Gun Life Prediction Model Based on Bayesian Optimization CNN-LSTM

Abstract

Composite steel is the most commonly used material for artillery barrels. Ablation and wear of the steel material during artillery firing affect the life of the barrel. In this paper, we propose a new model that combines the advantages of convolutional neural network (CNN) and long and short-term memory network (LSTM) in feature extraction and memory prediction, respectively, using Nadam (Nesterov-accelerated Adaptive Moment Estimation) algorithm and Bayesian Optimization (BO) to optimize the model parameters. The improved accuracy compared to other prediction models demonstrates the feasibility and superiority of the model.

Keywords:

• CNN-LSTM
• Nadam
• Bayesian Optimization
• service life prediction
• neural network