单 位：
合肥工业大学汽车工程技术研究院;合肥工业大学汽车与交通工程学院
关键词：
阻尼可调减振器;操稳性;平顺性;神经网络
摘 要：
基于一款在研高速电磁阀式三级阻尼可调减振器,利用试验台架测得其外特性曲线,与CDC减振器相比,具有相似的调节范围和更低的制造成本,具有大规模推广潜力.再通过BP神经网络建立其非参数模型,与试验对比可以很好地用于悬架动力学分析;采用MATLAB/Simulink搭建了装有此减振器的整车七自由度仿真模型,通过随机路面四轮激励输入得到车辆簧下质量加速度响应,训练了LSTM神经网络,通过验证集验证了该神经网络能有效识别路面等级;最后,利用训练好的LSTM神经网络和可调阻尼减振器进行实车随机路面和急加速急减速实验.证明了与传统被动悬架相比,基于LSTM神经网络控制的三级可调阻尼减振器悬架系统能有效改善汽车行驶时的操稳性与平顺性.
译 名：
Research on control of damped adjustable semi-active suspension system based on LSTM neural network
关键词：
damping adjustable shock absorber%operation stability%smoothness%neural network
摘 要：
Based on a high-speed solenoid valve type three-stage damping adjustable shock absorber,its external characteristic curve was measured by the test bench.Compared with CDC shock absorber,it had similar adjustment range and lower manufacturing cost,and had the potential to be popularized on a large scale.Then the non-parametric model was established by BP neural network,which could be used for suspension dynamics inverse analysis well compared with the test.A seven-degree-of-freedom simulation model of the vehicle equipped with the shock absorber was built using MATLAB/Simulink.The vehicle unsprung mass acceleration response was obtained by random four-wheel excitation input on the road surface,and the LSTM neural network was trained.The verification set verified that the neural network could effectively identify the road surface grade.Finally,LSTM neural network and adjustable damping damper were used to carry out random road surface and rapid acceleration and deceleration experiments of real cars.It was proved that compared with traditional passive suspension,the LSTM neural network controlled adjustable damping damper suspension system could effectively improve the operating stability and ride comfort of vehicles.
