Prediction of Nuclear Weapon Quantities Based on BP Neural Network and Time Series ARIMA Models

This paper focuses on the number of nuclear weapons and their potential impacts in depth. Using BP neural network model and time series ARIMA model, the nuclear blast assessment model was constructed. Firstly, by using MATLAB to process the data, the classification, screening and sorting of each labeled data were completed, and the countries with the largest range of changes in the growth rate and decline rate were identified; secondly, the optimal solution was obtained through the time series ARIMA model of machine learning, and a simple time prediction series model predicting the number of nuclear bombs possessed by each country in the next one hundred years was completed. Finally, this paper predicts that there will be 5,789 nuclear bombs in the world.