基于神经网络驾驶员模型的车间AGV控制算法研究

Research on Control Algorithm of AGV in Workshop Based on Neural Network Driver Model

人工神经网络作为理论上能够拟合任何非线性映射的算法被广泛用在驾驶员模型的构建之中,契合了驾驶员作为一个复杂、离散性和时变性系统的特点。本文研究的车间自动导航车辆(AGV)作为无人驾驶车辆的一种,具有应用场景相对单一和道路结构化的特点,对于人工神经网络拟合过程需要的训练样本量的要求不大。基于此构建驾驶员在环的神经网络训练样本采集试验平台,使用双层前馈神经网络进行车间AGV驾驶员模型拟合,并通过MATLAB/Simulink耦合Pr⁃escan软件进行算法仿真试验验证分析,试验结果验证了算法的有效性和一定的速度适应性,为车间AGV控制提供了一条可行的研究路径。

As an algorithm that can fit any nonlinear mapping,in theory,an artificial neural network is widely used in the construc⁃tion of the driver model,which fits the characteristics of a driver as a complex,discrete and time-varying system.As a kind of driv⁃erless vehicle,the workshop automatic navigation vehicle(AGV)studied in this paper has the characteristics of a single application scenario and structured road,and has little requirement for the training sample size of the artificial neural network fitting process.Based on this,a driver in the loop neural network training sample collection test platform is constructed,and the double-layer feed-forward neural network is used to fit the driver model of workshop AGV,and the algorithm simulation test is carried out by Matlab/Simulink coupled with Prescan software.The test results verify the validity and speed adaptability of the algorithm,which provides a feasible research path for workshop AGV control.