Research On Parameter Optimization Of Variable Pitch PID Based On Data Drive

As modern industrial technology continues to advance,the depletion and requirement for energy is increasing.However,the scarcity of natural resource and environmental contamination have caused to a search for new sources of energy to patch these problems.Against this background,wind energy is reckoned to be the foremost foreground technology for power generation at the present stage.Currently,the majority of turbines adopt a convention PID pitch control strategy to revision the power.However,owing to the random nature of the wind and the fact that wind turbines themselves are a multidisciplinary and comprehensive project with a wide range of internal parameters,it is hassle to realize well performance adopt legacy PID control.What kind of establish an accurate pitch system model and an effective control strategy has become a challenge for pitch control technology.This thesis therefore emphasis on the system identification and the control of the wind turbine as follows.Firstly,the composition of turbines is systematically presented,and the energy switch procedure of wind power and the parameter of turbines under different working conditions are analysed,thus determining the main working conditions of this thesis,the permanent power control working conditon for turbines.Then,in order to solve the problem that the traditional modelling cannot precise describe the state of the system,this article use a data-driven BP neural network system identification,using the unit operating data in the wind farm SCADA system as the neural network recognition sample,dividing the pitch angle according to the interval of 0～5 degrees,5～10 degrees and10～15 degrees,and carrying out the interval of the variable pitch system the modeling of the pitch system is fine-grained.The accuracy of the identified model is verified by comparing the error between the actual output power of the system and the output power of the model and the root of mean square error.This provides a neural network system identification model of the controlled object for the pitch system.Finally,an improved particle swarm variable pitch PID control method is proposed according to the constant power control requirements.The strategy addresses the problems of the particle swarm algorithm in the process of finding the optimum and ignoring the global adjustment,and uses linear inertia weights,local search operators and adaptive adjustment strategies to improve it.Four standard test functions are used to test the revamp particle swarm algorithm,and the test results verify the effectiveness of the revamp method,and then the improved strategy is used to the pitch system for power control.The results indicate that the revamp control approach can provide better stabilisation of the wind turbine output power and lower control errors.