Doubly-fed fan control parameter identification method and system based on improved black hole algorithm

Abstract

The invention provides a double-fed fan control parameter identification method and system based on an improved black hole algorithm, and relates to the technical field of new energy power generation parameter identification. Comprising the following steps: collecting system output observed quantity of a fan during a fault period as an initial data set, building a doubly-fed fan simulation model, determining key control parameters needing to be identified under a fault working condition, calculating track sensitivity of the parameters to be identified and the observed quantity output by the system, and determining optimal observed quantity; then an initial value is obtained by using a least square method in combination with RT-LAB measured data, and control parameter identification and optimization are carried out based on an improved black hole algorithm to search for an optimal control parameter, so that a locally optimal solution is effectively avoided, and a globally optimal or approximately optimal control parameter solution is provided, thereby reducing the calculation amount, saving the time and improving the control efficiency. The search efficiency and the solution quality are improved, the seeking efficiency of the optimal control parameters is improved, and the system stability is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of parameter identification of renewable energy power generation, and in particular to a control parameter identification method and system for a doubly-fed wind turbine based on an improved black hole algorithm. Background Art

[0002] Wind energy occupies an increasingly important proportion in the electric energy system. Doubly-fed wind turbines (DFIGs) have the advantages of wide speed regulation range and low price, and have become the main type of wind power generation. Wind turbines with intermittent and random characteristics are connected to the power grid in large quantities through nonlinear power electronic converters, which brings huge challenges to the safe and stable operation of the power system. In order to accurately analyze the impact of large-scale wind power on the power grid, it is necessary to establish an accurate wind turbine model. The accuracy of the wind turbine model mainly depends on the accuracy of the parameters. The manufacturer will not directly provide the control parameters and the control parameters have the "black box" attribute. To obtain the control parameters, identification methods are usually used. Based on this, in order to solve the problems existing in the existing parameter identification research of doubly-fed wind turbine control systems, a parameter identification method for doubly-fed wind turbine controllers based on an improved black hole algorithm is proposed, which effectively improves the convergence speed and accuracy of the parameter identification of doubly-fed wind turbine control systems. Summary of the invention

[0003] The main purpose of the present invention is to provide a control parameter identification method and system for a doubly fed wind turbine based on an improved black hole algorithm, solve the problem of a doubly fed wind turbine model established in PLECS (Piecewise Linear Electrical Circuit Simulation) based on a traditional black hole algorithm, and propose a method for identifying the control parameters of a doubly fed wind turbine model, which can accurately and reliably identify the key control parameters in the doubly fed wind turbine wind power generation system model.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is: a control parameter identification method of a doubly-fed wind turbine based on an improved black hole algorithm, comprising the following steps: S1: Collect the observed values ​​of the system output during the fault period of the wind turbine and save them as the initial data set. The observed values ​​include the active power and reactive power on the grid side. S2: Build a simulation model of a doubly-fed wind turbine according to the preset typical values ​​of control parameters, and determine the parameters to be identified in the initial data set; S3: Use trajectory sensitivity as an indicator to analyze the correlation between the parameters to be identified and the observations to be selected, determine the observation with the highest sensitivity as the optimal observation, and determine the identifiability of each parameter to be identified; S4: Using the least square method combined with the measured data under the RT-LAB fault condition, the initial values ​​of the rotor-side converter PI control parameters are obtained to determine the optimization interval; S5: According to the optimization interval in S4, based on the data obtained by simulating the improved black hole algorithm and the doubly fed wind turbine simulation model, the control parameters are identified and adjusted, and it is continuously iterated until the fitness function meets the requirements and the identification effect is obtained.

[0005] In the preferred solution, in step S2, the rotor-side converter control equation of the doubly-fed wind turbine simulation model is expressed as: ; ; Where: The rotor current is Reference values ​​of axis components; are the reference values ​​of stator active power and reactive power respectively; are the active power and reactive power output by the stator of the doubly-fed induction generator respectively; is the slip, ; is the synchronous angular velocity of the motor; is the rotor angular velocity; are the dq axis components of the rotor voltage respectively; The rotor current is Axis component; is the self-inductance of the rotor winding; is the self-inductance of the stator winding; is the mutual inductance between the stator and rotor windings; is the magnetic flux leakage coefficient, ; is the stator voltage; are the converter control parameters respectively.

