Active power control method and system for wind farm based on multi-objective and pi control

Abstract

The application discloses a wind farm active power control method and system based on multi-target and PI control, and the method comprises the following steps: configuring unit operation parameters according to the model of a wind turbine generator unit, and initializing algorithm parameters and a target function of a multi-target optimization algorithm based on decomposition; collecting key data of unit operation in real time and receiving a field-level active power target value; fitting the real-time theoretical power of each unit based on the key data of each wind turbine generator unit collected in real time and the configured unit operation parameters; judging a control trigger condition; performing multi-target optimization solving with each unit as an independent dimension, and issuing each unit after evaluating the solution quality; and calling PI control as a backup when the algorithm fails. The multi-target optimization algorithm based on decomposition adopted by the application is naturally compatible with multi-constraint multi-target optimization, and can guarantee engineering reliability through hierarchical cooperation of a main unit and a backup unit, realize unit adjustment and fine adjustment in combination with deviation layering and dynamic periods, effectively balance power grid friendliness, load balance and model adaptation, and has a good engineering application prospect.

Description

Technical Field

[0001] This invention relates to the technical field of wind power generation control, and in particular to a method and system for controlling the active power of a wind farm based on multi-objective and PI control. Background Technology

[0002] With the accelerated transformation of the global energy structure, wind power, as an important component of clean energy, has seen its installed capacity grow rapidly. Early wind power aimed for "full grid integration," with wind turbines mostly operating in maximum power point tracking (MPPT) mode to capture as much wind energy as possible. However, as the installed capacity of wind power continues to expand, insufficient grid absorption capacity has led to an increasingly prominent problem of "wind curtailment," and the control mode of wind farms has gradually shifted from "generating as much power as possible" to "operating with limited power according to dispatch instructions."

[0003] However, existing active power control technologies for wind farms have the following main shortcomings:

[0004] Traditional algorithms have significant limitations in multi-constraint extension, relying on manual weighting coefficients for objective trade-offs and being difficult to extend. They cannot adapt to the differentiated control characteristics of different models under complex operating conditions and the increasingly complex and diverse unit control requirements. When using multi-objective intelligent control algorithms, problems such as strong randomness in solution results and unstable convergence results are prone to occur, making it impossible to directly meet the requirements for engineering implementation. In addition, existing methods mostly use fixed control cycles, failing to fully consider differentiated control under operating conditions and the need for minimal unit adjustments, making it difficult to effectively reduce unit fatigue loads. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for controlling the active power of wind farms based on multi-objective and PI control.

[0006] To achieve the above objectives, the technical solution provided by this invention is as follows:

[0007] A wind farm active power control method based on multi-objective and PI control includes the following steps:

[0008] S1. Configure the unit operating parameters according to the model of each wind turbine in the wind farm, and initialize the algorithm parameters and objective function of the decomposition-based multi-objective optimization algorithm;

[0009] S2. Collect key data of each wind turbine in real time and receive the field-level active power target value issued by the wind farm-side active power automatic generation control system.

[0010] S3. Based on the key data of each wind turbine and the configured unit operating parameters collected in real time, obtain the real-time theoretical power of each wind turbine.

[0011] S4. Compare the field-level active power target value mentioned in step S2 with the field-level active power target value issued in the previous control cycle. If the change exceeds the preset threshold, or the deviation between the actual active power value obtained by the dynamic algorithm periodic monitoring and the field-level active power target value meets the control triggering condition, then proceed to step S5 to allocate active power using a decomposition-based multi-objective optimization algorithm; otherwise, each wind turbine unit continues to maintain its current state.

[0012] S5. Using the aforementioned decomposition-based multi-objective optimization algorithm, with the real-time theoretical power and key data of each wind turbine as the algorithm input, within the preset total iteration time and total number of iterations, the multi-objective optimization problem is decomposed into several single-objective sub-problems, and solved iteratively through neighborhood relationship co-evolution; after the iteration terminates, the obtained solution set is sorted by non-dominated order, dominant solutions are removed and a Pareto front is constructed; the quality of the non-dominated solutions in the Pareto front is evaluated, and when the evaluation result meets the preset control requirements, the non-dominated solutions are decoded into the active power target value of each wind turbine and sent to the corresponding turbine for execution;

[0013] When the iteration time of the decomposition-based multi-objective optimization algorithm reaches the total iteration time limit or the number of iterations reaches the total number of iterations limit, or when it is determined that the non-dominated solution does not meet the preset control requirements, the pre-stored PI control algorithm is invoked to perform fallback control on the wind turbine to ensure that the overall control action time of the active power of the wind farm meets the grid connection assessment requirements.

