight coefficients of the flexible regulation and decision module (103) according to the error size, and improves system regulation and control accuracy. 2.The power grid load-oriented active balancing ultra-fast charging facility flexible regulation system of claim 1, wherein, The multi-source data acquisition module (101) comprises a power distribution network data acquisition subunit and a facility parameter acquisition subunit; The power distribution network data acquisition subunit is used for acquiring node voltage, line active power, reactive power and load fluctuation of the power distribution network, and the sampling frequency is set to 50Hz-2kHz. The time series measurement data collected is preliminarily screened to remove obvious abnormal values. The facility parameter acquisition subunit is used for acquiring the current state of charge (SOC), battery temperature, rated charging power and pulse charging period of each super-fast charging facility, synchronously receiving operation state feedback signals of each facility, and ensuring real-time and integrity of parameter acquisition. 3.The power grid load oriented active balancing ultra-fast charging facility flexible regulation system of claim 1, wherein, The dynamic load prediction module (102) comprises a prediction cycle adjustment subunit and a load sequence prediction subunit; The prediction cycle adjustment subunit calculates a fluctuation coefficient through power distribution network load fluctuation: ； wherein, is the load fluctuation amount at time t, is the load fluctuation amount at time t-1; and according to the prediction period is set to take values , when ≤ 5% = 15 minutes, 5% when ≤ 15% = 5 minutes, when > 15% = 1 minute; The load sequence prediction unit: adopts a long short-term memory network (LSTM) model, inputs the standardized historical load data and the current SOC matrix output by the multi-source data acquisition module (101), and outputs the load prediction sequence in the future period through model training and iterative calculation to . 4.The power grid load oriented active balancing ultra-fast charging facility flexible regulation system of claim 1, wherein, The flexible regulation decision module (103) further comprises an optimization target construction subunit and an optimal power solution subunit; The optimization target construction subunit constructs a multi-objective optimization function: ； wherein, is the cost of load fluctuation of the distribution network, , t = 1 to ; i is the ultra-fast charging facility number, and a dynamic weight coefficient is set ; ; wherein is the average SOC, and + = 1; The optimal power solving subunit: with the upper limit of the distribution network line capacity, the maximum power of the super-fast charging facility and the battery temperature threshold as the constraint conditions, the particle swarm optimization algorithm is used to solve the optimization objective function, and the optimal total regulation and control power is obtained through iterative calculation , and output to the pulse power distribution module (104).

5. The power grid load oriented active balancing of super-fast charging facility flexible regulation system according to claim 1, characterized in that, The pulse power distribution module (104) further comprises a power distribution calculation subunit and a pulse parameter generation subunit; The power distribution calculation sub-unit: according to the optimal total regulation power output by the flexible regulation decision module (103) , in combination with the state of charge of each super-fast charging facility , calculate the distribution coefficient: ; And then the distribution power of each facility is obtained: ； The pulse parameter generating subunit: according to the distribution power With The pulse parameter is adjusted, When ≤30%, duty cycle D=0.8, 30% When ≤80%, D=0.5+0.3×( -30%) / 50%, When >80%, D=0.5 and amplitude I linearly reduces to 0.5I rated, generate instruction set containing pulse period, high level time ton, low level time toff and amplitude adjustment curve, ensure that each facility pulse high level period has no overlap. 6.The power grid load oriented active balancing ultra-fast charging facility flexible regulation system of claim 1, wherein, The real-time execution control module (105) further comprises a control signal generation subunit and a power output adjustment subunit; The control signal generation subunit receives the pulse parameter instruction set of the pulse power distribution module (104), performs signal conversion and processing through a digital signal processor (DSP), generates a pulse width modulation (PWM) control signal, and transmits the signal to a power electronic converter; The power output adjustment subunit controls the power electronic converter to realize accurate output of pulse current / voltage according to the PWM signal, simultaneously monitors the deviation of the output power and the instruction power in real time, dynamically corrects the PWM duty ratio through a PID controller, and ensures the power output accuracy.

7. The power grid load oriented active balancing of super-fast charging facility flexible regulation system according to claim 1, characterized in that, The closed-loop feedback optimization module (106) comprises an error calculation and analysis subunit and a parameter correction and execution subunit; The error calculation and analysis subunit: collects actual running power data of the real-time execution control module (105) The predicted power data of the dynamic load prediction module (102) Calculate the load prediction error Determine whether the error exceeds the preset threshold The parameter correction execution subunit: when >10%, the error value is input into the dynamic load prediction module (102), and the weight matrix of the LSTM model is corrected; when ≤10%, the weight coefficients and λ of the flexible regulation decision module (103) are fine-tuned, and the correction amount is Δ =0.1× ×sign( -0.5), and the system regulation accuracy is continuously optimized through parameter correction.

Images