【摘要】 在风力机运行过程中，偏侧风状态会导致风能利用率降低，同时降低风机的疲劳寿命。在动态风向变化下，随着侧风角度的变化，风力机叶片上的应变信号存在着规律性变化趋势，但由于应变信号呈现出非平稳、非线性和多种频率成分叠加的复杂特性，使得通过应变信号来揭示侧风角度变化规律有一定难度。文章提出了一种基于应变感知和BP神经网络的风力机风向追踪算法（SPWDP）。首先，通过设计风力机叶片应变测试方案，采集侧风角度变化下的叶片表面的多点应变数据，对应变信号进行变分模态分解（VMD），将叶片应变随侧风角度变化的规律提取出来，定义为风向-应变的特征；然后，使用BP神经网络作为建模算法，对风向-应变特征进行学习，并使用粒子群算法对BP神经网络的初始权值和阈值进行优化；最后，得到动态风向跟踪模型。经风洞实验数据验证，所提SPWPD算法在风力机侧风风向判断上具有可行性。更多还原

【Abstract】 In the process of wind turbine operation, the crosswind state will not only reduce the utilization rate of wind energy, but also reduce the fatigue life of wind turbine. And under the dynamic wind direction change, the strain signals on wind turbine blades have a regular trend of variation with the variation of crosswind Angle. However, it is difficult to reveal the variation of crosswind angle through the strain signals due to the non-stationary, nonlinear and complex superposition of multiple frequency components. In this paper, a Stress Perception based Wind Direction Prediction algorithm(SPWDP) is proposed based on strain perception and BP neural network. By designing a wind turbine blade strain test scheme, multi-point strain data on blade surface under the change of crosswind Angle were collected, and the strain signal was decomposed by VMD. The variation rule of blade strain with crosswind angle was extracted and defined as the wind direction-strain feature. BP neural network was used as the modeling algorithm to learn the wind direction and strain characteristics, and PSO algorithm was used to optimize the initial weights and thresholds of BP neural network. Finally, a dynamic wind direction tracking model is obtained. The SPWPD algorithm proposed in this paper is verified by wind tunnel experiment data to be feasible in wind turbine crosswind wind direction judgment.更多还原