基于自适应布谷鸟与反向传播协同搜索的病害识别系统  

Disease Recognition System Based on Cooperative Search of Cuckoo and BP Algorithm

作　　者：王秀清[1] 陈琪 杨世凤[1] WANG Xiuqing;CHEN Qi;YANG Shifeng(College of Electronic Information and Automation,Tianjin University of Science&Technology,Tianjin 300222,China)

机构地区：[1]天津科技大学电子信息与自动化学院,天津300222

出　　处：《天津科技大学学报》2020年第2期69-73,共5页Journal of Tianjin University of Science & Technology

基　　金：天津市应用基础与前沿技术研究计划资助项目(14JCZDJC39000)。

摘　　要：为实现番茄病害快速诊断识别,提出了一种基于自适应布谷鸟搜索算法(cuckoo search,CS)与反向传播(back propagation,BP)协同搜索的病害识别算法(ASCS-BPCA)。首先,该算法将全局搜索能力强的CS与BP中反向传播算法结合,协同搜索最优识别参数,并在此基础上引入自适应调节步长机制加快布谷鸟搜索算法收敛速度。然后,以3种番茄病害(灰霉病、白粉病和晚疫病)叶片及正常叶片为研究对象,提取病斑特征集构建ASCS-BPCA病害识别模型,与标准CS-BP网络进行结果对比分析。仿真结果表明:ASCS-BPCA网络平均正确识别率达90%以上,优于同等条件下CS-BP算法,且更加稳定高效。In order to diagnose and classify tomato diseases rapidly,this research developed an algorithm(ASCS-BPCA)to collaborate self-adaptive step cuckoo search(CS)and back propagation(BP).Firstly,the algorithm combined the global search capability of cuckoo with back-propagation algorithm of BP to optimize weights and thresholds,based on which,it introduced self-adaptive step into the cooperative algorithm to accelerate the convergence speed of cuckoo algorithm.Three tomato diseased leaves and normal leaves were taken as research subjects and classification features of disease spots were extracted as a sample set to construct an ASCS-BPCA classifier.Finally,the classification accuracy of ASCS-BPCA network was compared with that of CS-BP network used in other researches.The results show that the average correct recognition rate of ASCS-BPCA is more than 90%,better than CS-BP algorithm’s performance under the same conditions,and ASCSBPCA is more stable and efficient.

关 键 词：协同搜索 自适应步长 特征提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]