An identification control and detection method for an offshore wind power virtual inertia frequency modulation device
By using the bat algorithm to optimize the Hammerstein model in the virtual inertia control system of offshore wind power, we can optimize the NFM parameters and linear dynamic link parameters, and solve the problems of low control accuracy and low convergence of identification parameters in the existing technology, achieving more efficient frequency Adjustment and active power control improve the system's anti-interference ability and control accuracy.
Patent Information
- Application Number
- CN202111428370.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-29
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2041-11-29
AI Technical Summary
When facing noise and disturbance, the existing offshore wind power virtual inertia control system has low control accuracy and is prone to fall into the local optimal solution. The identification parameters of the least squares method are low, making it difficult to achieve effective frequency adjustment in large interference environments. Active power control.
The Hammerstein model based on the bat algorithm is adopted, and NFM parameters and linear dynamic link parameters are optimized through the arma algorithm, combined with the bat optimization algorithm, the fitting and parameter identification of nonlinear links is achieved, limitations are avoided, and model accuracy and control effect are improved.
More accurate system parameter identification and control is achieved in a heavy noise environment, improving the anti-interference ability and control accuracy of virtual inertia frequency regulation of offshore wind power, and ensuring the stable operation of the system under large interference.
Smart Images

Figure CN114094602B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to an offshore wind power control device, and in particular to an identification control and detection method for an offshore wind power virtual inertia frequency modulation device. Background Art
[0002] In offshore wind power systems, the doubly-fed induction motors currently used are mainly divided into three parts: mechanical, electrical and control. When a disturbance occurs in the power grid, the doubly-fed motor system can quickly change the speed by quickly changing the rotor side frequency, making full use of the rotor kinetic energy to release or absorb energy and compensate for the grid disturbance.
[0003] At present, the primary frequency regulation of wind power systems mainly focuses on power reserve control and rotor kinetic energy control. Offshore wind turbines can effectively participate in system frequency regulation. Virtual inertia control is mainly achieved by designing additional frequency response modules to absorb or release the rotational kinetic energy of wind turbines. Existing virtual inertia research mainly focuses on frequency regulation. Inertial response temporarily releases the kinetic energy stored in the rotating mass to prevent the frequency from dropping in the initial stage of the disturbance.
[0004] In actual industrial processes, the Hammerstein model can be used to transform the nonlinear system predictive control problem into the linear system control problem. On this basis, the bat algorithm, as a heuristic algorithm with global optimization capabilities, does not rely on specific statistical characteristics and is an effective means to deal with such complex problems. Virtual inertia control based on the Hammerstein model can solve the frequency support problem of microgrids, thereby adjusting and predicting the frequency of the power grid within the allowable range of variation and controlling the output of active power. However, in the actual doubly fed induction generator system operation process, there are still large noises and disturbances, large nonlinear effects, and the control quantity output by the general controller will produce large spikes.
[0005] In addition, the identification parameters of the Hammerstien model were originally identified using the least squares method, but the least squares method has low convergence and low precision, and may fall into a local optimal solution, which greatly reduces the performance. The DFIG wind power generation system has strong nonlinearity, large disturbances, and measurement noise in the system. Summary of the invention
[0006] Purpose of the invention: The purpose of the present invention is to provide an identification and control method for an offshore wind power virtual inertia frequency modulation device with strong anti-interference ability while retaining the optimal control rate and anti-interference ability, as well as a detection method with higher detection accuracy and more efficiency.
[0007] Technical solution: The identification and control method of the offshore wind power virtual inertia frequency modulation device of the present invention establishes a mathematical model of the virtual inertia controller based on the Hammerstein model of the bat algorithm. In the Hammerstein model, the method of using the bat algorithm to simultaneously optimize the NFM parameters and the linear dynamic link parameters is: using the minimum value method of the arma algorithm, the objective function of the bat algorithm is obtained as follows:
[0008]
[0009] In the above formula, is the estimated value of Φ(b) at the kth iteration, For the best position of each bat at the kth iteration, the best position of the global bat is:
[0010]
[0011] According to the constraints and fitness function indicators, the optimal solution under the current number of iterations is obtained, and the above Hammerstein model is normalized.
