免疫蚂蚁算法优化的RBF网络用于模拟电路故障诊断.pdfVIP

学兔兔 第31卷 第6期 仪 器 仪 表 学 报 V01．31 No．6 2010年6月 Chinese Journal of Scientific Instrument Jun．2010 免疫蚂蚁算法优化的RBF网络用于模拟电路故障诊断木 李 晴，何怡刚，包 伟 (湖南大学电气与信息工程学院 长沙 410082) 摘 要：针对传统蚂蚁算法中出现的易停滞现象，引入免疫机制，提出了一种新的融合算法，即免疫-蚂蚁算法，用其对RBF网 络中心参量进行优化，并将优化后的模型用于模拟电路故障诊断。详述了算法原理，构造网络模型的步骤以及故障诊断步骤， 并给出了仿真实例。仿真结果表明该方法可以有效克服基本蚂蚁算法的缺陷，优化得到的RBF网络实现了对隐层中心参量的 自适应获取，提高泛化能力，用于故障诊断有速度快、准确率高的特点。 关键词：蚂蚁算法；免疫算法；RBF网络；模拟电路；故障诊断 中图分类号：TN71 1 文献标识码：A 国家标准学科分类代码：520．2040 Immune—ant algorithm based RBFNN for fault diagnosis of analog circuits Li Qing，He Yigang，Bao Wei (College of Electrical and Information Engineering，Hunan University，Changsha 410082，China) Abstract：Aiming at the phenomena of stagnation in traditional ant colony system(AS)，the immune algorithm is introduced，and a new mixed algorithm，i．e．the ant colony system based on immune principle(ACS-IP)，is pro— posed．The proposed approach was applied to optimize the selection of data center parameters of radial basis function neural network(RBFNN)．And the optimized model was applied to analog circuit fault diagnosis．The algorithm principle，model construction steps and diagnosis steps are described in this paper，and simulation examples are giv— en as wel1．Simulation results show that this method can effectively overcome the shortcomings in AS；the optimized RBFNN can achieve data acquisition with self-adapting．Also，when applied in diagnosis，the model is more effec— tire and accurate，and has enhanced generalization capability． Key words：ant colony system；immune algorithm；RBF neural network；analog circuit；fault diagnosis 优于BP网络，而被广泛应用于函数逼近、模式识别、信号