Study mechanical transmission gearbox vibration frequency too high detection problem.The traditional artificial time to test the gearbox vibration frequency is too high,inefficient,and technical requirements of personnel.Due to the gear box itself to complex struc-ture,signal abnormalities,strong nonlinear relation of artificial detection accuracy is difficult to guarantee.Put forward a kind of gearbox vi-bration frequency is too high based on decision tree analysis found that method,time domain and frequency domain at the time of the gearbox vibration characteristic signal is extracted,the signal after data preprocessing and discrete classified by ID3decision tree structure vibration frequency too high diagnosis model,using particle swarm optimization method for decision tree structure calculation of the attribute informa-tion gain weight coefficient optimization,improved the gearbox vibration frequency is too high detection accuracy.Experimental results show that after optimization,the decision tree algorithm is high diagnostic accuracy of gearbox vibration frequency is as high as 97%,2%lower er-ror between the branches of the cumulative,have very strong practicability.%研究机械传动设备齿轮箱振动频率过高检测问题；传统的人工定时对齿轮箱的振动频率过高进行检测，效率低下，且对人员的技术要求较高；由于齿轮箱本身结构复杂、信号异常情况较多，非线性关系较强，人工检测的准确率难以保证；提出一种基于决策树分析的齿轮箱振动频率过高发现方法，对齿轮箱振动时时域与频域的特征信号进行提取，信号经过数据预处理与离散归类后通过ID3决策树构造振动频率过高诊断模型，使用粒子群优化的方法对决策树构造中的属性信息增益计算进行权重系数进行优化，提高了齿轮箱振动频率过高检测的准确率。实验表明，经过优化后的决策树发现算法对齿轮箱的振动频率过高诊断准确率高达97％，树枝之间的误差累积降低2％，具有很强的实用性。
展开▼
