摘要
为了提高雷电过电压的识别率,满足雷电过电压识别的实时性,提出了粒子群优化算法和最小二乘支持向量机的雷电过电压识别模型。首先提取多种特征作为雷电过电压识别的输入向量,然后采用最小二乘支持向量机设计雷电过电压识别的分类器,采用粒子群优化算法确定最合理的分类器参数,最后通过实验分析其有效性和优越性。结果表明,PSOLSSVM可以描述雷电过电压信号与特征间变化关系,提高了雷电过电压识别率,加快了雷电过电压识别速度,识别结果优于其他模型。
In order to improve identification rate and satisfy the identification real-time performance of lightning over- voltage, an identification model （PSO-LSSVM） of lightning over-voltage based on particle swarm optimizing algorithm （PSO） and least square support vector machine （LSSVM） is proposed in this paper. Firstly, multi-features are extracted and combined as input vectors for lightning over-voltage, and secondly, LSSVM is used to design the classifier for identifica- tion of lightning over-voltage and PSO is used to determine the most reasonable classifier parameters, finally, the validity and superiority of PSO-LSSVM are verified by the lightning over- voltage identification experiment. The results show that PSO- LSVM can better describe relationship between lightning over- voltage signal and features, and the identification rate is imp- roved and identification speed accelerated with identification results superior to those by other models.
出处
《电网与清洁能源》
北大核心
2016年第6期35-40,共6页
Power System and Clean Energy
基金
国家自然科学基金资助项目(31302231)
浙江省教育厅科研项目(Y201226043)
宁波市自然科学基金资助项目(2012A610110)~~
关键词
雷电过电压
最小二乘支持向量机
特征提取
粒子群优化算法
lightning over-voltage
least square support vector machine
extracting features
particle swarm optimizing algorithm
