摘要
采集真实电解加工过程中阴极表面上的力信号,利用小波变换对采集的信号进行降噪处理,得到一条随着间隙减小而光滑增大的力信号趋势曲线.用这些力间隙数据对训练一个BP神经网络,通过训练好的网络和在线测得的力实现间隙的在线预报,并设计了一个模糊控制器.利用上述从试验中获得的间隙力关系训练另一个BP神经网络,实现间隙向力信号的映射,把间隙的误差转化为力的误差及误差的变化信号,以此作为模糊控制器的输入,以加工电压的增量作为模糊控制器输出,实现对间隙的控制.在M atlab的simulink模块中建立了由神经网络、模糊控制器和电解加工系统联合组成的智能控制系统的仿真模型,进行了仿真试验,试验结果表明对间隙的控制效果满意,特别是快速性和鲁棒性好.
The force signal on the surface of the cathode, which is led by the electrolyte, is gathered in the test and treated by wavelet transform to decrease noise and get a smooth curve of force signal which tends to ascend as the interelectrode gap decreases. This relationship of force-gap can be used to train a BP neural network (NN), then the interelectrode gap can be predicted on line by the trained BP NN and the force signal gathered on line. A fuzzy controller is designed, of which the input is the force error and the change of force error which are transfered from the error of interelectrode gap by using another BP NN trained by the pair of the gap-force data gathered in the test. The output of the controller is the increment of machining voltage. The simulation model based on the simulink module of Matlab is established. The fuzzy controller gives satisfactory performance.
出处
《东南大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2005年第5期719-723,共5页
Journal of Southeast University:Natural Science Edition
基金
国家自然科学基金资助项目(50275077)
高等学校博士点学科专项科研基金资助项目(20020287004)
航空科学基金资助项目(02H52048)
关键词
电解加工
间隙检测
力信号
智能控制
小波变换
electrochemical machining
measure of the interelectrode gap
force signal
intelligent control
wavelet transform