Improved particle swarm optimization-based parallel cooperative control method for network-forming type fans

Abstract

The invention discloses a network construction type fan parallel cooperative control method based on an improved particle swarm algorithm, and relates to the technical field of control, and the method specifically comprises the following steps: building a target function by improving the system frequency dynamic response during load step disturbance and reducing the output voltage deviation of a grid-side converter; iterative solution is carried out by using an improved particle swarm algorithm until a satisfactory fitness value is obtained or the maximum number of iterations is reached; introducing the optimized voltage deviation and angular frequency deviation into a control loop; and compensating the voltage drop caused by the added virtual impedance. According to the method, the optimization objective function is established, the frequency dynamic response capability of the system under load step disturbance is improved, the output voltage deviation of the grid-side converter is reduced, the control precision of the grid-forming converter is improved, and a new technical reference is provided for stable operation of a power system.

Description

Technical Field

[0001] The present invention belongs to the field of control technology, and in particular relates to a parallel collaborative control method of grid-type fans based on an improved particle swarm algorithm. Background Art

[0002] In the field of power systems, the parallel operation of grid-type wind turbines is one of the key technologies to improve power supply reliability and system flexibility. However, this mode of operation also brings a series of control challenges, especially when the load undergoes a step change, the system frequency dynamic response capability is insufficient, resulting in frequency fluctuations and affecting the stability of the power system. In addition, due to the mismatch of the line impedance at the outlet of the grid-side converter, reactive circulating current problems will occur when grid-type wind turbines are operated in parallel, which not only affects system performance, but may also cause voltage deviations, further affecting the power quality. Traditional control strategies have limitations in dealing with these challenges, especially when faced with problems caused by load disturbances and line impedance mismatch, and lack effective solutions.

[0003] Therefore, in order to solve these problems, a new control method is needed, which can not only improve the frequency dynamic response capability of the system, but also reduce the output voltage deviation of the converter, and at the same time ensure the stable operation of the power system. Summary of the invention

[0004] In order to solve the problems in the prior art, the present invention provides a parallel cooperative control method for grid-type wind turbines based on an improved particle swarm algorithm. By establishing an optimization objective function, the frequency dynamic response capability of the system under load step disturbance is improved, and the output voltage deviation of the grid-side converter is effectively reduced. This strategy uses an improved particle swarm algorithm for iterative solution, and finally introduces the optimized voltage deviation and frequency deviation into the control loop, while compensating for the voltage drop caused by the introduction of virtual impedance. This scheme not only improves the control accuracy of the grid-type converter, but also provides a new technical reference for the stable operation of the power system.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The present invention provides a parallel coordinated control method for grid-type fans based on an improved particle swarm algorithm, which specifically includes the following steps:

[0007] S1, establishing an objective function to improve the dynamic response of system frequency during load step disturbance and reduce the output voltage deviation of the grid-side converter;

[0008] S2, using the improved particle swarm algorithm to iteratively solve the problem until a satisfactory fitness value is obtained or the maximum number of iterations is reached;

[0009] S3, introducing the optimized voltage deviation and angular frequency deviation into the control loop;

[0010] S4. Compensate for the voltage drop caused by the added virtual impedance.

[0011] Preferably, in S1, the step of establishing the objective function includes:

[0012] S11. Assume that the system under study is a plurality of grid-type wind turbines of the same capacity connected in parallel. According to the impedance value of each grid-side converter outlet line, the initial virtual impedance is set respectively. The virtual impedance is used to balance the circulation problem caused by the mismatch of each line impedance.

[0013] Assume that the virtual inductance of the jth converter with the largest line impedance is L vj , then the virtual inductance of the i-th converter is

[0014] S12, the objective function is:

[0015]

[0016] Where: α, β, γ are weight coefficients, α+β+γ=1, Δω is the frequency deviation after considering load disturbance, ΔU is the output voltage deviation of each grid-side converter, N is the number of wind turbines in parallel, W j is the penalty function for node voltage exceeding the limit, and the frequency deviation Δω is

[0017]

[0018] Where: ΔP load is the load change, Δω i is the change of output angular frequency of each fan;

[0019] The voltage deviation ΔU is

[0020]

[0021] Where: U jmax , U jmin The upper and lower limits of voltage are

[0022] Penalty function W for node voltage exceeding the limit j for

[0023]

[0024] Preferably, in S2, the operation steps of the improved particle swarm algorithm are as follows:

