基于APSO算法的拖拉机牵引性能预测通用模型建立与试验

General Model Building and Experiment on Traction Performance Prediction Based on APSO Algorithm

摘要:

针对现有轮式拖拉机牵引性能预测模型通用性差、预测精度低等问题，提出了一套涵盖系统建模、预测优化、实例验证全过程的适用于四轮驱动与二轮驱动拖拉机的牵引性能预测通用模型。通过深入分析土壤力学、轮胎力学、传动系统之间的相互作用，将拖拉机牵引性能抽象为轮-壤模型、驱动力模型、滑转率模型、牵引力模型4个基本模型，以建立适用于四轮驱动与二轮驱动拖拉机的整机牵引性能预测通用模型。为了提高预测精度，以整机滑转率为优化目标，建立基于自适应粒子群优化算法(APSO)的牵引性能预测优化方法。通过线上优化，验证了模型准确性和通用性。为了进一步验证该通用模型优越性和工程实用性，以东方红某105kW拖拉机作为试验样机，在中国一拖集团有限公司田间全地型试验场，完成线下试验。试验结果表明，与现有预测模型相比，对于四轮驱动拖拉机，基于APSO的牵引性能预测方法的滑转率和滚动阻力平均绝对误差分别为1.9%和0.18kN。对于二轮驱动拖拉机，相应的平均绝对误差分别为2.7%和0.25kN，精度大幅提升。

Abstract:

Aiming at the problems of poor generality and low prediction accuracy of existing models for traction performance of wheeled tractors, a set of general model for traction performance prediction of four-wheel drive and two-wheel drive tractors was proposed, which covered the whole process of system modeling, prediction optimization and case verification. By analyzing the interaction of many physical fields such as soil mechanics, tire mechanics and transmission system, the tractor traction performance was abstracted into four basic models, namely wheel-soil model, driving force model, slip rate model and tractive force model, in order to establish a general model for the whole machine traction performance prediction of four-wheel drive and twowheel drive tractors. In order to improve the prediction accuracy, the traction performance prediction optimization algorithm based on adaptive particle swarm optimization (APSO) was established with the overall machine slip rate as the optimization objective. Through on-line optimization, the accuracy and universality of the model were verified. In order to further verify its superiority and engineering practicability, a 105kW tractor of YTO was used as a test prototype to complete the offline test in the whole field test site. The experimental results showed that compared with the existing prediction models, the error of slip rate and rolling resistance of the APSO-based prediction method was 1.9% and 0.18kN, respectively. For two-wheel drive tractors, the corresponding errors were 2.7% and 0.25kN, respectively, and the accuracy was greatly improved.The general model of traction performance prediction for four-wheel drive and two-wheel drive tractors was studied, which had certain research significance in the fields of traction control and performance of wheeled tractors.