Improvement of GM (1,1) model optimization based on the hill-climbing algorithm and its application in vehicle traffic accident studies

In order to accurately predict the change trend of the number of vehicle traffic accidents, this paper proposes an improved GM (1,1) model based on the hill-climbing method to solve the problem that the traditional GM (1,1) model has low accuracy in predicting data with large fluctuations. The heuristic method is used to optimize the parameter calculation method of the original model, improve the accuracy of curve fitting, and reduce the average relative residual error of the model prediction. On this basis, using the multiple sets of data of the number of real historical vehicle traffic accidents in the past ten years, the accident prediction accuracy of the traditional GM (1,1) model and the improved GM (1,1) model is compared. The experimental results show that the improved GM (1,1) model based on hill-climbing method proposed in this paper has higher fitting accuracy, and the average relative residual is lower than that of the traditional model. It is less affected by the fluctuation of data sequence and can better meet the needs of practical engineering. It can provide some support for the development of related application research and has certain practical value.