[0014] Furthermore, in step S1, the objective function includes a grid-friendly objective function and an objective function for reducing unit fatigue load;

[0015] The grid-friendly objective function includes a subproblem of minimizing tracking error and a subproblem of prioritizing control of generating units with a large proportion of remaining capacity.

[0016] The objective function for reducing unit fatigue load includes the sub-problem of minimizing downtime and the number of control consoles, and the sub-problem of prioritizing wind turbine units with high control effectiveness evaluation values.

[0017] Furthermore, the objective function of the subproblem of minimizing tracking error is... as follows:

[0018] ;

[0019] in, For tracking error; For decision variables; express The optimal solution; To be allocated to the Active power increment of typhoon generators:

[0020] ;

[0021] For the first Typhoon generator power-up rate; For the first Typhoon generator power reduction rate; For the first The current active power generated by the typhoon generator; For the first Available active power of typhoon generator units; For the first Minimum active power of typhoon generator units; To control the cycle;

[0022] The automatic active power generation control system on the wind farm side controls the total active power increment:

[0023] ;

[0024] The target value of active power at the farm level is issued by the automatic active power generation control system on the wind farm side. This represents the total number of wind turbine units.

[0025] The objective function of the subproblem of prioritizing units with a remaining capacity percentage greater than a set value. as follows:

[0026] ;

[0027] in, For the first Rated active power of typhoon generator units.

[0028] Furthermore, the objective function of the minimum shutdown and console number subproblem is... as follows:

[0029] ;

[0030] in, This refers to the number of shutdowns and control consoles. This is a level 1 penalty coefficient; This is a secondary penalty coefficient; This represents the total number of wind turbine units. For the first State variables indicating whether the typhoon generator is shut down:

[0031] ;

[0032] For the first State variables indicating whether the typhoon generator is operating:

[0033] ;

[0034] The objective function of the wind turbine sub-problem with the highest priority control effectiveness evaluation value as follows:

[0035] ;

[0036] For the first The efficiency score of a typhoon generator ranges from 0 to 1. A higher efficiency score indicates a healthier generator and a lower probability of power fluctuations.

[0037] Further, step S3 includes:

[0038] Based on the key data collected in real time for each wind turbine, when each wind turbine is in non-power curtailment mode, the wind speed and actual active power of the corresponding wind turbine are added to a historical fitting sample cyclic queue of a preset wind speed sub-box. The real-time theoretical power of each wind turbine is obtained by fitting based on the historical fitting samples.

[0039] Furthermore, the control triggering conditions are determined based on a dynamic control cycle method for active power execution deviation classification, including:

[0040] The deviation between the obtained actual active power value and the target value of the field-level active power. As the object, based on the assessment error band of the wind farm-side active power automatic generation control system, the deviation range is divided into five continuous regions, which are defined in order from negative to positive as: over-generation warning zone, upper under-regulation zone, balance zone, lower under-regulation zone, and under-generation warning zone.

[0041] when deviation When the system is in the over-issue warning zone or the under-issue warning zone, a decomposition-based multi-objective optimization algorithm is immediately invoked for intervention and adjustment, and the time interval between two consecutive control actions based on the decomposition-based multi-objective optimization algorithm is not less than a preset time threshold.

[0042] when deviation When the control is in the upper or lower adjustment zone, an adjustment strategy is adopted to extend the control period by a preset multiple based on the original control period, thereby achieving adjustment control.

[0043] when deviation When the system is in the equilibrium zone, no further adjustment based on decomposition-based multi-objective optimization algorithm is performed.

[0044] Furthermore, less than -1% The interval is set as the over-issuance warning zone; [-1%] -0.5% The interval of [-0.5%] is set as the upper adjustment zone; the interval of [-0.5%] is set as the lower adjustment zone. 0.5% The interval [0.5%] is set as the equilibrium region; 1% The interval [] is set as the lower adjustment zone; greater than 1% The interval is designated as a low-risk warning zone; This refers to the active capacity.