[0012] Through the above technical solution, when ARMA performs parameter optimization, bats are combined to enable more accurate identification in heavy noise environments. By fitting the nonlinear links through NFM, the limitations of the algorithm are avoided and the overall optimal solution is achieved.
[0013] Specifically, in this scheme, a large amount of system input and output data is first obtained in real time and sent to the Hammerstein model, the parameters to be identified in the original algorithm are sent to the position to be updated of the bat optimization algorithm, the scale, parameter dimension, average loudness, population frequency, initial speed and position of the bat algorithm are initialized, and then the data is output through the Hammerstein model, the deviation value of the current fitness function of the bat is calculated, the speed and position of the bat, the optimal position of each bat algorithm and the optimal position of the entire group are updated, and a random number is generated to compare the current fitness value with its individual historical best fitness value. Subsequently, according to the loudness and fitness value of the population sound, the global best fitness and its location are judged and updated, and the corresponding value of the best fitness value is assigned to the corresponding parameter. Finally, the algorithm is completed by satisfying the conditions, and the minimum coefficient of the least squares method required for the global optimal position is output, which solves the problem of the least squares method in the Hammerstein model, and also enables the model to be better used to predict the system output, so as to achieve the purpose of obtaining a better control effect under large interference.
[0014] Preferably, the Hammerstein expression is:
[0015]
[0016] In the above formula, w i is the weight of NFM, u(k) and x(k) are the input and output of the nonlinear part respectively. Therefore, the linear dynamic link can be expressed as:
[0017]
[0018] in, In the above formula, y(k) is the output of the model, x(k) is the input of the linear part, and d is the system delay.
[0019] Preferably, the identification control method comprises the following steps:
[0020] (1) In an offshore wind power system, when a frequency event occurs in the power grid, the model collects power grid frequency and active power data in real time;
[0021] (2) Input the collected sampled data into the Hammerstein model, use NFM to fit the nonlinear link, and use the bat algorithm to optimize the NFM parameters of the nonlinear link and the dynamic parameters of the linear link at the same time. The optimal parameters are used as the mathematical model system parameters to obtain the optimal transfer function.
[0022] (3) Detect the frequency prediction results, calculate the system active power prediction value through the frequency prediction value, and control the frequency support of the doubly-fed induction motor to the power grid according to the identified Hammerstein model.
[0023] On the whole, the model identification control method constructs an auxiliary model, sets the parameters of the auxiliary model as the position vector of the individual bat, sets the error between the output of the assignment model and the real system as the fitness value of the bat, and minimizes the output error through the position iteration of the bat algorithm. At this time, the position vector of the optimal individual is the system parameter to be identified.
[0024] Preferably, in step (1), the model collects grid frequency and active power data in real time, inertial support provided by the doubly-fed induction motor when a frequency event occurs, and outputs corresponding active power in real time according to the frequency change of the system.
[0025] A detection method for an offshore wind power virtual inertia frequency modulation device adopts the above-mentioned Hammerstein model instead of a fuzzy logic controller.
[0026] Preferably, the per-unit value of the rotor speed is set to 0.7pu. When a frequency event occurs in the power grid, the rotor speed range comparator is used to control the power grid frequency support. The system frequency modulation status is reflected by detecting the actual value of the rotor speed. If the system rotor speed is lower than the rated speed, the system exits the frequency modulation. If the system rotor speed is higher than the rated speed, the system starts the frequency modulation until the system frequency becomes stable and then exits the frequency modulation.
[0027] Generally, the stability of the doubly-fed induction motor is related to the size of the frequency control coefficient. If the frequency coefficient is too large or too small, the doubly-fed induction motor will be unstable, and the wind power system will have a power event, causing system instability. Therefore, in order to eliminate the large deceleration of the rotor speed, the moment of inertia control is activated only when the rotor speed exceeds the per-unit value. Therefore, a rotor speed range comparator is set here. When the actual rotor speed is less than the minimum speed of 0.7, the output signal of the range comparator is zero, and the system exits the frequency modulation.