[0025] S21, Initialization: Set the initial position and velocity of the particle swarm, as well as the parameters of the algorithm, including ω max ,ω min 、c min 、c max, T;

[0026] S22, iterative process: in each iteration, the inertia weight and learning factor are dynamically calculated;

[0027] S23, update particle state: use the updated parameters to adjust the particle speed and position according to the standard update rule of the PSO algorithm;

[0028] S24, evaluation and update: evaluate the fitness of each particle and update the individual and global optimal solutions according to the fitness;

[0029] S25, convergence judgment: check whether the convergence conditions are met, such as reaching the maximum number of iterations or the fitness reaching a preset threshold. If so, stop the iteration; otherwise, return to step 22 to continue the iteration.

[0030] Preferably, the updating formula of the particle swarm velocity in the particle swarm algorithm is:

[0031] V id (t+1)=ωV id (t)+c 1 r 1 [P bestd (t)-X id (t)]+c 2 r 2 [P optd (t)-X id (t)]

[0032] Where: V id (t) is the velocity value of the i-th particle in the d-th dimension, X id (t) is the position of the ith particle in the dth dimension and the tth generation, P bestd (t) is the optimal individual position of the tth generation of the dth dimension of the ith particle, P optd (t) is the global optimal position of the tth generation of the particle swarm in the dth dimension, w is the inertia weight of the particle, c 1 is the weight coefficient of the particle tracking its own historical optimal value, c 2 is the weight coefficient of the particle with the best historical quality, r 1 and r 2 is a random number between [0,1].

[0033] Preferably, the update formula of the particle swarm position in the particle swarm algorithm is:

[0034] X id (t+1)=X id (t)+V id (t+1)

[0035] The formula for adaptive inertia weight in particle swarm optimization is:

[0036]

[0037] Where: min With ω max are the minimum and maximum values ​​of the inertia weight respectively; f n is the fitness of the particle at the nth iteration; and are the minimum and maximum fitness values ​​of all particles at the nth iteration, respectively; t and T represent the current iteration number and the maximum iteration number, respectively.

[0038] Preferably, the learning factor formula in the particle swarm algorithm is:

[0039]

[0040] Where: c min With c max Represent the maximum and minimum values ​​of the learning factor respectively.

[0041] Preferably, the specific operation steps of introducing the optimized voltage deviation and angular frequency deviation into the control loop in S3 are as follows:

[0042] S31, introducing the voltage deviation into the virtual impedance link, so as to dynamically adjust the value of the virtual impedance according to the current voltage deviation, thereby reducing the voltage deviation output by the grid-side converter and suppressing the circulating current,

[0043] The dynamic virtual impedance is expressed as:

[0044] Where: X set is the static virtual impedance value set initially; k v is the gain coefficient;

[0045] S32, introducing the angular frequency deviation into the active loop of the virtual synchronous control link, and timely compensating to suppress the drop of transient frequency, thereby improving the frequency dynamic response of the grid connection point.

[0046] Preferably, the specific operation method for compensating the voltage drop caused by the added virtual impedance in step S4 is as follows:

[0047] The voltage drop caused by the virtual impedance is introduced into the reactive loop of the virtual synchronous control link, and Voltage compensation is performed on each converter.

[0048] Where: Q ref is the reactive reference power of the grid-side converter; U is the voltage value output by the converter.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] (1) The present invention improves the particle swarm algorithm by introducing adaptive inertia weight and dynamic adjustment parameters, thereby improving the inertia weight, optimizing the learning factor, and improving the optimization ability of the entire algorithm. This can solve the problem that the standard particle swarm algorithm cannot take into account both optimization speed and optimization accuracy.

[0051] (2) Frequency compensation is used to suppress transient frequency drops, thereby improving the system's frequency dynamic response and solving the problem of a large transient drop in the grid-connected point frequency during the inertia support stage due to load step disturbances.

[0052] (3) Impedance matching is achieved by introducing dynamic virtual impedance, circulating current is suppressed, and voltage compensation is performed to improve the voltage quality of the grid connection point. The problem of reactive circulating current occurring when grid-connected wind turbines are connected in parallel due to impedance mismatch of the grid-side converter outlet line is solved, which affects the performance of the system.

