Abstract: In order to accurately predict the trend of the development of production safety accidents in our country, an improved GM(1,1)-Markov prediction model was established based on the traditional GM(1,1) model and the Markov model, and the actual application of this improved model was discussed by using the original sequence about the numbers of production safety accidents from 2005 to 2018. This study differs from traditional grey residual correction theory, and it selected the relative error of the gray prediction result as the correction index, and applied the Markov model to optimized the prediction of the relative error state and the error symbol state respectively, and the accuracy of this model was evaluated by the average relative error and probability of small error. The results showed that the improved GM(1,1)-Markov model had a relative error of 3.0%, and compared with a single gray prediction model, the prediction error was reduced by 19.5%, which prediction accuracy was significantly improved, and the numbers of production safety accidents in China will reach 479 in 2019.

Key words: production safety accidents, prediction, GM(1,1), Markov chain