Research on Heavy Truck Load Estimation Based on Artificial Intelligence Algorithm

For the estimation of heavy truck load, this article will conduct research as a regression scenario. Firstly, it is found that the correlation between engine data and load can be enhanced by filtering, and the filtering principle is that the correlation sum of feature data is the largest. Secondly, the tree-based GBDT (Gradient Boosting Decision Tree) algorithm is selected to study the impact of data processing methods and parameter optimization on the estimation accuracy. It is found that the mean error decreases from 3.21 tons to 1.11 tons under the optimal filtering constant, and from 1.11 tons to 0.0069 tons after hyper-parameters optimization. LDA (Linear Discriminant Analysis) feature conversion method is the best; Finally, based on LDA feature transformation, the applicability of BP neural network algorithm and SVM (Support Vector Machine) algorithm to load estimation is studied and compared. It is found that the average error of BP (Back Propagation) neural network is 0.76 tons, the memory occupation is the smallest, the error distribution is the most discrete. The load estimation has almost no time series attribute. The mean error of SVM algorithm is 0.85 tons, the memory occupation is medium, and the error distribution is between GBDT and BP. SVM model is suitable for ECU terminal integration. GBDT model is suitable for cloud integration.