焊接电源电信号滤波方法  

Electrical signal filtering method of welding power supply

作　　者：李小伟 宋芳[2] 

机构地区：[1]黄河交通学院,焦作市454950 [2]河南工业贸易职业学院,郑州市450007

出　　处：《焊接》2016年第1期64-67,72,共4页Welding & Joining

摘　　要：为了提高焊接过程中焊接电流控制精度,从众多干扰因素中得到高品质的电流信号,在保证滤波效果的同时,又能保证滤波后电流波形具有较好的跟踪和在线实时处理性能,提出了一种改进的神经网络加权卡尔曼焊接电源滤波算法。将神经网络引入到卡尔曼滤波器中,通过遗传算法对BP神经网络的初始值和阀值进行全局精度调节,通过神经网络在线实时调整自适应卡尔曼滤波器的加权因子,以提高滤波器整体性能。最后对提出的算法进行仿真,仿真结果表明:新型神经网络加权卡尔曼滤波方法能够有效滤除各种噪声,具有良好的适应性。In order to improve the control accuracy of welding current,obtain the high quality current signal from many interference factors,and guarantee the real-time performance of the filtered waveform to be tracked,an improved neural network weights Calman welding power source filter was proposed. The design principle of Calman filter was introduced,and the recurrence formula of the discrete weighted Calman filter was given. By introducing neural network to Calman filter,adjusting the overall accuracy of the initial values and threshold for BP neural network through genetic algorithm,the overall performance of the filter was improved by real time adjusting the weighted factor of adaptive Calman filter with neural network. Finally,the proposed algorithm was simulated,and the simulation results show that the new neural network weighted Calman filtering method can effectively remove all kinds of noise and has good adaptability.

关 键 词：焊接电流 卡尔曼滤波器 神经网络 仿真 

分 类 号：TG409[金属学及工艺—焊接]