[0028] Preferably, a DSP switching system controller is adopted, and a DSP digital signal controller is used to detect the system frequency modulation working condition, which can greatly improve the frequency modulation efficiency and make the system run more smoothly.
[0029] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: 1. By optimizing the least squares method based on the bat algorithm and using NFM to fit the nonlinear link, the nonlinear link can be effectively avoided from being difficult to parameterize; 2. By simultaneously optimizing the NFM parameters and the parameters of the linear dynamic link using the bat optimization algorithm, a more accurate controller identification model can be obtained, thereby improving the model accuracy; 3. The rotor speed range comparator is used to monitor the frequency modulation device, and the monitoring effect is more accurate. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 It is a schematic diagram of the structure of the Hammerstein model in the present invention;
[0031] Figure 2 This is a flowchart of the bat optimization algorithm in the present invention;
[0032] Figure 3 This is the frequency modulation control diagram of the rotor speed control system in the present invention;
[0033] Figure 4 It is a schematic diagram of the structure of the frequency modulation hardware controller using different rotation speeds in the present invention. DETAILED DESCRIPTION
[0034] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.
[0035] The identification and control method of the offshore wind power virtual inertia frequency modulation device comprises the following steps:
[0036] (1) In an offshore wind power system, when a frequency event occurs in the power grid, the model collects the power grid frequency and active power data in real time, wherein the model collects the power grid frequency and active power data in real time, the inertial support provided by the doubly-fed induction motor when a frequency event occurs, and outputs the corresponding active power in real time according to the frequency change of the system;
[0037] (2) Input the collected sampled data into the Hammerstein model, use NFM to fit the nonlinear link, and use the bat algorithm to optimize the NFM parameters of the nonlinear link and the dynamic parameters of the linear link at the same time. The optimal parameters are used as the mathematical model system parameters to obtain the optimal transfer function.
[0038] (3) Detect the frequency prediction results, calculate the system active power prediction value through the frequency prediction value, and control the frequency support of the doubly-fed induction motor to the power grid according to the identified Hammerstein model.
[0039] In the above steps, the mathematical model of the virtual inertia controller is established by the Hammerstein model based on the bat algorithm. In the Hammerstein model, the method of using the bat algorithm to simultaneously optimize the NFM parameters and the linear dynamic link parameters is: using the minimum value method of the arma algorithm, the objective function of the bat algorithm is obtained as follows:
[0040]
[0041] In the above formula, is the estimated value of Φ(b) at the kth iteration, For the best position of each bat at the kth iteration, the best position of the global bat is:
[0042]
[0043] According to the constraints and fitness function indicators, the optimal solution under the current number of iterations is obtained, and the above Hammerstein model is normalized.
[0044] At the same time, the Hammerstein expression is:
[0045]
[0046] In the above formula, w i is the weight of NFM, u(k) and x(k) are the input and output of the nonlinear part respectively. Therefore, the linear dynamic link can be expressed as:
[0047]
[0048] in, In the above formula, y(k) is the output of the model, x(k) is the input of the linear part, and d is the system delay.
[0049] like Figure 1 , 2 As shown in the figure, this figure is a schematic diagram of the structure of the Hammerstein model in this scheme. Figure 2 This is the flowchart of the bat optimization algorithm in this scheme.
[0050] In this scheme, a large amount of system input and output data is first obtained in real time and sent to the Hammerstein model, the parameters to be identified in the original algorithm are sent to the position to be updated of the bat optimization algorithm, the scale, parameter dimension, average loudness, population frequency, initial speed and position of the bat algorithm are initialized, and then the data is output through the Hammerstein model, the deviation value of the current fitness function of the bat is calculated, the speed and position of the bat, the optimal position of each bat algorithm and the optimal position of the entire group are updated, and a random number is generated to compare the current fitness value with its individual historical best fitness value. Subsequently, according to the loudness and fitness value of the population sound, the global optimal fitness and its locked position are judged and updated, and the corresponding value of the optimal fitness value is assigned to the corresponding parameter. Finally, the algorithm is completed by satisfying the conditions, and the minimum coefficient of the least squares method required for the global optimal position is output, which solves the problem of the least squares method in the Hammerstein model, and also enables the model to be better used to predict the system output, so as to achieve the purpose of obtaining a better control effect under large interference.
