基于BP神经网络的汽车变速箱试验台开发与试验研究

Study on the Development and Experimental of Automobile Gearbox Test-bed based on BP Neural Network

研制了一种通用型的变速箱在线检测试验台,该试验台利用工控机控制机械臂实现变速器的自动换挡。系统利用BP神经网络算法建立数据分析模型,通过梯度下降法完成神经网络训练,该网络可成功判别变速箱的优劣。利用建立好的专家系统解决换挡策略的非线性问题,可以准确的判断挡位参数。该试验台可进行疲劳测试、传动效率测试等多个测试项目,具有可靠性好、智能化高的特点,在变速箱自动检测与故障诊断方面有很好的应用价值。

A universal online detection test--bed of transmission is developed, the automatic shift of transmission is realized by the test--bed using the industry control computer control mechanical arm. Data analysis model is established based on neural network algorithm, and neural network train- ing is completed by the gradient descent method. The network can discriminate transmission quality successfully. The establishment of expert system is good to solve nonlinear problems of shift strategy and determine the gear parameters accurately. The fatigue test, transmission efficiency test and other test items can be completed by the test--bed. The system has good reliability, high intelligence and has the good application value in the aspect of automatic detection and fault diagnosis of gearbo~