基于HMM-SVM旋转风电装备叶片覆冰状态的评估

Study on prediction of blade icing state of rotary wind power equipment based on HMM-SVM

风电装备叶片覆冰严重影响风力发电机组的安全与稳定运行。为准确评估覆冰条件下旋转风电装备叶片覆冰状态,本文研究以旋转风电装备叶片覆冰监测数据为基础,建立了基于隐马尔科夫算法(HMM)和支持向量机算法(SVM)的旋转风电装备叶片覆冰状态评估模型;采用时域和频域方法综合分析旋转风电装备叶片覆冰振动信号,选取频率均方根RMS i、时域峰值XP i、速度V、位移S四个特征量为模型的输入,利用HMM算法将收集到的振动信号进行处理,并与风电装备叶片覆冰各状态进行匹配,将各状态观察序列概率优化后通过SVM分类器进行覆冰状态识别分类。实验结果显示:HMM-SVM模型评估风电装备叶片覆冰状态准确率达90.83%,比单独使用HMM模型进行叶片覆冰状态评估的准确度更高。

The icing of wind turbine blades seriously affects the safe and stable operation of wind turbine.In order to accurately assess the ice-covering status of rotating wind power equipment blades under ice-covering conditions,based on the ice-covering monitoring system of rotating wind power equipment blades.An ice-coated state assessment model for rotating wind power equipment was established based on HMM and SVM.The time domain and frequency domain methods are used to comprehensively analyze the ice-covered vibration signals of rotating wind power equipment blades.Four characteristic variables,root mean square frequency RMS i,time domain peak XP i,velocity V and displacement S were selected as the input of the model.Using HMM algorithm,the collected vibration signals are processed.Then,it is matched with each state of icing on wind power equipment blade.After the probability optimization of each state observation sequence,SVM classifier was used to identify and classify the ice-covered states.The experimental results show that the accuracy of the HMM SVM model in evaluating the ice-covered status of wind power equipment is 90.83%,it is more accurate than using HMM model alone to assess the ice-covering status of blades.