[0045] Furthermore, in step S5, the quality assessment of the non-dominated solution includes: performing a preliminary logical judgment on the active power output of each wind turbine, which includes judging whether the sum of the target output values ​​of each wind turbine satisfies the total target value, judging whether the power rise and fall direction of a single wind turbine is consistent with the direction of change of the total active power target of the field level, and judging whether the output power of a single wind turbine satisfies the constraints set by the algorithm; and evaluating the convergence accuracy and distribution uniformity of the solution through the generation distance GD and the spatial distribution index SP.

[0046] Furthermore, step S1 also includes: providing a human-machine interface through the system's backend management interface to set the objective function online and to set or enable / disable the turbine control function online, so as to meet the diverse needs of wind farms and turbine control.

[0047] Furthermore, to achieve the above objectives, the present invention also provides a wind farm active power control system based on multi-objective and PI control, used to implement the above-mentioned wind farm active power control method based on multi-objective and PI control. The system includes:

[0048] The initialization module is used to configure the unit operating parameters according to the model of each wind turbine in the wind farm, and to initialize the algorithm parameters and objective function of the decomposition-based multi-objective optimization algorithm.

[0049] The data acquisition module is used to collect key data of each wind turbine in real time and receive the field-level active power target value issued by the wind farm-side active power automatic generation control system.

[0050] The real-time theoretical power calculation module is used to obtain the real-time theoretical power of each wind turbine based on the key data of each wind turbine collected in real time and the configured unit operating parameters.

[0051] The control trigger judgment module is used to determine whether the main control module should be triggered; the judgment process includes:

[0052] The target value of active power at the field level is compared with the target value of active power at the field level issued in the previous control cycle. If the change exceeds the preset threshold, or if the deviation between the actual active power value obtained by the periodic monitoring based on the dynamic algorithm and the target value of active power at the field level meets the control triggering condition, the main control module is triggered. Otherwise, the main control module is not triggered, and each wind turbine continues to maintain its current state.

[0053] The main control module is used to employ the decomposition-based multi-objective optimization algorithm, taking the real-time theoretical power and key data of each wind turbine as algorithm input. Within the preset total iteration time and total number of iterations, the multi-objective optimization problem is decomposed into several single-objective sub-problems, which are solved iteratively through neighborhood relationships and co-evolution. After the iteration terminates, the obtained solution set is sorted by non-dominated solution, dominant solutions are removed, and a Pareto front is constructed. The quality of the non-dominated solutions in the Pareto front is evaluated. When the evaluation result meets the preset control requirements, the non-dominated solutions are decoded into the active power target value of each wind turbine and sent to the corresponding turbine for execution.

[0054] The backup control module is used to call a pre-stored PI control algorithm to perform fallback control on the wind turbine when the iteration time of the decomposition-based multi-objective optimization algorithm reaches the total iteration time limit or the number of iterations reaches the total number of iterations limit, or when it is determined that the non-dominated solution does not meet the preset control requirements.

[0055] Compared with existing technologies, the principles and advantages of this technical solution are as follows:

[0056] 1. Unlike traditional methods that transform multi-objective problems into single-objective problems based on linear weighting, this technical solution employs a decomposition-based multi-objective optimization algorithm to decompose the multi-objective optimization problem into several single-objective sub-problems. It utilizes neighborhood relationships for co-evolution, naturally supporting parallel optimization solutions with multiple constraints and control objectives, eliminating the need for manually preset fixed weight coefficients for objective trade-offs. Furthermore, during the initialization phase, based on the configuration and operating parameters of each wind turbine model (including active power capacity, ramp / deceleration rate, minimum active power setpoint, control mode, etc.), the physical constraints and control characteristics of different turbine models are directly embedded into the solution framework of the decomposition-based multi-objective optimization algorithm. This allows the algorithm to automatically adapt to the differentiated constraint boundaries of different turbine models during the optimization process. The system's backend management interface provides a human-machine interface, allowing online setting or enabling / disabling of objective functions and turbine control functions to meet the needs of different wind farms and turbine control requirements, achieving synergy between the algorithm model and the human-machine configuration mechanism.