[0053] The method of the present application can significantly improve the frequency dynamic response of the grid connection point, reduce the circulating current caused by line impedance mismatch, and thus improve the power quality and stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a system topology diagram of a grid-type wind turbine parallel cooperative control method based on an improved particle swarm algorithm of the present invention;

[0055] Figure 2 To improve the particle swarm algorithm flow chart;

[0056] Figure 3 This is a control block diagram of a parallel coordinated control method of meshed fans based on an improved particle swarm algorithm of the present invention;

[0057] Figure 4 This is a comparison chart of the simulation results of circulation suppression when two grid-type wind turbines are connected in parallel;

[0058] Figure 5 This is a comparison chart of the frequency compensation simulation results when two grid-type wind turbines are connected in parallel;

[0059] Figure 6 This is a comparison chart of the voltage compensation simulation results when two grid-type wind turbines are connected in parallel. DETAILED DESCRIPTION

[0060] The technical solutions in the embodiments of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0061] This embodiment provides a parallel coordinated control method for grid-type wind turbines based on an improved particle swarm algorithm, comprising the following steps:

[0062] S1. Establishing the objective function to improve the dynamic response of the system frequency and reduce the output voltage deviation of the grid-side converter during load step disturbance. The specific steps are as follows:

[0063] First, assume that the system under study is a parallel connection of multiple grid-type wind turbines of the same capacity. First, according to the impedance value of the outlet line of each grid-side converter, the initial virtual impedance is set respectively. This virtual impedance is mainly to balance the circulating current problem caused by the mismatch of the line impedance. Assume that the virtual inductance of the jth converter with the largest line impedance is L vj , then the virtual inductance of the i-th converter is

[0064] Secondly, in order to effectively suppress the system frequency fluctuation caused by the parallel connection of grid-type wind turbines and improve the stability of the power grid, the objective function is to minimize the frequency deviation after load disturbance; in order to suppress the circulating current problem caused by the impedance mismatch of each line, the objective function is to minimize the outlet voltage deviation of each grid-side converter; considering the possible node voltage exceeding the limit during the iteration process, a penalty function for node voltage exceeding the limit is introduced.

[0065] Finally, based on the above, the objective function is obtained:

[0066]

[0067] Where: α, β, γ are weight coefficients, α+β+γ=1, Δω is the frequency deviation after considering load disturbance, ΔU is the voltage deviation of each grid-side converter outlet, N is the number of wind turbines in parallel, W j is the penalty function for node voltage exceeding the limit.

[0068] The frequency deviation Δω is

[0069]

[0070] Where: ΔP load is the load change, Δω i is the change of fan output angular frequency.

[0071] The voltage deviation ΔU is

[0072]

[0073] Where: U jmax , U jmin The upper and lower limits of voltage are

[0074] Penalty function W for node voltage exceeding the limit j for

[0075]

[0076] S2. The specific operation steps of improving the particle swarm algorithm are as follows:

[0077] according to Figure 2 The improved particle swarm algorithm flowchart is iteratively solved. The particle swarm velocity and position update formulas and the adaptive inertia weight and learning factor formulas required in the iterative process are as follows:

[0078] The update formula of particle swarm velocity in particle swarm algorithm is:

[0079] V id (t+1)=ωV id (t)+c 1 r 1 [P bestd (t)-X id (t)]+c 2 r 2 [P optd (t)-X id (t)]

[0080] Where: V id (t) is the velocity value of the i-th particle in the d-th dimension, X id (t) is the position of the ith particle in the dth dimension and the tth generation, P bestd (t) is the optimal individual position of the tth generation of the dth dimension of the ith particle, P optd (t) is the global optimal position of the tth generation of the particle swarm in the dth dimension, w is the inertia weight of the particle, c 1 is the weight coefficient of the particle tracking its own historical optimal value, c 2 is the weight coefficient of the particle with the best historical quality, r 1 and r 2 is a random number between [0,1].

[0081] The update formula of particle swarm position in particle swarm algorithm is:

[0082] X id (t+1)=X id (t)+V id (t+1)

[0083] The formula for adaptive inertia weight in particle swarm optimization is:

[0084]

[0085] Where: min With ω max are the minimum and maximum values ​​of the inertia weight respectively; f n is the fitness of the particle at the nth iteration; and are the minimum and maximum fitness values ​​of all particles at the nth iteration, respectively; t and T represent the current iteration number and the maximum iteration number, respectively.

