基于PSO-SVM的电机故障检测

传统智能故障检测模型中算法初始参数复杂,选取难度较大,缺乏自学习、自组织能力、泛化能力弱,极易陷入局部极小值、算法单一等缺点.组合应用智能检测算法可整合不同算法优势,避免单一算法缺点,为此,文中提出支持向量机算法与改进粒子群算法相结合的电机故障检测模型:以电机故障特征频率特征数据为基础,首先使用改进全局求解性能的粒子群算法求解影响支持向量机分类检测性能的最佳参数,然后把最佳参数应用于的擅长模式识别的支持向量机算法,进行样本数据的训练,构建故障检测模型;最后,使用故障检测模型对电机的状态进行预测.实验结果表明,采用该方法进行故障检测的准确率,比传统的神经网络方法提高17%,比纯支持向量机算法提高3.33%.

Mine Based Motor Fault Detection Model Based on PSO-SVM

Traditional intelligent fault detection model such as neural network has some faults such as lacking of self-learning and self-organization, weak generalization ability, easy to fall into local minimum value and single. Intelligent detection algorithm of combination application can integrate advantages of different algorithms and avoid the disadvantages of single algorithm. Therefore, this paper proposes a mine based motor fault detection model based on combination of vector machine(SVM) algorithm and improved particle swarm optimization(PSO) algorithm. Firstly, the optimal parameters for SVM is got by using improved particle swarm optimization(PSO) algorithm, which has better inspiration performance and relapses into local optimal solution less. Secondly, the optimal parameters are used by SVM algorithm to train sample data for data classification, because SVM algorithm is good at pattern recognition. At last, a fault diagnosis model has built up. The experimental results show that the method can improve the accuracy of fault detection by 3.33%-17%.