欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对汽车发动机失火故障问题，提出一种新的智能诊断方法。建立了汽车尾气中各气体的体积分数与失火故障原因的映射关系，对归一化处理的数据进行机器训练，将训练好的相关向量机模型应用于故障分类诊断。算法中的惩罚因子和径向基核函数参数对分类准确率有着很大的影响，利用粒子群算法对超参数进行了优化。将优化训练后的相关向量机模型与目前较成熟的遗传优化的神经网络及支持向量机方法进行了对比，实验结果表明新方法比传统方法在诊断精度和鲁棒性方面均有一定的提高。
英文摘要：
To solve the problems of the misfiring errors of an automobile engine, the authors, put forward a new in-telligent fault diagnosis method. A mapping relation is established the volume fraction of gases in the exhaust of the automobile and the cause of the misfire. Machine training is applied to normalized data and the trained relevance vector machine model is applied to the fault classification and diagnosis. The penalty factor and the RBF kernel pa-rameters in the algorithm greatly affect the classification accuracy. The particle swarm algorithm is used to optimize the super-parameters;in addition, the relevance vector machine model having experienced optimization training is compared with the presently mature genetic optimized neural network and support vector machine method. The ex-perimental results show that the new method improves the diagnosis accuracy and robustness.
同期刊论文项目
同项目期刊论文
期刊信息
位置：