[0086] The learning factor formula in the particle swarm algorithm is:

[0087]

[0088] Where: c min With c max Represent the maximum and minimum values ​​of the learning factor respectively.

[0089] S3. The specific process of introducing the optimized voltage deviation and angular frequency deviation into the control loop is as follows:

[0090] Firstly, the iterated voltage deviation is introduced into the virtual impedance link, so as to dynamically adjust the value of the virtual impedance according to the current voltage deviation, thereby reducing the voltage deviation output by the grid-side converter and suppressing the circulating current.

[0091] The dynamic virtual impedance is expressed as:

[0092] Where: X set is the static virtual impedance value set initially; k v is the gain coefficient;

[0093] like Figure 4 As shown, the reactive circulating current of the traditional control strategy is large. The control method of the present invention can significantly suppress the circulating current, verifying the effectiveness of the control method and improving the performance of the system. In order to illustrate the principle of the strategy adjustment, it is assumed that the output voltage of the grid-side converter is lower than the average value. According to the above formula, a value of increasing the virtual impedance correction term is generated. The increase of the correction term will reduce the adaptive virtual impedance value, thereby reducing the equivalent line impedance of the converter. The reduction of the equivalent line impedance will reduce the voltage drop, thereby increasing the output voltage of the converter, and finally achieving the consistency of the output voltage of each grid-side converter, thereby suppressing the circulating current.

[0094] Secondly, the angular frequency deviation is introduced into the active loop of the virtual synchronous control link, such as Figure 3 As shown, timely compensation is used to suppress the drop of transient frequency, thereby improving the frequency dynamic response of the grid connection point.

[0095] like Figure 5 As shown in the figure, when the load step increases suddenly, the frequency of the traditional control strategy drops by about 0.2Hz, and the transient adjustment time is longer; the frequency of the improved control strategy proposed in this paper drops by about 0.05Hz, and the transient adjustment time is shorter. The simulation results show that compared with the traditional control strategy, the control method proposed in this invention can shorten the transient response time of the system and improve the robustness of the system under transient conditions.

[0096] S4. The specific process of compensating the voltage drop caused by the added virtual impedance is as follows:

[0097] The voltage drop caused by the virtual impedance is introduced into the reactive loop of the virtual synchronous control link, such as Figure 3 As shown, and in accordance with Perform voltage compensation on each grid-side converter.

[0098] like Figure 6 As shown in the figure, when virtual impedance is added at the 1st second, if the voltage compensation link in the fourth step is not performed, the voltage at the grid connection point will have a voltage drop of about 13V, while after voltage compensation, the voltage at the grid connection point will only have a voltage drop of about 3V. The simulation results show that the voltage compensation method proposed in the present invention improves the power quality at the end of the line and ensures the robustness of the output voltage of the grid-side converter.

[0099] Where: Q ref is the reactive reference power of the grid-side converter; U is the voltage value output by the converter.

[0100] This embodiment improves the frequency dynamic response capability of the system under load step disturbance and effectively reduces the output voltage deviation of the grid-side converter by establishing an optimization objective function. This strategy iterates and solves the problem through an improved particle swarm algorithm, and finally introduces the optimized voltage deviation and frequency deviation into the control loop, while compensating for the voltage drop caused by the introduction of virtual impedance. This solution not only improves the control accuracy of the grid-type converter, but also provides a new technical reference for the stable operation of the power system.

[0101] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A parallel coordinated control method for grid-type wind turbines based on an improved particle swarm algorithm, characterized in that: The specific steps include: S1, establishing an objective function to improve the dynamic response of system frequency during load step disturbance and reduce the output voltage deviation of the grid-side converter; S2, using the improved particle swarm algorithm to iteratively solve the problem until a satisfactory fitness value is obtained or the maximum number of iterations is reached; S3, introducing the optimized voltage deviation and angular frequency deviation into the control loop; S4. Compensate for the voltage drop caused by the added virtual impedance.