[0006] In the preferred solution, in step S3, obtaining the sensitivity between each parameter to be identified and the observed value of the system output includes: S301: The expression of trajectory sensitivity is:

[0007] In the formula, , For parameters The initial value of is the change of the corresponding parameter; is the selected observation; is the time t The trajectory sensitivity of the parameters; S302: Select the average value of trajectory sensitivity as the final criterion for measurement. The calculation formula for the average value of the absolute value of the trajectory sensitivity of each observation to the parameter to be identified is as follows: ; In the formula, is the average sensitivity; is the number of samples; S303: Analyze the average sensitivity calculation results to obtain the optimal observation quantity and sensitivity.

[0008] In the preferred embodiment, step S4 includes: S401: Use the least square method as the fitness function to quantify the black hole algorithm, and set the star with the optimal fitness value as a black hole. The formula is: ; Where: and The power output value obtained by the identification model and the actual power measurement value, n is the number of samples; S402: Establish the optimization interval and use it as the upper and lower limits of the black hole search space. The formula is: ; In the formula, is the solution in the search space, i.e. value; is the lower limit set, The upper limit is set; For interval Random numbers within; is the number of planets; .

[0009] In the preferred embodiment, step S5 specifically includes: S501: Use Logistic chaos mapping to initialize the planet positions, the formula is: ; Where: is the chaotic mapping parameter, is the solution in the search space; S502: Greedy retention strategy and Levy flight mechanism are introduced to improve the formula for updating the position of stars in the traditional black hole algorithm. The formula of the greedy retention strategy is: ; Where: is the position of the i-th planet after t iterations according to the greedy retention strategy; The Levy flight mechanism formula is: ; Where: is the parameter that obeys the normal distribution, that is . and They are ; Where: is the Gamma function; ; Finally, the planet position update is optimized as follows: ; Where: Obey for step length Random search vector for distribution ; is a vector operation; S503: Calculate and compare fitness, determine whether the black hole is replaced, and calculate the distance between the star and the black hole , When a new star is generated, the formula is: ; Where: m is the dimension of the star; is the jth component of the i-th star; K is a constant that adjusts the black hole absorption rate; is the fitness of the black hole; N is the total number of stellar variables; is the fitness value of the i-th planet; S504: If the convergence condition is met or the number of iterations is reached, the iteration is stopped and the optimal solution is output as the identification result.

[0010] The preferred solution also includes step S6: simulating and verifying the parameter identification model based on the improved black hole algorithm and comparing it with the traditional black hole algorithm and the particle swarm algorithm to determine its effectiveness, specifically: S601: Identify the PI control parameters of the rotor-side converter using the observed quantity, and perform simulation verification through the double-fed wind turbine simulation model built in S2; S602: A symmetrical three-phase short-circuit fault is set at the outlet of the fan to make the voltage drop to 50% of the initial value: a short-circuit fault is applied at t=1.5s, and the fault disappears at t=2s; the fan operation data between 0 and 3s is collected, and the wind speed is set to 10m / s. At this time, the fan works in the MPPT state; the IBHA algorithm is used to identify the parameters to be identified, the number of iterations is 50 times, the algorithm's optimization range is set to 0.5 to 1.5 times the initial identification result, and the average value of 100 identification results is taken; S603: In order to further verify the accuracy of parameter identification, the identified parameters are applied to the simulation model, and the active and reactive power curves obtained by simulation are compared with the curves of the original model; S604: Select the number of algorithm iterations for each optimization as 100, then perform 100 simulation tests, take the average value as the result of parameter identification, compare the obtained control parameter identification result with the control parameter identification results using the particle swarm algorithm and the traditional black hole algorithm, and obtain the objective function convergence curves of the three different optimization algorithms when solving the control parameters.

