风电叶片多点静力测试神经网络PID解耦控制

Neural Network PID Decoupling Control for Multi-Point Static Test of Wind Turbine Blades

风电叶片多点静力加载测试中由于存在各加载点之间的交联耦合现象,其加载精度难以得到保证。为此在研究耦合机理和解耦控制的基础上,提出变步长BP神经网络PID自整定算法。建立多点静力加载控制系统耦合模型,通过仿真并与实际实验数据比较,验证该算法的有效性。结果表明该算法可有效降低各加载点之间的耦合效应,有利于提高网络的稳定性和搜索速度,同时还具有自学习功能以适应不同尺寸和参数的叶片。

In the multi-point static loading test of wind turbine blades,cross-linking phenomenon exists among the loading points,the loading precision is difficult to be guaranteed.Therefore,based on the study of coupling mechanism and decoupling control,a variable step size BP neural network PID self-tuning algorithm is proposed.A coupling model of the multi-point static loading control system is established,the effectiveness of the algorithm is verified by simulation and comparison with actual experimental data.The results show that the algorithm can effectively reduce the coupling effect between the loading points,which is beneficial to improve the stability and search speed of the network,and it also has a self-learning function to adapt to blades of different sizes and parameters.