基于极限学习机的入土切割过程数据驱动模型研究

The Buried Cutting Process Data-Driven Model Based on Extreme Learning Machine

在甘蔗收割机切割过程中,入土切割能有效提高甘蔗切割质量,但入土切割深度过大极易造成推土现象,不利于控制刀盘升降。为了能够对刀盘入土切割深度进行有效控制,采用基于极限学习机的神经网络模型对影响刀盘负载压力的主要参数与刀盘入土深度之间的非线性关系进行拟合预测,并利用粒子群优化算法(PSO)对模型参数进行优化以提高极限学习机的泛化能力。通过仿真分析并与传统BP神经网络进行比较分析。由分析结果可知基于极限学习机的数据模型优于基于BP神经网络的数据模型,有良好的泛化能力,从而为对入土切割刀盘自动控制系统的研发提供了依据。

In the process of sugarcane harvester cutting,the buried cutting process which can effectively improve the quality of the sugarcane cutting,which also can easily cause bulldozing phenomenon when it cuts too deep.In order to be able to effectively control capacity of cutter cutting depth,the neural network model based on extreme learning machine main parameters affecting the load pressure capacity and the cutter ‘s depth the nonlinear relationship between forecast,and by using particle swarm optimization algorithm for model parameter optimization,which can improve the generalization ability. It will compare with the traditional BP neural network by the simulation analysis.the data model based on extreme learning machine is better than on BP neural network according to the analysis results,which has good generalization ability,which provides the basis for the research and development of the automatic control system.