[0011] A control parameter identification system for a doubly-fed wind turbine based on an improved black hole algorithm includes the following modules: A data acquisition module is used to collect the observed values ​​of the system output of the wind turbine during the fault period and save them as the initial data set. The observed values ​​include the active power and reactive power on the grid-connected side. A model building module is used to build a simulation model of a doubly-fed wind turbine according to typical values ​​of preset control parameters and determine the parameters to be identified in the initial data set; A sensitivity calculation module is used to analyze the correlation between the parameters to be identified and the observations to be selected using trajectory sensitivity as an indicator, determine the observation with the highest sensitivity as the optimal observation, and determine the identifiability of each parameter to be identified; The optimization interval module is used to obtain the initial value of the rotor-side converter PI control parameter by using the least square method combined with the measured data under the RT-LAB fault condition, so as to determine the optimization interval; The identification module is used to identify and adjust the control parameters according to the optimization interval in the optimization interval module, based on the data obtained by simulating the improved black hole algorithm and the doubly fed wind turbine simulation model, and continuously iterate until the fitness function meets the requirements to obtain the identification effect.

[0012] The present invention provides a control parameter identification method for a doubly-fed wind turbine based on an improved black hole algorithm, which collects system output observations of the wind turbine during a fault period for parameter identification and optimization in subsequent steps, builds a simulation model of the doubly-fed wind turbine, determines key control parameters that need to be identified under fault conditions, and then calculates the trajectory sensitivity of the parameters to be identified and the system output to determine the optimal observations; then, the least squares method is used in combination with RT-LAB measured data to obtain initial values, and control parameter identification and optimization are performed based on the improved black hole algorithm to search for optimal control parameters, effectively avoiding local optimal solutions and providing globally optimal or approximately optimal control parameter solutions, thereby reducing the amount of calculation, saving time, and improving search efficiency and solution quality; initial values ​​are provided by sensitivity analysis and the least squares method, combined with the optimization of the improved black hole algorithm, to improve the efficiency of seeking optimal control parameters and improve system stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 It is a parameter identification flow chart of the present invention; Figure 2 It is a sensitivity analysis result diagram of the present invention; Figure 3 It is the flow chart of the improved black hole algorithm of the present invention; Figure 4 is the power curve fitting degree under short-circuit fault of the present invention; Figure 5 It is the convergence diagram of the three algorithms of the present invention; Figure 6 This is a comparison chart of the simulation and test results of error analysis condition 1 of the present invention; Figure 7 This is a comparison chart of the simulation and test results of error analysis condition 2 of the present invention; Figure 8 This is a comparison chart of the simulation and test results of error analysis condition 3 of the present invention; Fig. 9 It is a comparison chart of the simulation and test results of error analysis condition 4 of the present invention; Fig.10 It is a comparison chart of the simulation and test results of error analysis condition 5 of the present invention; Fig.11 It is a comparison chart of the simulation and test results of error analysis condition 6 of the present invention; Fig.12 It is a comparison chart of the simulation and test results of error analysis condition 7 of the present invention; Fig.13 It is a comparison chart of the simulation and test results of error analysis condition 8 of the present invention. DETAILED DESCRIPTION

[0014] Example 1 This embodiment proposes a method for identifying control parameters of a doubly-fed wind turbine model established in PLECS based on a traditional black hole algorithm, so as to accurately and reliably identify key control parameters in a doubly-fed wind turbine wind power generation system model.

[0015] like Figure 1-13 As shown, a control parameter identification method for a doubly-fed wind turbine based on an improved black hole algorithm comprises the following steps: S1: Collect the observed values ​​of the system output during the fault period of the wind turbine and save them as the initial data set. The observed values ​​include the active power and reactive power on the grid-connected side.

[0016] S2: Build a doubly-fed wind turbine simulation model according to the preset typical values ​​of the control parameters, and determine the parameters to be identified in the initial data set.

[0017] S3: Use trajectory sensitivity as an indicator to analyze the correlation between the parameters to be identified and the observations to be selected, determine the observation with the highest sensitivity as the optimal observation, and determine the identifiability of each parameter to be identified.

[0018] S4: The least squares method is combined with the measured data under RT-LAB (Real-Time Laboratory) fault conditions to obtain the initial values ​​of the rotor-side converter PI (proportional-integral) control parameters to determine the optimization interval.

[0019] S5: According to the optimization interval in S4, based on the data obtained by simulating the improved black hole algorithm and the doubly fed wind turbine simulation model, the control parameters are identified and adjusted, and it is continuously iterated until the fitness function meets the requirements and the identification effect is obtained.

