基于粒子群算法的组合风速模型参数提取方法

Parameters Extraction Method for the Combined Wind Speed Model by the Particle Swarm Optimization Algorithm

真实风速模型构建是风电机组控制系统在实验环境下实现故障检测的重要保障。针对风电机组控制系统经验参数提取方法的不足,提出一种基于粒子群算法(Particle Swarm Optimization,PSO)的四分量组合风速模型多参数自动提取方法。该方法分为两步,以三分量模型参数和随机风分量模型参数为多维输入,风速预测值与真实值之间的均方根误差(RMSE)为目标输出,通过建立PSO算法寻优RMSE的最小值得到最佳的四分量模型参数;将得到的参数代入四分量组合模型得到风速预测值,并和实际测得的风速进行对比验证,使得四分量模型参数的提取更加智能化,提高其准确性并降低风速预测过程的时间成本。

The construction of real wind speed model is an important guarantee for fault detection of wind turbine control system in the experimental environment.The approximation degree of wind speed prediction model and natural wind directly affects the accuracy of fault detection of wind turbine.At present,the widely used wind speed four-component combined model contains many empirical parameters,which are not universal and difficult to reflect the authenticity of the operating environment of the wind turbine.The research of empirical parameter extraction method is still insufficient.Therefore,a four-component combined wind speed model multi-parameter automatic extraction method based on particle swarm optimization(PSO)is proposed in this paper.The method is divided into two steps.The three-component model parameters and the random wind component model parameters are taken as multi-dimensional input respectively.Root mean square error(RMSE)between the predicted wind speed and the true wind speed is used as the target output,and the PSO algorithm is established to find the minimum value of RMSE to obtain the best four-component model parameters.The obtained parameters are substituted into the four-component combined model to get the wind speed prediction value and then compare with the actual measured wind speed.It makes the four-component model parameter extraction more efficient and intelligent,improves its accuracy and reduces the time cost of wind speed prediction process.