基于数据驱动的履带车辆冻土路面牵引性能分析与预测

Data-driven Analysis and Prediction of Traction Performance of Tracked Vehicles on Frozen Ground

摘要:

松软路面冻结会对履带式农用机械机动性能产生重要影响，分析环境因素对履带车辆冻土路面牵引性能影响十分必要。依据气温、冷冻时长、降水量等对土壤冻结有关键影响的环境数据，选取冷冻温度、含水率、冷冻时间等环境变量，制作试验冻土土样，通过直剪试验测量土样抗剪强度指标，分析指标变化趋势;建立履带车辆牵引力模型，分析环境变量对冻土路面牵引性能的影响;基于数据驱动算法对试验数据进行学习，实现环境变量对强度指标的精准预测，最终实现由环境变量到牵引性能的端到端预测。研究结果表明，环境变量对冻土抗剪强度指标有较大影响，影响机理复杂，对履带车辆牵引性能也有较大影响。通过数据驱动算法能够精准预测抗剪强度指标，实现对履带车辆冻土牵引性能预测。

Abstract:

When travelling on soft ground, tracked agricultural machinery is highly prone to slipping and becoming bogged down, which severely limits its maneuverability. The frozen surface has a significant impact on the maneuverability of tracked agricultural machinery on soft ground;therefore, it is essential to analyze the influence of environmental factors on the traction performance of tracked vehicles on frozen ground, Utilizing environmental data such as air temperature, duration of freezing, and precipitation, which had a key influence on soil freezing. Environmental variables, including freezing temperature, moisture content, and freezing duration were selected. Test frozen soil samples were prepared, and their shear strength was measured by using direct shear tests to analyze trends in these indicators. A traction model for tracked vehicles was established to analyze the impact of environmental variables on traction performance on frozen ground. Using data-driven algorithms to analyze the experimental data, accurate predictions of strength indices were achieved based on environmental variables, ultimately realizing end-to-end prediction from environmental variables to traction performance. The results indicated that the three environmental variables, moisture content, freezing temperature and freezing duration had a coupled effect on the shear strength of frozen soil, with freezing duration having a relatively minor influence;both freezing temperature and moisture content exerted a significant impact on the indices. Traction exhibited a distinct critical low-point region with respect to temperature and stabilized after approximately 8 hours with respect to time;meanwhile, the higher the moisture content was, the greater the loss of traction at low temperatures compared with ambient temperatures. By training the data-driven algorithm on experimental data, precise predictions of strength indicators based on environmental variables were achieved;a prediction framework for the traction performance of tracked vehicles on frozen ground based on environmental variables was established, enabling end-to-end precise prediction from environmental variables to traction performance.