[0020] S6: The parameter identification model based on the improved black hole algorithm is simulated and verified and compared with the traditional black hole algorithm and particle swarm algorithm to determine its effectiveness.

[0021] In this embodiment, the system output observations of the wind turbine during the fault period are collected as the initial data set for parameter identification and optimization in subsequent steps, a doubly fed wind turbine simulation model is built, and the key control parameters that need to be identified under fault conditions are determined. Then, the trajectory sensitivity of the parameters to be identified and the system output (such as active power and reactive power) is calculated to determine the optimal observations; then, the least squares method is used in combination with the RT-LAB measured data to obtain the initial value, and the control parameters are identified and optimized based on the improved black hole algorithm to search for the optimal control parameters; finally, simulation verification and algorithm comparison are performed, and the effectiveness of the improved algorithm is further verified by comparing the performance difference between the improved black hole algorithm and the particle swarm algorithm in control parameter identification. This embodiment adopts an improved black hole algorithm to effectively avoid local optimal solutions and provide globally optimal or approximately optimal control parameter solutions, thereby reducing the amount of calculation, saving time, and improving search efficiency and solution quality; by providing initial values ​​through sensitivity analysis and least squares method, combined with the optimization of the improved black hole algorithm, the efficiency of seeking optimal control parameters is improved, and the system stability is improved; by comparing with traditional algorithms, the effectiveness of the improved black hole algorithm in the identification of doubly fed wind turbine control parameters is verified, and it has a strong engineering application value.

[0022] In step S1, P and Q are the initial data sets, that is, the basic observable quantities of the wind turbine during the fault period.

[0023] like Figure 4 As shown in the figure, it is the power curve fitting degree under short-circuit fault in this implementation scenario. It can be seen that: The measured curve is the data obtained by hardware-in-the-loop semi-physical simulation in RT-LAB, and the simulation curve is the data obtained by simulating the doubly fed wind turbine simulation model built in PLECS and filling in the identified parameters. It can be seen that due to the high accuracy of the identified PI control parameters, the fit between the measured data curve and the simulation data curve is very high, which preliminarily proves the feasibility of the proposed improved black hole algorithm in control parameter identification.

[0024] According to the typical structure and control block diagram of the rotor-side converter control, the RSC controller structure is a double closed-loop PI control system consisting of a current inner loop and a power outer loop.

[0025] In the preferred solution, in step S2, the rotor-side converter control equation of the doubly-fed wind turbine simulation model is expressed as: ; ; Where: The rotor current is Reference values ​​of axis components; are the reference values ​​of stator active power and reactive power respectively; are the active power and reactive power output by the stator of the doubly-fed induction generator respectively; is the slip, ; is the synchronous angular velocity of the motor; is the rotor angular velocity; are the dq axis components of the rotor voltage respectively; The rotor current is Axis component; is the self-inductance of the rotor winding; is the self-inductance of the stator winding; is the mutual inductance between the stator and rotor windings; is the magnetic flux leakage coefficient, ; is the stator voltage; are the converter control parameters respectively.

[0026] In this embodiment, data collection and simulation modeling, namely steps S1 and S2, provide important basic data for the subsequent optimization process. The collected data during the fan failure period and the simulation model construction are the prerequisites for control parameter identification.

[0027] In the preferred solution, in step S3, obtaining the sensitivity between each parameter to be identified and the observed value of the system output includes: S301: The expression of trajectory sensitivity is: ; In the formula, , For parameters The initial value of is the change of the corresponding parameter; is the selected observation; is the time t The trajectory sensitivity of the parameters.

[0028] The initially selected observation quantities are the active power P, reactive power Q, grid-connected voltage amplitude U, current amplitude I and speed of the DFIG (Doubly-Fed Induction Generator) grid-connected output.

[0029] S302: Select the average value of trajectory sensitivity as the final criterion for measurement. The calculation formula for the average value of the absolute value of the trajectory sensitivity of each observation to the parameter to be identified is as follows: ; In the formula, is the average sensitivity; is the number of samples.

[0030] S303: Analyze the average sensitivity calculation results to obtain the optimal observation quantity and sensitivity.