[0051] like Figure 3 As shown, this figure is the frequency control diagram of the rotor speed control system in this scheme. The detection method of a virtual inertia frequency regulation device for offshore wind power first uses the above-mentioned Hammerstein model to replace the fuzzy logic controller, and then sets the per-unit value of the rotor speed to 0.7pu. When a frequency event occurs in the power grid, the rotor speed range comparator is used to control the power grid frequency support. The actual value of the rotor speed is detected to reflect the system frequency regulation status. If the system rotor speed is lower than the rated speed, the system exits the frequency regulation. If the system rotor speed is higher than the rated speed, the system starts the frequency regulation until the system frequency reaches stability and then exits the frequency regulation. A DSP digital signal controller can be used to detect the system frequency regulation condition, which can greatly improve the efficiency of frequency regulation and make the system run more smoothly.
[0052] In addition, if Figure 4As shown in the figure, this is a schematic diagram of the frequency modulation hardware controller structure using different speed conditions. The DI port of the controller is connected to the output port of the speed range comparator, and the DO port is connected to the active power of the frequency modulation system. When a frequency event occurs in the system, the DSP digital signal processor is controlled by the size 0 or 1 of the speed range comparator to control the on and off of the system frequency modulation mode.
Claims
1. A method for identifying and controlling an offshore wind power virtual inertia frequency modulation device, characterized in that: (1) In an offshore wind power system, when a frequency event occurs in the power grid, the Hammerstein model collects the grid frequency and active power data in real time, the inertial support provided by the doubly-fed induction motor when a frequency event occurs, and outputs the corresponding active power in real time according to the frequency change of the system; the Hammerstein expression is: In the above formula, w i is the weight of NFM, u(k) and x(k) are the input and output of the nonlinear part respectively. Therefore, the linear dynamic link can be expressed as: in, In the above formula, y(k) is the output of the model, x(k) is the input of the linear part, and d is the system delay; (2) Input the collected sampled data into the Hammerstein model, use NFM to fit the nonlinear link, and use the bat algorithm to optimize the NFM parameters of the nonlinear link and the dynamic parameters of the linear link at the same time. The optimal parameters are used as the mathematical model system parameters to obtain the optimal transfer function. (3) Detect the frequency prediction results, calculate the system active power prediction value through the frequency prediction value, and control the frequency support of the doubly-fed induction motor to the power grid according to the identified Hammerstein model; Among them, the mathematical model of the virtual inertia controller is established by the Hammerstein model based on the bat algorithm. In the Hammerstein model, the method of using the bat algorithm to simultaneously optimize the NFM parameters and the linear dynamic link parameters is: using the minimum value method of the arma algorithm, the objective function of the bat algorithm is obtained as follows: In the above formula, is the estimated value of Φ(b) at the kth iteration, For the best position of each bat at the kth iteration, the best position of the global bat is: According to the constraints and fitness function indicators, the optimal solution under the current number of iterations is obtained, and the above Hammerstein model is normalized.
2. A detection method for an offshore wind power virtual inertia frequency modulation device, characterized in that: The Hammerstein model described in claim 1 above is used to replace the fuzzy logic controller; then the per-unit value of the rotor speed is set to 0.7pu. When a frequency event occurs in the power grid, the rotor speed range comparator is used to control the power grid frequency support. The system frequency modulation status is reflected by detecting the actual value of the rotor speed. If the system rotor speed is lower than the rated speed, the system exits the frequency modulation. If the system rotor speed is higher than the rated speed, the system starts the frequency modulation until the system frequency reaches stability and then exits the frequency modulation.
3. The detection method of the offshore wind power virtual inertia frequency modulation device according to claim 2 is characterized in that: Adopt DSP switching system controller.
