改进鲸鱼算法优化SVDD的辊道窑窑温异常检测  

Roller kiln abnormal temperature detection based on SVDD optimized by improved whale optimization algorithm

作　　者：董明明 印四华 朱成就 金熹 DONG Mingming;YIN Sihua;ZHU Chengjiu;JIN Xi(School of Computers,Guangdong University of Technology,Guangzhou 510006,China;School of Electromechanical Engineering,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]广东工业大学计算机学院,广东广州510006 [2]广东工业大学机电工程学院,广东广州510006

出　　处：《现代电子技术》2021年第21期90-95,共6页Modern Electronics Technique

基　　金：国家自然科学基金重点项目-广东省联合基金(U1501248)。

摘　　要：针对现有辊道窑窑温异常检测是依靠经验设置固定阈值,存在效率低、容易造成误警等问题,提出改进鲸鱼算法优化支持向量数据描述方法(IWOA-SVDD)。采用自适应反向学习策略使鲸鱼算法搜索效率提升,加快收敛速度,又引入高斯变异算子,避免在迭代后期陷入局部最优解而过早收敛。采用改进鲸鱼算法对SVDD的核心参数惩罚常数c和核宽度σ进行寻优。使用UCI数据集进行验证,结果表明,改进鲸鱼算法对SVDD的优化相比粒子群算法、遗传算法和传统鲸鱼算法效果更好,识别精度更高。与常用的过程监控方法——核主成分分析和核偏最小二乘法进行对比实验,结果表明,所提方法检测结果更加准确、误警率更低,验证了所提出算法的优越性。The existing abnormal temperature detection for roller kiln relies on experience to set the fixed threshold value,which is inefficient and prone to causing false alarm,so a support vector data description(SVDD)method optimized by the improved whale optimization algorithm(IWOA)is proposed.The adaptive opposition-based learning strategy is used to improve the search efficiency and speed up the convergence speed of the whale optimization algorithm(WOA).The Gaussian mutation operator is introduced to avoid premature convergence due to falling into the local optimal solution at the later stage of iteration.The IWOA is used to optimize the kernel parameters(penalty constant c and kernel widthσ)of SVDD.The UCI(University of California-Irvine)data set is adopted for verification.The verification results show that,in comparison with the particle swarm optimization(PSO)algorithm,the genetic algorithm(GA)and the traditional WOA,the IWOA has a better optimization effect and higher recognition accuracy for SVDD.In comparison with the commonly-used process monitoring methods,for instance,kernel principal component analysis(KPCA)and kernel partial least squares(KPLS),the SVDD method′s detection results are more accurate and its false alarm rate is lower,which has verified the superiority of the proposed algorithm.

关 键 词：窑温异常检测 辊道窑 支持向量数据描述 改进鲸鱼优化算法 参数优化 数据检测 

分 类 号：TN98-34[电子电信—信息与通信工程] TP3[自动化与计算机技术—计算机科学与技术]