Optimisation Study of Pyrolysis Based on Genetic Algorithm and BP Neural Networks

Focusing on the analysis of joint pyrolysis process with machine learning algorithms such as genetic algorithms and BP neural networks, this paper firstly revealed a significant regression relationship between n-hexane insolubles and tar and coke residue yields, while there was no significant correlation with aquatic product yields, using ridge regression analysis. Next, a multiple linear regression model was constructed by applying the least squares method to analyse the ration interaction effect in order to optimise the thermal mixing ratio and maximise product utilisation and energy efficiency. In addition, the parameters were optimised by genetic algorithm to determine the optimal mixing ratios, and the differences between the experimental and theoretical calculated values were analysed using the Wilcoxon signed rank test. Finally, the yield was predicted based on a BP neural network model. This study demonstrates the potential application of advanced computer algorithms in energy efficiency optimisation, providing important theoretical support and practical guidance for related fields.