2. According to claim 1, a method for parallel coordinated control of grid-type wind turbines based on improved particle swarm algorithm is characterized in that: In S1, the step of establishing the objective function includes: S11. Assume that the system under study is a plurality of grid-type wind turbines of the same capacity connected in parallel. According to the impedance value of each grid-side converter outlet line, the initial virtual impedance is set respectively. The virtual impedance is used to balance the circulation problem caused by the mismatch of each line impedance. Assume that the virtual inductance of the jth converter with the largest line impedance is L vj , then the virtual inductance of the i-th converter is S12, the objective function is: Where: α, β, γ are weight coefficients, α+β+γ=1, Δω is the frequency deviation after considering load disturbance, ΔU is the output voltage deviation of each grid-side converter, N is the number of wind turbines in parallel, W j is the penalty function for node voltage exceeding the limit, and the frequency deviation Δω is Where: ΔP load is the load change, Δω i is the change of output angular frequency of each fan; The voltage deviation ΔU is Where: U jmax , U jmin The upper and lower limits of voltage are Penalty function W for node voltage exceeding the limit j for 3. The method for parallel coordinated control of grid-type wind turbines based on improved particle swarm algorithm according to claim 1 is characterized in that: In S2, the operation steps of the improved particle swarm algorithm are as follows: S21, Initialization: Set the initial position and velocity of the particle swarm, as well as the parameters of the algorithm, including ω max ,ω min 、c min 、c max , T; S22, iterative process: in each iteration, the inertia weight and learning factor are dynamically calculated; S23, update particle state: use the updated parameters to adjust the particle speed and position according to the standard update rule of the PSO algorithm; S24, evaluation and update: evaluate the fitness of each particle and update the individual and global optimal solutions according to the fitness; S25, convergence judgment: check whether the convergence conditions are met, such as reaching the maximum number of iterations or the fitness reaching a preset threshold. If so, stop the iteration; otherwise, return to step 22 to continue the iteration.

4. The method for parallel coordinated control of grid-type wind turbines based on improved particle swarm algorithm according to claim 3 is characterized in that: The updating formula of particle swarm velocity in the particle swarm algorithm is: V id (t+1)=ωV id (t)+c1r1[P bestd (t)-X id (t)]+c2r2[P optd (t)-X id (t)] Where: V id (t) is the velocity value of the i-th particle in the d-th dimension, X id (t) is the position of the ith particle in the dth dimension and the tth generation, P bestd (t) is the optimal individual position of the tth generation of the dth dimension of the ith particle, P optd (t) is the global optimal position of the tth generation of the particle swarm in the dth dimension, w is the inertia weight of the particle, c1 is the weight coefficient of the particle tracking its own historical optimal value, c2 is the weight coefficient of the particle for the historical best quality, and r1 and r2 are random numbers between [0,1].

5. The method for parallel coordinated control of grid-type wind turbines based on improved particle swarm algorithm according to claim 3 is characterized in that: The updating formula of the particle swarm position in the particle swarm algorithm is: X id (t+1)=X id (t)+V id (t+1) The formula for adaptive inertia weight in particle swarm optimization is: Where: min With ω max are the minimum and maximum values ​​of the inertia weight respectively; f n is the fitness of the particle at the nth iteration; and are the minimum and maximum fitness values ​​of all particles at the nth iteration, respectively; t and T represent the current iteration number and the maximum iteration number, respectively.

6. The method for parallel coordinated control of grid-type wind turbines based on improved particle swarm algorithm according to claim 3 is characterized in that: The learning factor formula in the particle swarm algorithm is: Where: c min With c max Represent the maximum and minimum values ​​of the learning factor respectively.

7. The method for parallel coordinated control of grid-type wind turbines based on improved particle swarm algorithm according to claim 1, characterized in that: The specific operation steps of introducing the optimized voltage deviation and angular frequency deviation into the control loop in S3 are as follows: S31, introducing the voltage deviation into the virtual impedance link, so as to dynamically adjust the value of the virtual impedance according to the current voltage deviation, thereby reducing the voltage deviation output by the grid-side converter and suppressing the circulating current, The dynamic virtual impedance is expressed as: Where: X set is the static virtual impedance value set initially; k v is the gain coefficient; S32, introducing the angular frequency deviation into the active loop of the virtual synchronous control link, and timely compensating to suppress the drop of transient frequency, thereby improving the frequency dynamic response of the grid connection point.

8. The method for parallel coordinated control of grid-type wind turbines based on improved particle swarm algorithm according to claim 1 is characterized in that: The specific operation method of compensating the voltage drop caused by the added virtual impedance in step S4 is as follows: The voltage drop caused by the virtual impedance is introduced into the reactive loop of the virtual synchronous control link, and Voltage compensation is performed on each converter. Where: Q ref is the reactive reference power of the grid-side converter; U is the voltage value output by the converter.

Images