摘要
研究了局部放电图像组合识别特征提取和反向传播算法神经网络分类器设计方法 ,根据变压器局部放电在线监测的要求 ,设计了 5种放电模型并进行了模拟实验。 5种放电模型数据识别结果说明 :与分别采用分形特征和统计特征的识别结果相比 。
PARTIAL DISCHARGE (PD) PATTERN RECOGNITION IS AN IMPORTANT METHOD FOR INSULATION DIAGNOSIS OF ELECTRICAL EQUIPMENT. IN THIS PAPER, THE COMBINATION FEATURES AND BACK PROPAGATION NEURAL NETWORK(BPNN)ARE STUDIED FOR PD PATTERN REMOTE RECOGNITION SYSTEM. ACCORDING TO THE REQUIREMENT OF ON LINE PD MONITORING FOR TRANSFORMER, SEVERAL DISCHARGE MODELS ARE DESIGNED AND THE RELEVANT EXPERIMENT METHODS ARE PROJECTED. WITH DISCHARGE MODEL TESTES, A LOT OF DISCHARGE SAMPLE DATA IS ACQUIRED. IT CAN BE SHOWN FROM ANALYSIS OF THE RECOGNITION RESULTS OF LARGE QUANTITIES OF THE PD SAMPLES THAT THE HIGHER RECOGNITION RATIO IS ACHIEVED IN USE OF COMBINATION FEATURES THAN THAT IN USE OF FRACTAL FEATURES OR STATISTICAL FEATURES SEPARATELY.
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2004年第6期11-13,共3页
High Voltage Engineering
基金
重庆大学骨干教师资助计划项目
关键词
局部放电
模式识别
组合特征
反向传播算法
神经网络
PARTIAL DISCHARGE PATTERN RECOGNITION COMBINED FEATURES BACK PROPAGATION NEURAL NETWORK(BPNN)