[0057] 2. Construct a hierarchical collaborative mechanism of "main control + PI control backup". The main control adopts a decomposition-based multi-objective optimization algorithm for refined multi-objective optimization; the backup control pre-stores traditional PI control algorithms and seamlessly switches in case of main control failure, ensuring that the overall active power control action time of the wind farm always meets the grid connection assessment requirements. After the algorithm stops iterating, Pareto front is constructed by judging non-dominated relationships and dominant solutions are eliminated, while global non-dominated solutions are retained; at the same time, the generation distance (GD) index is used to evaluate convergence accuracy and the spatial distribution index (SP) is used to evaluate distribution uniformity. Only when the non-dominated solution meets the preset control requirements is it issued to each unit PLC for execution; otherwise, it is rejected. Before issuing the solutions of each dimension as the active power target value of the wind turbine, the output value of each wind turbine is preliminarily judged: whether the sum of the target values ​​of each wind turbine meets the total target value, whether the power rise and fall direction of a single wind turbine is consistent with the farm level, and whether the output power of a single wind turbine meets the constraints set by the algorithm, to ensure the rationality and executability of the issued instructions. When a decomposition-based multi-objective optimization algorithm fails to obtain a solution that meets the preset control requirements within the set iteration time or number of iterations, the PI control algorithm (logic) is automatically invoked for fallback allocation. Through a multi-guarantee mechanism of "quality assessment + access control + PI fallback," the problem of strong randomness and unstable convergence in the solution results of multi-objective intelligent algorithms is effectively overcome, making the entire control process free from human intervention, and ensuring reliable control logic and stable execution.

[0058] 3. By combining deviation stratification with dynamic cycle coordination, the system achieves refined control that enables "rapid response to large deviations, minimal adjustments for small deviations, and steady-state operation when there are no deviations." Under the premise of meeting grid-friendly and grid connection assessment requirements, the system effectively reduces the frequency of operation of the unit's electrical components, lowers the fatigue load on the unit, and extends the service life of the equipment. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0060] Figure 1 This is a flowchart illustrating the principle of the active power control method for wind farms based on multi-objective and PI control, as described in an embodiment of the present invention.

[0061] Figure 2 This is a block diagram showing the connections between modules in a wind farm active power control system based on multi-objective and PI control, according to an embodiment of the present invention. Detailed Implementation

[0062] The present invention will be further described below with reference to specific embodiments:

[0063] like Figure 1 As shown, the active power control method for wind farms based on multi-objective and PI control described in this embodiment includes S1, configuring the unit operating parameters according to the model of each wind turbine in the wind farm, and initializing the algorithm parameters and objective function of the decomposition-based multi-objective optimization algorithm (MOEA / D algorithm);

[0064] In this step, the configured unit operating parameters include active power capacity, data acquisition points, active power control mode, unit operating status mapping logic, unit power curtailment status mapping logic, minimum active power control value for a single unit, and single unit ramp-up / ramp-down rate, etc.

[0065] Algorithm parameters include:

[0066] Total iteration duration and iteration count limits: The system defaults to a total iteration duration of 3 seconds and an algorithm iteration count of 20,000. These limits can be changed based on the response lag requirements or response arrival time requirements of the active power automatic generation control system on the regional wind farm side, as well as a comprehensive evaluation of the number of wind turbine units and the measured algorithm performance.

[0067] The objective functions include a grid-friendly objective function and an objective function for reducing unit fatigue load;

[0068] The grid-friendly objective function includes the subproblem of minimizing tracking error and the subproblem of prioritizing control of generating units with a large proportion of remaining capacity.

[0069] The objective function of minimizing the tracking error subproblem as follows:

[0070] ;

[0071] in, For tracking error; For decision variables; express The optimal solution; To be allocated to the Active power increment of typhoon generators:

[0072] ;

[0073] For the first Typhoon generator power-up rate; For the first Typhoon generator power reduction rate; For the first The current active power generated by the typhoon generator; For the first Available active power of typhoon generator units; For the first Minimum active power of typhoon generator units; To control the cycle;

[0074] The automatic active power generation control system on the wind farm side controls the total active power increment:

[0075] ;

[0076] The target value of active power at the farm level is issued by the automatic active power generation control system on the wind farm side. This represents the total number of wind turbine units.