[0031] like Figure 2 The figure shows the final average sensitivity calculation results in three wind speed ranges. It is concluded that: taking P and Q as observation quantities, their corresponding trajectory sensitivities are much higher than U and I. At the same time, it can be seen that the active power output value on the rotor side has a significant influence on the control parameters. Relatively sensitive, reactive power output value to control parameters Relatively sensitive. The sensitivity values ​​of these parameters are greater than other parameters, and the recognition is higher.

[0032] Next, the least squares method is used in combination with the measured data under RT-LAB fault conditions to obtain the initial values ​​of the rotor-side converter PI control parameters, so as to determine the optimization interval. Then, based on the improved black hole algorithm and the data obtained from simulation, the control parameters are identified and adjusted. The algorithm is continuously iterated until the fitness function meets the requirements and achieves the best identification effect.

[0033] In the preferred embodiment, step S4 includes: S401: Use the least square method as the fitness function to quantify the black hole algorithm, and set the star with the optimal fitness value as a black hole. The formula is: ; Where: and To identify the power output value obtained by the model and the actual power measurement value, n is the number of samples.

[0034] S402: Establish the optimization interval and use it as the upper and lower limits of the black hole search space. The formula is: ; In the formula, is the solution in the search space, i.e. value; is the lower limit set, The upper limit is set; For interval Random numbers within; is the number of planets; .

[0035] In this embodiment, the lower limit and upper limit of the parameter are set, and the boundary value is determined according to the actual system situation and the range of the initial value. The planet is initialized (search solution) by chaos mapping to enhance the randomness and exploration of the search.

[0036] The fitness function evaluates the quality of the solution based on the difference between the power output value obtained by the identification model and the actual power measurement value.

[0037] In the preferred embodiment, step S5 specifically includes: S501: Use Logistic chaos mapping to initialize the planet positions, the formula is: ; Where: is the chaotic mapping parameter, is a solution in the search space.

[0038] S502: Greedy retention strategy and Levy flight mechanism are introduced to improve the formula for updating the position of stars in the traditional black hole algorithm. The formula of the greedy retention strategy is: ; Where: is the position of the i-th planet after t iterations according to the greedy retention strategy.

[0039] The Levy flight mechanism formula is: ; Where: is the parameter that obeys the normal distribution, that is . and They are: ; Where: is the Gamma function; .

[0040] Finally, the planet position update is optimized as follows: ; Where: Obey for step length Random search vector for distribution ; It is a vector operation.

[0041] S503: Calculate and compare fitness, determine whether the black hole is replaced, and calculate the distance between the star and the black hole , When a new star is generated, the formula is: ; Where: m is the dimension of the star; is the jth component of the i-th star; K is a constant that adjusts the black hole absorption rate; is the fitness of the black hole; N is the total number of stellar variables; is the fitness value of the i-th planet; S504: If the convergence condition is met or the number of iterations is reached, the iteration is stopped and the optimal solution is output as the identification result.

[0042] Otherwise, go to step S4 until the convergence condition is met or the number of iterations is reached, then stop the iteration.

[0043] In this embodiment, a greedy retention strategy and a Levy flight mechanism are introduced to improve the position update formula, thereby improving the convergence speed and global search capability of the traditional black hole algorithm. The improved black hole algorithm finds the optimal solution through continuous iteration and evaluation of fitness values, thereby providing an efficient and accurate method for identifying fan control parameters.

[0044] If the fitness value of a planet is better than the current black hole fitness value, the position of the black hole is updated; if the change in the fitness function is less than the preset threshold (convergence condition) or the maximum number of iterations is reached, the optimization process is stopped.

[0045] Example 2 Based on the previous embodiment, this embodiment provides a comparative verification of a parameter identification method for a doubly-fed wind turbine converter control system.

[0046] The preferred solution also includes step S6: simulating and verifying the parameter identification model based on the improved black hole algorithm and comparing it with the traditional black hole algorithm and the particle swarm algorithm to determine its effectiveness, specifically: S601: Identify the PI control parameters of the rotor-side converter using the observed quantity, and perform simulation verification through the doubly-fed wind turbine simulation model built in S2.

