综合传动换挡品质的误差逆传播神经网络评价方法

Shifting Evaluation of Integrated Transmission Based on Back Propagation Neural Network

从换挡平顺性的角度出发,确定涡轮转速变化率的峰值、涡轮转速平均变化率及挡位3个参数作为客观评价指标,并将其细化定义;基于神经网络的基本原理,利用误差逆传播神经网络(BPNN)对履带车辆综合传动装置换挡品质评价进行客观描述;应用Matlab/Simulink软件的神经网络工具箱函数,建立用于评价车辆换挡品质的神经网络模型,并借助于GUI工具箱开发换挡品质客观评价等级软件进行仿真研究。通过神经网络测试结果与主观评价对比,证明此方法与传统主观评价方法具有较好的一致性,增强了换挡品质评价的客观性、针对性和灵活性。

To improve the shifting comfort, the peak of the turbine speed change rate, the average rate of the turbine speed change and gear are determined as three objective evaluation indices, which are defined in detail, and an objective description of shift quality evaluation for tracked vehicle equipped with integrated automatic transmission is given using back propagation neural network （BPNN） according to its basic principle. A BPNN model used for evaluating shift quality is established and trained by using the BPNN toolbox in Matlab/Simulink software, and an objective evaluation shift software is designed specially using GUI box in Matlab environment. The comparison of the test results with the subjective ratings shows that the developed evaluation method has a better consistency with the traditional one, which can also enhance the objectivity, pertinency and flexibility of this system.