[0077] The objective function of the subproblem of prioritizing units with a remaining capacity percentage greater than a set value. as follows:

[0078] ;

[0079] in, For the first Rated active power of typhoon generator units.

[0080] The objective function for reducing unit fatigue load includes the sub-problem of minimizing downtime and the number of control consoles, and the sub-problem of wind turbine units with the highest control performance evaluation value.

[0081] The objective function of the minimum shutdown and console number subproblem is... as follows:

[0082] ;

[0083] in, This refers to the number of shutdowns and control consoles. As the first-level penalty coefficient, take ; The second-level penalty coefficient is taken as... ; This represents the total number of wind turbine units. For the first State variables indicating whether the typhoon generator is shut down:

[0084] ;

[0085] For the first State variables indicating whether the typhoon generator is operating:

[0086] ;

[0087] The objective function of the wind turbine sub-problem with the highest priority control effectiveness evaluation value as follows:

[0088] ;

[0089] For the first The efficiency score of a typhoon generator ranges from 0 to 1. A higher efficiency score indicates a healthier generator and a lower probability of power fluctuations.

[0090] This step also includes providing a human-machine interface through the system's backend management interface to set the objective function online and to set or enable / disable the turbine control function online, so as to meet the diverse needs of wind farms and turbine control.

[0091] S2. Collect key data of each wind turbine in real time and receive the field-level active power target value issued by the wind farm-side active power automatic generation control system.

[0092] Key data in this step include operating status, power rationing status, blade angle, wind speed, actual active power, target active power, and ambient temperature.

[0093] S3. Based on the key data and configured operating parameters of each wind turbine collected in real time, the real-time theoretical power of each wind turbine is obtained. The specific process is as follows:

[0094] Based on the key data collected in real time for each wind turbine, when each wind turbine is in non-power curtailment mode, the wind speed and actual active power of the corresponding wind turbine are added to a historical fitting sample cyclic queue of a 0.5m / s wind speed sub-box. The real-time theoretical power of each wind turbine is obtained by fitting based on the historical fitting samples.

[0095] S4. Compare the field-level active power target value mentioned in step S2 with the field-level active power target value issued in the previous control cycle. If the change exceeds the preset threshold, or the deviation between the actual active power value obtained by the dynamic algorithm periodic monitoring and the field-level active power target value meets the control triggering condition, then proceed to step S5 to allocate active power using a decomposition-based multi-objective optimization algorithm; otherwise, each wind turbine unit continues to maintain its current state.

[0096] In this step, the control triggering conditions are determined based on the dynamic control cycle method of active power execution deviation classification, including:

[0097] The deviation between the obtained actual active power value and the target value of the field-level active power. As the object, based on the assessment error band of the wind farm-side active power automatic generation control system, the deviation range is divided into five continuous regions, which are defined in order from negative to positive as: over-generation warning zone, upper under-regulation zone, balance zone, lower under-regulation zone, and under-generation warning zone.

[0098] when deviation When the system is in the over-issue warning zone or the under-issue warning zone, a decomposition-based multi-objective optimization algorithm is immediately invoked for intervention and adjustment, and the time interval between two consecutive control actions based on the decomposition-based multi-objective optimization algorithm is not less than a preset time threshold.

[0099] when deviation When the control is in the upper or lower adjustment zone, an adjustment strategy is adopted to extend the control period by a preset multiple based on the original control period, thereby achieving adjustment control.

[0100] when deviation When the system is in the equilibrium zone, no further adjustment based on decomposition-based multi-objective optimization algorithm is performed.

[0101] More specifically, less than -1% The interval is set as the over-issuance warning zone; [-1%] -0.5% The interval of [-0.5%] is set as the upper adjustment zone; the interval of [-0.5%] is set as the lower adjustment zone. 0.5% The interval [0.5%] is set as the equilibrium region; 1% The interval [] is set as the lower adjustment zone; greater than 1% The interval is designated as a low-risk warning zone; This refers to the active capacity.