[0047] S602: A symmetrical three-phase short-circuit fault is set at the outlet of the wind turbine to make the voltage drop to 50% of the initial value: a short-circuit fault is applied at t=1.5s, and the fault disappears at t=2s; the wind turbine operation data between 0 and 3s is collected, and the wind speed is set to 10m / s. At this time, the wind turbine operates in the MPPT (Maximum Power Point Tracking) state; the IBHA (Improved Black Hole Algorithm) algorithm is used to identify the parameters to be identified, the number of iterations is 50, the algorithm's optimization range is set to 0.5 to 1.5 times the preliminary identification result, and the average value of 100 identification results is taken.

[0048] S603: In order to further verify the accuracy of parameter identification, the identified parameters are applied to the simulation model, and the active and reactive power curves obtained by simulation are compared with the curves of the original model.

[0049] S604: Select the number of algorithm iterations for each optimization as 100, then perform 100 simulation tests, take the average value as the result of parameter identification, compare the obtained control parameter identification result with the control parameter identification results using the particle swarm algorithm and the traditional black hole algorithm, and obtain the objective function convergence curves of the three different optimization algorithms when solving the control parameters.

[0050] In this embodiment, the low voltage crossing error assessment standard commonly used in engineering is used for measurement. In this standard, the error is divided into four categories: average deviation, average absolute deviation, maximum deviation and weighted average deviation. The output characteristics during low voltage crossing are classified by time period: pre-fault interval a, fault period interval b, and post-fault interval c. For the three steady-state intervals, the steady-state average deviation F1, the steady-state average absolute deviation F3, and the steady-state maximum deviation F5 are calculated. For the two transient intervals, the transient average deviation F2 and the transient average absolute deviation F4 are calculated. Then, the F3 of the three intervals a, b, and c are weighted averaged according to the weights of 10%, 60%, and 30%, and the weighted average absolute deviation FG of the entire period is obtained.

[0051] In this embodiment, the data and comparison algorithm used are shown in Table 1 below.

[0052] Appendix 1 Model verification results

[0053] Appendix 2 Low Pressure Ride-Through Verification Conditions

[0054] As shown in Table 2, it is a table of various working conditions used in the verification test.

[0055] By comparing the experimental results, we can draw the following conclusions: Figure 5 As shown, there are three algorithms to solve the convergence diagram: Improved Black Hole Algorithm (IBHA): Compared with the traditional black hole algorithm and particle swarm algorithm, the improved black hole algorithm may have advantages in convergence speed and accuracy, especially after the introduction of the greedy retention strategy and Levy flight mechanism, it can more effectively avoid local optimal solutions and better find the global optimal solution.

[0056] Traditional BHA (Black Hole Algorithm): It can find a better solution to a certain extent, but due to its slow convergence speed, it is easy to fall into the local optimum and its performance is not as good as the improved version.

[0057] Traditional PSO (Particle Swarm Optimization): Although the particle swarm algorithm performs well in global search, it may have slow convergence problems when dealing with more complex wind turbine control problems, especially in high-dimensional search space.

[0058] like Figure 6-Figure 13 As shown in the figure, through simulation verification, the improved black hole algorithm shows better performance than the traditional black hole algorithm and particle swarm algorithm in the identification of control parameters of doubly fed wind turbines. It can accurately identify the control parameters in a short time and effectively optimize the system performance.

[0059] By comparing with traditional optimization algorithms, the effectiveness and advantages of the improved black hole algorithm in wind turbine control parameter identification are further verified.

[0060] Based on PLECS, low voltage ride-through experiments were conducted under high wind and low wind conditions. A three-phase symmetrical short-circuit fault was set at the grid connection point, and the grid voltage was dropped to 35%UN, 50%UN, 60%UN and 80%UN under each condition, respectively. UN is the rated voltage of the grid. The measurement error was analyzed. The test conditions are shown in Table 2. The waveform comparison results of the simulation conditions to be verified in the present invention are shown in Table 2. Figure 6-13 The deviations of the PSO and BHA identification models were verified under the same working conditions, and the model verification results were obtained, as shown in Table 1. From the above verification results, it can be seen that in terms of the identification of the control parameters of the doubly fed wind turbine, the results obtained by directly using the traditional PSO and BHA for identification have large errors, while the control parameter identification using the IBHA proposed in this paper meets the specified error standards. Compared with the PSO algorithm and BHA, the errors of the IBHA identification model in each interval in the low pressure ride-through verification are smaller than those of the former two, and the total errors are reduced by 0.123 and 0.088 respectively, achieving better results.

