文档介绍：河北工业大学
硕士学位论文
变桨距风力发电机数据采集与预测的研究与设计
姓名:郭红梅
申请学位级别:硕士
专业:通信与信息系统
指导教师:刘艳萍
20081101
河北工业大学硕士学位论文
变桨距风力发电机数据采集与预测
的研究与设计
摘要
在风力发电机组整体技术不断进步的今天,变桨距型风力发电机组以其风能利用率
高、起动性能与制动性能好和机组脱网时不需经历突甩负载过程等优点,越来越受到人们
的青睐。变桨距系统作为大型风电机组控制系统的核心部分之一,对机组安全、稳定、高
效的运行有十分重要的作用。
本文对变桨距风力发电机组核心技术之一的变桨距系统进行了理论分析,提出了基于
神经网络算法预测桨距角的方法,并设计了 DSP 控制的高速数据采集系统。
研究了基于 BP 算法、LM 算法的神经网络和 RBF 神经网络三种具体模型,并对桨距
角进行了预测。主要包括风速样本集的选取与预处理,BP 神经网络隐含层节点数的选取、
RBF 神经网络的模型参数设计,以及待识风速的预测等过程。实际应用结果表明基于神经
网络算法的桨距角预测方法具有较高的预测精度和重要的应用价值。
高速数据采集系统的核心运算单元为 TI 公司的 DSP TMS320F2812 芯片。外围电路包
括交流信号调理电路,锁相环频率跟踪和采样同步脉冲产生电路,光电编码器和风速风向
传感器与 DSP 接口电路设计等部分,可以实现对三相电压、三相电流、频率、采样脉冲、
转子转速、风速风向及实际桨距角的采样,初步达到了设计的目的。
关键词:变桨距,风力发电机,神经网络,DSP,数据采集
i
变桨距风力发电机数据采集与预测的研究与设计
RESEARCH AND DESIGN
ON DATA ACQUISITION AND PREDICTION FOR
PITCH-CONTROLLED WIND TURBINE
ABSTRACT
The technology in the field of wind turbine has been greatly improved in recent years.
Pitch-controlled wind turbine is believed to be superior to other types of wind turbine due to its
features such as high efficiency and ideal starting and braking performance. Pitch-controlled
system, the very core of a large-scale wind turbine control system, is playing a very important
role in the security, stability and efficient operation of the units.
The pitch-controlled system is theoretically analyzed and the method of predicting the pitch
angle reference value neural based work algorithms are raised in this paper, and the
DSP-controlled high-speed data acquisition system is designed.
Three idiographic models, such as the work based on BP algorithm, neural
network based on LM algorithm and RBF work, are discussed in this paper, and the
pitch angle reference value is predicted. The main steps includes, selection and pre-processing of
the sample set, selection of hidden-layer nodes in BP work, and design of model
parameters in RBF work, predictio