CAJ | 학술논문

提出利用经遗传算法优化的 BP 神经网络辨识矿山路面的方法。建立了14自由度自卸车仿真模型，将仿真得到的座椅加速度作为网络理想输入样本，基于逆变换原理拟合出的路面不平度作为网络理想输出样本，通过网络训练，建立了两者之间非线性映射模型。对拟合出的不同等级路面、各种凹坑路面及自卸车不同载重下路面不平度进行辨识，辨识路面与测试路面相关系数高、相对误差小，验证了该方法具有对复杂矿山路面的辨识能力。通过整车道路试验，证明了该方法的准确性。与自卸车常用 C 级路面下的平顺性仿真结果的对比显示，采用该方法得到辨识路面更加接近实际路面，达到了提高模型仿真精度的目的。
제출이용경유전산법우화적 BP 신경망락변식광산로면적방법。건립료14자유도자사차방진모형，장방진득도적좌의가속도작위망락이상수입양본，기우역변환원리의합출적로면불평도작위망락이상수출양본，통과망락훈련，건립료량자지간비선성영사모형。대의합출적불동등급로면、각충요갱로면급자사차불동재중하로면불평도진행변식，변식로면여측시로면상관계수고、상대오차소，험증료해방법구유대복잡광산로면적변식능력。통과정차도로시험，증명료해방법적준학성。여자사차상용 C 급로면하적평순성방진결과적대비현시，채용해방법득도변식로면경가접근실제로면，체도료제고모형방진정도적목적。
BP neural network optimized by GA was used to identify the mining road.A fourteen degree-of-freedom vehicle vibration model was set up.The vehicle seat acceleration obtained by simu-lation was regarded as an ideal input sample of neural network,and the fitting road roughness was re-garded as an ideal output sample of neural network based on inverse transformation principles,then the nonlinear mapping model between them was built by network training.Road roughness was iden-tified under the conditions of different grade roads through fitting,various pit roads and different loads of dump truck.Identification ability was verified for complex mining roads due to high correla-tion coefficient and small relative error in this method.The accuracy of the method was verified through vehicle road test.Compared with simulation results of ride comfort under common C-class roads,it is shown that identification road is more closer to actual one,so as to achieve the purpose of improving the simulation accuracy of the models.