[0061] Example 3 In combination with Example 1, a control parameter identification system for a doubly-fed wind turbine based on an improved black hole algorithm is further described, and includes the following modules: The data acquisition module is used to collect the observed quantities of the system output during the fault period of the wind turbine and save them as the initial data set. The observed quantities include the active power and reactive power on the grid-connected side.

[0062] The model building module is used to build a doubly-fed wind turbine simulation model according to the preset typical values ​​of the control parameters and determine the parameters to be identified in the initial data set.

[0063] The sensitivity calculation module is used to use trajectory sensitivity as an indicator to analyze the correlation between the parameters to be identified and the observations to be selected, determine the observation with the highest sensitivity as the optimal observation, and determine the identifiability of each parameter to be identified.

[0064] The optimization interval module is used to obtain the initial values ​​of the rotor-side converter PI control parameters by using the least squares method combined with the measured data under the RT-LAB fault condition, so as to determine the optimization interval.

[0065] The identification module is used to identify and adjust the control parameters according to the optimization interval in the optimization interval module, based on the data obtained by simulating the improved black hole algorithm and the doubly fed wind turbine simulation model, and continuously iterate until the fitness function meets the requirements to obtain the identification effect.

[0066] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A control parameter identification method for a doubly-fed wind turbine based on an improved black hole algorithm is characterized in that: The following steps are involved: S1: Collect the observed values ​​of the system output during the fault period of the wind turbine and save them as the initial data set. The observed values ​​include the active power and reactive power on the grid side. S2: Build a simulation model of a doubly-fed wind turbine according to the preset typical values ​​of control parameters, and determine the parameters to be identified in the initial data set; S3: Use trajectory sensitivity as an indicator to analyze the correlation between the parameters to be identified and the observations to be selected, determine the observation with the highest sensitivity as the optimal observation, and determine the identifiability of each parameter to be identified; S4: Using the least square method combined with the measured data under the RT-LAB fault condition, the initial values ​​of the rotor-side converter PI control parameters are obtained to determine the optimization interval; S5: According to the optimization interval in S4, based on the data obtained by simulating the improved black hole algorithm and the doubly fed wind turbine simulation model, the control parameters are identified and adjusted, and it is continuously iterated until the fitness function meets the requirements and the identification effect is obtained.

2. The control parameter identification method of a doubly-fed wind turbine based on an improved black hole algorithm according to claim 1 is characterized in that: In step S2, the rotor-side converter control equation of the doubly-fed wind turbine simulation model is expressed as: ； ； Where: The rotor current is Reference values ​​of axis components; are the reference values ​​of stator active power and reactive power respectively; are the active power and reactive power output by the stator of the doubly-fed induction generator respectively; is the slip, ; is the synchronous angular velocity of the motor; is the rotor angular velocity; are the dq axis components of the rotor voltage respectively; The rotor current is Axis component; is the self-inductance of the rotor winding; is the self-inductance of the stator winding; is the mutual inductance between the stator and rotor windings; is the magnetic flux leakage coefficient, ; is the stator voltage; are the converter control parameters respectively.

3. The control parameter identification method of a doubly-fed wind turbine based on an improved black hole algorithm according to claim 1 is characterized in that: In step S3, the sensitivity between each parameter to be identified and the observed value of the system output is obtained, including: S301: The expression of trajectory sensitivity is: In the formula, , For parameters The initial value of is the change of the corresponding parameter; is the selected observation; is the time t The trajectory sensitivity of the parameters; S302: Select the average value of trajectory sensitivity as the final criterion for measurement. The calculation formula for the average value of the absolute value of the trajectory sensitivity of each observation to the parameter to be identified is as follows: ； In the formula, is the average sensitivity; is the number of samples; S303: Analyze the average sensitivity calculation results to obtain the optimal observation quantity and sensitivity.

