装载机作业对象种类多、工况复杂、环境恶劣、载荷变化剧烈，准确预测作业阻力是实现自主作业的前提和关键。本项目在前期载荷谱采集与分析的基础上，进一步研究基于产品运行大数据的作业阻力关键参数挖掘、表征以及精确预测模型构建的理论与方法，为装载机自主作业奠定基础。具体工作包括：开展作业扭矩表征建模与评估方法研究，解决作业扭矩难以准确估计的问题；使用单目和双目视觉挖掘运行数据特征，并构建物料参数与运行数据特征的映射关系模型，丰富与拓展预测模型的可用特征；利用迁移学习进行载荷谱与产品运行大数据之间迁移建模，实现快速准确的作业阻力预测，解决基于载荷谱的作业阻力预测模型实际应用中精确度低、适用性小的问题，并探索与分析作业阻力与其影响因素之间的内在联系及关联度。研究成果可为挖掘机、推土机等其他工程机械的作业阻力预测建模提供思路和方法，为工程机械的自主作业提供决策依据，具有重要的应用价值和科学意义。

Wheel loaders are featured by various types of work objects, complicated operational conditions, harsh working environment, and dramatically-varied working loads. Therefore, accurate prediction of the working resistance force in advance is the premise to achieve autonomous loading operation. Based on our previous acquired load spectrum and corresponding analysis results, this project further introduces operational big data for identifying, characterizing and predicting key parameters that affect the working resistance force, which lays a foundation for intelligent wheel loaders. The research contents include: (1) characterize and evaluate the working torques to ensure the accuracy of the predictive model; (2) acquire operational data features using monocular and binocular visual analysis and build the relationship between operational data features and material parameters; (3) develop a predictive model using transfer learning from operational big data to solve the problem of low accuracy and less applicability of working resistance force prediction models based on load spectrum, and explore the intrinsic relationship and correlation between the working resistance force and the sensitive parameters. The research results can offer a practical method for working resistance force prediction of excavators, bulldozers and other construction machinery, and a basis for decision-making in autonomous loading control of construction machinery.