[0102] S5. Using the aforementioned decomposition-based multi-objective optimization algorithm, with the real-time theoretical power and key data of each wind turbine as the algorithm input, within the preset total iteration time and total number of iterations, the multi-objective optimization problem is decomposed into several single-objective sub-problems, and solved iteratively through neighborhood relationship co-evolution; after the iteration terminates, the obtained solution set is sorted by non-dominated order, dominant solutions are removed and a Pareto front is constructed; the quality of the non-dominated solutions in the Pareto front is evaluated, and when the evaluation result meets the preset control requirements, the non-dominated solutions are decoded into the active power target value of each wind turbine and sent to the corresponding turbine for execution;

[0103] In this step, the quality assessment of the non-dominated solution includes: performing a preliminary logical judgment on the active power output of each wind turbine, which includes judging whether the sum of the target output values ​​of each wind turbine meets the total target value, judging whether the power rise and fall direction of a single wind turbine is consistent with the direction of change of the total active power target of the field level, and judging whether the output power of a single wind turbine meets the constraints set by the algorithm; and evaluating the convergence accuracy and distribution uniformity of the solution through the generation distance GD and the spatial distribution index SP.

[0104] When the iteration time of the decomposition-based multi-objective optimization algorithm reaches the total iteration time limit or the number of iterations reaches the total number of iterations limit, or when it is determined that the non-dominated solution does not meet the preset control requirements, the pre-stored PI control algorithm is invoked to perform fallback control on the wind turbine to ensure that the overall control action time of the active power of the wind farm meets the grid connection assessment requirements.

[0105] In addition, this embodiment also includes a wind farm active power control system based on multi-objective and PI control, used to implement the above-mentioned wind farm active power control method based on multi-objective and PI control, such as... Figure 2 As shown, the system includes:

[0106] The initialization module is used to configure the unit operating parameters according to the model of each wind turbine in the wind farm, and to initialize the algorithm parameters and objective function of the decomposition-based multi-objective optimization algorithm.

[0107] The data acquisition module is used to collect key data of each wind turbine in real time and receive the field-level active power target value issued by the wind farm-side active power automatic generation control system.

[0108] The real-time theoretical power calculation module is used to obtain the real-time theoretical power of each wind turbine based on the key data of each wind turbine collected in real time and the configured unit operating parameters.

[0109] The control trigger judgment module is used to determine whether the main control module should be triggered; the judgment process includes:

[0110] The target value of active power at the field level is compared with the target value of active power at the field level issued in the previous control cycle. If the change exceeds the preset threshold, or if the deviation between the actual active power value obtained by the periodic monitoring based on the dynamic algorithm and the target value of active power at the field level meets the control triggering condition, the main control module is triggered. Otherwise, the main control module is not triggered, and each wind turbine continues to maintain its current state.

[0111] The main control module is used to employ the decomposition-based multi-objective optimization algorithm, taking the real-time theoretical power and key data of each wind turbine as algorithm input. Within the preset total iteration time and total number of iterations, the multi-objective optimization problem is decomposed into several single-objective sub-problems, which are solved iteratively through neighborhood relationships and co-evolution. After the iteration terminates, the obtained solution set is sorted by non-dominated solution, dominant solutions are removed, and a Pareto front is constructed. The quality of the non-dominated solutions in the Pareto front is evaluated. When the evaluation result meets the preset control requirements, the non-dominated solutions are decoded into the active power target value of each wind turbine and sent to the corresponding turbine for execution.

[0112] The backup control module is used to call a pre-stored PI control algorithm to perform fallback control on the wind turbine when the iteration time of the decomposition-based multi-objective optimization algorithm reaches the total iteration time limit or the number of iterations reaches the total number of iterations limit, or when it is determined that the non-dominated solution does not meet the preset control requirements.

[0113] The above-described embodiments are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Therefore, any changes made in accordance with the shape and principle of the present invention should be covered within the protection scope of the present invention.

