基于粒子群优化支持向量机决策树的齿轮箱故障诊断方法

发布时间：2018-01-10 01:22

本文关键词：基于粒子群优化支持向量机决策树的齿轮箱故障诊断方法　出处：《太原理工大学》2012年硕士论文　论文类型：学位论文

【摘要】：齿轮箱作为机械设备中一种必不可少的连接和传递动力的通用零件,在现代工业发展中具有广泛的应用。在工业生产过程中,大约有80%的机械故障是由齿轮箱故障引起的,这些故障的发生将直接影响机器寿命、生产安全与生产效益等。因此,对其进行故障诊断,以保证设备正常运行具有非常重要的实际意义。本文针对齿轮箱的故障机理,研究了齿轮箱故障特征提取技术,特别是针对现有支持向量机在利用决策树策略构造多类分类器的过程中存在的问题,提出基于粒子群优化决策树的支持向量机模型,以解决齿轮箱故障难以及时有效地辨识的问题。首先,本文分析齿轮箱振动机理和常见的几种故障类型,并在此基础上,搭建了齿轮箱试验平台,设计了齿轮箱故障诊断试验方案,通过测试得到了不同故障下齿轮箱的振动响应加速度信号。其次,通过粒子群优化算法的基本原理,分析了粒子群算法的参数对算法精度与效率的影响,以及参数的选择原则。为了验证带惯性权重PSO算法的性能,本文选取了两个常用的标准函数做仿真,通过实验表明,惯性权重PSO算法在满足收敛精度的条件下,能够快速收敛到全局最优解。最后,将利用惯性权重PSO算法优良的全局搜索性能优化支持向量机决策树,生成最优决策树,构造出多元分类器,并将其代入齿轮箱故障诊断的试验数据中,结果表明,本论文研究的方法不但能够有效地实现故障诊断,并且提高了故障诊断的正确率,该研究方法为机械设备故障诊断提供了一种更加准确、更加有效的方法。
[Abstract]:In the process of industrial production , about 80 % of the mechanical faults are caused by the failure of the gear box , and the occurrence of these faults will directly affect the machine life , the production safety and the production benefit , etc . Therefore , the fault feature extraction technology of the gear box will be directly affected by the failure mechanism of the gear box , and the support vector machine model based on the particle swarm optimization decision tree is proposed to solve the problem that the gear box is difficult to identify effectively in time . First of all , this paper analyzes the vibration mechanism of gear box and several common types of faults . On the basis of this , the gearbox test platform is built , the fault diagnosis test scheme of the gearbox is designed , and the vibration response acceleration signal of the gear box under different faults is obtained through testing . Secondly , through the basic principle of particle swarm optimization algorithm , the influence of parameters on the accuracy and efficiency of the algorithm and the selection principle of the parameters are analyzed . In order to verify the performance of the PSO algorithm with inertia weight , two commonly used standard functions are selected as the simulation , and the experiment shows that the inertia weight PSO algorithm can converge quickly to the global optimal solution under the condition of meeting the convergence accuracy . Finally , the optimal decision tree is optimized by using the global search performance of the inertial weight PSO algorithm to generate the optimal decision tree . The multi - classifier is constructed and substituted into the test data of the fault diagnosis of the gear box . The results show that the method not only can effectively realize fault diagnosis , but also improves the accuracy of fault diagnosis . The research method provides a more accurate and more effective method for fault diagnosis of mechanical equipment .

【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3;TP18

【参考文献】