摘要
为提高制冷系统故障诊断的准确率,提出一种基于改进引力搜索算法(IGSA)优化的最小二乘支持向量机(LSSVM)的制冷系统故障诊断方法;首先,引入粒子群算法的速度更新机制对引力搜索算法进行改进,增加粒子的记忆性和信息共享能力,提高了算法的收敛速度和搜索精度;其次,利用IGSA对LSSVM的核参数与正则化参数进行优化,得到最优的IGSALSSVM故障诊断模型;最后,利用故障模拟实验台模拟制冷系统的4种典型故障,将优化好的LSSVM模型对其进行分类识别,并与标准LSSVM、GSA-LSSVM和PSO-LSSVM模型进行比较;仿真结果表明,基于IGSA优化的LSSVM方法具有良好的辨识能力和泛化能力,能够更好地对制冷系统故障进行诊断。
To improve the diagnosis accuracy of refrigeration system faults,an optimized Least Squares Support Vector Machine(LSSVM)based fault diagnosis method using the improved gravity search algorithm(IGSA)was proposed.Firstly,to increase the memory and information sharing ability of the particles,the gravitational search algorithm was further developed using the speed updating mechanism in particle swarm optimization algorithm,so that its calculation convergence and the search accuracy were improved.Through optimizing the kernel parameters and regularization parameters of LSSVM using IGSA,the proposed IGSA-LSSVM fault diagnosis model was then developed.Finally,using the experimental data obtained from a real refrigeration system,four typical faults of the refrigeration system were successfully identified by the optimized IGSA-LSSVM model.The fault diagnosis results,in comparison with that using the standard LSSVM,GSA-LSSVM and PSO-LSSVM models,showed that the proposed IGSA-LSSVM method was better as expressed in terms of its identification ability and generalization ability.
引文
[1]Beghi A,Brignoli R,Cecchinato L,et al.Data-driven Fault Detection and Diagnosis for HVAC water chillers[J].Control Engineering Practice,2016,53:79-91.
[2]文成林,吕菲亚,包哲静,等.基于数据驱动的微小故障诊断方法综述[J].自动化学报,2016(9):1285-1299.
[3]梁晴晴,韩华,崔晓钰,等.基于主元分析-概率神经网络的制冷系统故障诊断[J].化工学报,2016,(3):1022-1031.
[4]肖赋,王盛卫,徐新华,等.基于主成分分析法的空调系统传感器自动故障诊断[J].建筑科学,2008(6):34-34-39.
[5]汪海燕,黎建辉,杨风雷.支持向量机理论及算法研究综述[J].计算机应用研究,2014,31(5):1281-1286.
[6]J.A.K.Suykens,J.Vandewalle.Least Squares Support Vector Machine Classifiers[M].Kluwer Academic Publishers,1999,293-300.
[7]Olivier Chapelle,Vladimir Vapnik,Olivier Bousquet,et al.Choosing Multiple Parameters for Support Vector Machines[J].Machine Learning,2002,46(1-3):131-159.
[8]胡天骐,单剑锋,宋晓涛.基于改进PSO-LSSVM的模拟电路诊断方法[J].计算机技术与发展,2015(6):193-196.
[9]周竹,李小昱,李培武,等.基于GA-LSSVM和近红外傅里叶变换的霉变板栗识别[J].农业工程学报,2011(3):331-335.
[10]Xiaohui Yuan,Chen Chen,Yanbin Yuan,et al.Short-term wind power prediction based on LSSVM-GSA model[J].Energy Conversion&Management,2015,101:393-401.
[11]Esmat Rashedi,Hossein Nezamabadi-pour,Saeid Saryazdi.GSA:A Gravitational Search Algorithm[J].Information Sciences,2009,179(13):2232-2248.
[12]谷文祥,李向涛,朱磊,等.求解流水线调度问题的万有引力搜索算法[J].智能系统学报,2010(5):411-418.
[13]李航.统计学习方法[M].北京:清华大学出版社,2012.
[14]张树团,张晓斌,雷涛,等.基于粒子群算法和支持向量机的故障诊断研究[J].计算机测量与控制,2008,16(11):1573-1574.
[15]Mohammad-Ali Ahmadi,Alireza Bahadori.A LSSVM approach for determining well placement and conning phenomena in horizontal wells[J].Fuel,2015,153:276-283.
[16]Amarjeet Singh,Kusum Deep,Atulya Nagar.A New Improved Gravitational Search Algorithm for Function Optimization Using a Novel"Best-So-Far"Update Mechanism[J].2015:35-39.
[17]李晓宇,张新峰,沈兰荪.支持向量机(SVM)的研究进展[J].测控技术,2006,25(5):7-12.