Claims

1. A wind farm active power control method based on multi-objective and PI control, characterized in that, Includes the following steps: S1. Configure the unit operating parameters according to the model of each wind turbine in the wind farm, and initialize the algorithm parameters and objective function of the decomposition-based multi-objective optimization algorithm; S2. Collect key data of each wind turbine in real time and receive the field-level active power target value issued by the wind farm-side active power automatic generation control system. S3. Based on the key data of each wind turbine and the configured unit operating parameters collected in real time, obtain the real-time theoretical power of each wind turbine. S4. Compare the field-level active power target value mentioned in step S2 with the field-level active power target value issued in the previous control cycle. If the change exceeds the preset threshold, or the deviation between the actual active power value obtained by the dynamic algorithm periodic monitoring and the field-level active power target value meets the control triggering condition, then proceed to step S5 to allocate active power using a decomposition-based multi-objective optimization algorithm; otherwise, each wind turbine unit continues to maintain its current state. S5. Using the aforementioned decomposition-based multi-objective optimization algorithm, with the real-time theoretical power and key data of each wind turbine as the algorithm input, within the preset total iteration time and total number of iterations, the multi-objective optimization problem is decomposed into several single-objective sub-problems, and solved iteratively through neighborhood relationship co-evolution; after the iteration terminates, the obtained solution set is sorted by non-dominated order, dominant solutions are removed and a Pareto front is constructed; the quality of the non-dominated solutions in the Pareto front is evaluated, and when the evaluation result meets the preset control requirements, the non-dominated solutions are decoded into the active power target value of each wind turbine and sent to the corresponding turbine for execution; When the iteration time of the decomposition-based multi-objective optimization algorithm reaches the total iteration time limit or the number of iterations reaches the total number of iterations limit, or when it is determined that the non-dominated solution does not meet the preset control requirements, the pre-stored PI control algorithm is invoked to perform fallback control on the wind turbine to ensure that the overall control action time of the active power of the wind farm meets the grid connection assessment requirements.

2. The wind farm active power control method based on multi-objective and PI control according to claim 1, characterized in that, In step S1, the objective function includes a grid-friendly objective function and an objective function for reducing unit fatigue load; The grid-friendly objective function includes a subproblem of minimizing tracking error and a subproblem of prioritizing control of generating units with a large proportion of remaining capacity. The objective function for reducing unit fatigue load includes the sub-problem of minimizing downtime and the number of control consoles, and the sub-problem of prioritizing wind turbine units with high control effectiveness evaluation values.

3. The wind farm active power control method based on multi-objective and PI control according to claim 2, characterized in that, The objective function of the subproblem of minimizing tracking error as follows: ； in, For tracking error; For decision variables; express The optimal solution; To be allocated to the Active power increment of typhoon generators: ； The power-up rate of the i-th wind turbine unit; For the first Typhoon generator power reduction rate; For the first The current active power generated by the typhoon generator; For the first Available active power of typhoon generator units; For the first Minimum active power of typhoon generator units; To control the cycle; The automatic active power generation control system on the wind farm side controls the total active power increment: ； The target value of active power at the farm level is issued by the automatic active power generation control system on the wind farm side. This represents the total number of wind turbine units. The objective function of the subproblem of prioritizing units with a remaining capacity percentage greater than a set value. as follows: ； in, For the first Rated active power of typhoon generator units.

4. The wind farm active power control method based on multi-objective and PI control according to claim 3, characterized in that, The objective function of the minimum shutdown and console number subproblem as follows: ； in, This refers to the number of shutdowns and control consoles. This is a level 1 penalty coefficient; This is a secondary penalty coefficient; This represents the total number of wind turbine units. For the first State variables indicating whether the typhoon generator is shut down: ； For the first State variables indicating whether the typhoon generator is operating: ； The objective function of the wind turbine sub-problem with the highest priority control effectiveness evaluation value as follows: ； For the first The efficiency score of a typhoon generator ranges from 0 to 1. A higher efficiency score indicates a healthier generator and a lower probability of power fluctuations.

5. The wind farm active power control method based on multi-objective and PI control according to claim 1, characterized in that, Step S3 includes: Based on the key data collected in real time for each wind turbine, when each wind turbine is in non-power curtailment mode, the wind speed and actual active power of the corresponding wind turbine are added to a historical fitting sample cyclic queue of a preset wind speed sub-box. The real-time theoretical power of each wind turbine is obtained by fitting based on the historical fitting samples.

