神经网络与DE算法在自适应滤波中的应用  

Application of Neural Network and Differential Evolution Algorithm in Adaptive Filtering

作　　者：张鸣[1] 韦国勋[1] 杨煜普[1] 

机构地区：[1]上海交通大学电子信息与电气工程学院,上海200240

出　　处：《自动化仪表》2010年第4期8-11,共4页Process Automation Instrumentation

基　　金：上海市科学技术委员会科技成果转化资金资助项目(编号:041111047)

摘　　要：为了满足电流互感器自动测试系统中处理采集数据的需求,提出了一种基于神经网络与微分进化算法的自适应滤波器。在传统自适应滤波器结构中引入神经网络,提高了非线性处理能力;而微分进化算法则能保证全局最优解。试验表明,将改进后的滤波器应用于测试系统中具有较好的滤波效果,验证了该滤波方法应用于工业测控领域的正确性与可行性。In order to meet the requirements of processing collected data in automatic test system for current transformers, an adaptive filter based on neural network and differential evolution algorithm is proposed. Neural network has been introduced in traditional adaptive filter to enhance non-linear processing capability; while the differential evolution algorithm can ensure global optimal solution. Having been used in the test system , the better filtering result of the improved filter are indicated, and the feasibility and correction of applying this filtering method in industrial measurement and control areas is verified.

关 键 词：神经网络 微分进化 自适应滤波 电流互感器 虚拟仪器 励磁特性 

分 类 号：TP216.1[自动化与计算机技术—检测技术与自动化装置]