基于神经网络的风力发电机组变桨距复合控制

Compound pitch-control of wind turbine generator based on neural network

在分析风力发电机组系统特性和变桨距控制要求的基础上,提出了一种基于神经网络的分段复合控制方法进行变桨距控制,以解决风力发电系统的多干扰、时滞性、非线性等控制问题。该方法利用神经网络进行模型辨识,再依据运行工况分别进行模型预测控制和前馈控制,不但解决了风力发电机组系统模型难以精确建立的困难,而且去除了可测量的主要外扰——风速随机变化对系统动态控制品质的影响,从而提高了变距系统的响应快速性和抗干扰能力。最后,利用考虑尾涡效应的机组动态模型作为应用实例对复合控制方法进行仿真,结果表明提出方法的有效性与实用性。

In order to solve non-linear, multi-disturb and time lag problem of wind turbine generator （WTG）, a compound approach is investigated based on network to control pitch angle after analyzing characteristics of WTG and requisitions of pitch control. In this approach, by application of neural network model identification, model predictive control and feed-forward control are distinguishingly executed according to running state. So it both solves the problem of modeling WTG, and erases the influence of the main interference, variable wind speed, on control quality. As a resuit, it enhances rapidity and anti-interference capability of pitch system. Finally, utilizing the model considering trailing vortex as example, the simulation shows the effectiveness and practicability of the proposed method.