6. The wind farm active power control method based on multi-objective and PI control according to claim 1, characterized in that, The control triggering conditions are determined based on a dynamic control cycle method with graded active power execution deviations, including: The deviation between the obtained actual active power value and the target value of the field-level active power. As the object, based on the assessment error band of the wind farm-side active power automatic generation control system, the deviation range is divided into five continuous regions, which are defined in order from negative to positive as: over-generation warning zone, upper under-regulation zone, balance zone, lower under-regulation zone, and under-generation warning zone. when deviation When the system is in the over-issue warning zone or the under-issue warning zone, a decomposition-based multi-objective optimization algorithm is immediately invoked for intervention and adjustment, and the time interval between two consecutive control actions based on the decomposition-based multi-objective optimization algorithm is not less than a preset time threshold. when deviation When the control is in the upper or lower adjustment zone, an adjustment strategy is adopted to extend the control period by a preset multiple based on the original control period, thereby achieving adjustment control. when deviation When the system is in the equilibrium zone, no further adjustment based on decomposition-based multi-objective optimization algorithm is performed.

7. The wind farm active power control method based on multi-objective and PI control according to claim 6, characterized in that, Less than -1% The interval is set as the over-issuance warning zone; [-1%] -0.5% The interval of [-0.5%] is set as the upper adjustment zone; the interval of [-0.5%] is set as the lower adjustment zone. 0.5% The interval [ ] is set as the equilibrium region; (0.5%) 1% The interval [] is set as the lower adjustment zone; greater than 1% The interval is designated as a low-risk warning zone; This refers to the active capacity.

8. The wind farm active power control method based on multi-objective and PI control according to claim 1, characterized in that, In step S5, the quality assessment of the non-dominated solution includes: performing a preliminary logical judgment on the active power output of each wind turbine, which includes judging whether the sum of the target output values ​​of each wind turbine satisfies the total target value, judging whether the power rise and fall direction of a single wind turbine is consistent with the direction of change of the total active power target of the field level, and judging whether the output power of a single wind turbine satisfies the constraints set by the algorithm; and evaluating the convergence accuracy and distribution uniformity of the solution through the generation distance GD and the spatial distribution index SP.

9. The wind farm active power control method based on multi-objective and PI control according to claim 1, characterized in that, Step S1 also includes: providing a human-machine interface through the system's backend management interface to set the objective function online and to set or enable / disable the turbine control function online, so as to meet the diverse needs of wind farms and turbine control.

10. A wind farm active power control system based on multi-objective and PI control, used to implement the wind farm active power control method based on multi-objective and PI control as described in any one of claims 1-9, characterized in that, include: The initialization module is used to configure the unit operating parameters according to the model of each wind turbine in the wind farm, and to initialize the algorithm parameters and objective function of the decomposition-based multi-objective optimization algorithm. The data acquisition module is used to collect key data of each wind turbine in real time and receive the field-level active power target value issued by the wind farm-side active power automatic generation control system. The real-time theoretical power calculation module is used to obtain the real-time theoretical power of each wind turbine based on the key data of each wind turbine collected in real time and the configured unit operating parameters. The control trigger judgment module is used to determine whether the main control module should be triggered; The judgment process includes: The target value of active power at the field level is compared with the target value of active power at the field level issued in the previous control cycle. If the change exceeds the preset threshold, or if the deviation between the actual active power value obtained by the periodic monitoring based on the dynamic algorithm and the target value of active power at the field level meets the control triggering condition, the main control module is triggered. Otherwise, the main control module is not triggered, and each wind turbine continues to maintain its current state. The main control module is used to employ the decomposition-based multi-objective optimization algorithm, taking the real-time theoretical power and key data of each wind turbine as algorithm input. Within the preset total iteration time and total number of iterations, the multi-objective optimization problem is decomposed into several single-objective sub-problems, which are solved iteratively through neighborhood relationships and co-evolution. After the iteration terminates, the obtained solution set is sorted by non-dominated solution, dominant solutions are removed, and a Pareto front is constructed. The quality of the non-dominated solutions in the Pareto front is evaluated. When the evaluation result meets the preset control requirements, the non-dominated solutions are decoded into the active power target value of each wind turbine and sent to the corresponding turbine for execution. The backup control module is used to call a pre-stored PI control algorithm to perform fallback control on the wind turbine when the iteration time of the decomposition-based multi-objective optimization algorithm reaches the total iteration time limit or the number of iterations reaches the total number of iterations limit, or when it is determined that the non-dominated solution does not meet the preset control requirements.

Images