4. The control parameter identification method of a doubly-fed wind turbine based on an improved black hole algorithm according to claim 1 is characterized in that: Step S4 includes: S401: Use the least square method as the fitness function to quantify the black hole algorithm, and set the star with the optimal fitness value as a black hole. The formula is: ； Where: and The power output value obtained by the identification model and the actual power measurement value, n is the number of samples; S402: Establish the optimization interval and use it as the upper and lower limits of the black hole search space. The formula is: ； In the formula, is the solution in the search space, i.e., value; is the lower limit set, The upper limit is set; For interval Random numbers within; is the number of planets; .

5. The control parameter identification method of a doubly-fed wind turbine based on an improved black hole algorithm according to claim 1 is characterized in that: Step S5 specifically includes: S501: Use Logistic chaos mapping to initialize the planet positions, the formula is: ； Where: is the chaotic mapping parameter, is the solution in the search space; S502: Greedy retention strategy and Levy flight mechanism are introduced to improve the formula for updating the position of stars in the traditional black hole algorithm. The formula of the greedy retention strategy is: ； Where: is the position of the i-th planet after t iterations according to the greedy retention strategy. The Levy flight mechanism formula is: ； Where: is the parameter that obeys the normal distribution, that is ; and They are ； Where: is the Gamma function; ; Finally, the planet position update is optimized as follows: ； Where: Obey for step length Random search vector for distribution ; is a vector operation; S503: Calculate and compare fitness, determine whether the black hole is replaced, and calculate the distance between the star and the black hole , When a new star is generated, the formula is: ； Where: m is the dimension of the star; is the jth component of the i-th star; K is a constant that adjusts the black hole absorption rate; is the fitness of the black hole; N is the total number of stellar variables; is the fitness value of the i-th planet; S504: If the convergence condition is met or the number of iterations is reached, the iteration is stopped and the optimal solution is output as the identification result.

6. The control parameter identification method of a doubly-fed wind turbine based on an improved black hole algorithm according to claim 1 is characterized in that: The method further includes step S6: simulating and verifying the parameter identification model based on the improved black hole algorithm and comparing it with the traditional black hole algorithm and the particle swarm algorithm to determine its effectiveness, specifically: S601: Identify the PI control parameters of the rotor-side converter using the observed quantity, and perform simulation verification through the double-fed wind turbine simulation model built in S2; S602: A symmetrical three-phase short-circuit fault is set at the outlet of the fan to make the voltage drop to 50% of the initial value: a short-circuit fault is applied at t=1.5s, and the fault disappears at t=2s; the fan operation data between 0 and 3s is collected, and the wind speed is set to 10m / s. At this time, the fan works in the MPPT state; the IBHA algorithm is used to identify the parameters to be identified, the number of iterations is 50 times, the algorithm's optimization range is set to 0.5 to 1.5 times the initial identification result, and the average value of 100 identification results is taken; S603: In order to further verify the accuracy of parameter identification, the identified parameters are applied to the simulation model, and the active and reactive power curves obtained by simulation are compared with the curves of the original model; S604: Select the number of algorithm iterations for each optimization as 100, then perform 100 simulation tests, take the average value as the result of parameter identification, compare the obtained control parameter identification result with the control parameter identification results using the particle swarm algorithm and the traditional black hole algorithm, and obtain the objective function convergence curves of the three different optimization algorithms when solving the control parameters.

7. A control parameter identification system for a doubly-fed wind turbine based on an improved black hole algorithm, characterized in that: Includes the following modules: A data acquisition module is used to collect the observed values ​​of the system output of the wind turbine during the fault period and save them as the initial data set. The observed values ​​include the active power and reactive power on the grid-connected side. A model building module is used to build a simulation model of a doubly-fed wind turbine according to typical values ​​of preset control parameters and determine the parameters to be identified in the initial data set; A sensitivity calculation module is used to analyze the correlation between the parameters to be identified and the observations to be selected using trajectory sensitivity as an indicator, determine the observation with the highest sensitivity as the optimal observation, and determine the identifiability of each parameter to be identified; The optimization interval module is used to obtain the initial value of the rotor-side converter PI control parameter by using the least square method combined with the measured data under the RT-LAB fault condition, so as to determine the optimization interval; The identification module is used to identify and adjust the control parameters according to the optimization interval in the optimization interval module, based on the data obtained by simulating the improved black hole algorithm and the doubly fed wind turbine simulation model, and continuously iterate until the fitness function meets the requirements to obtain the identification effect.

Images