Prediction and Optimisation Models for Carbon Trading Prices - A Study Based on BP Neural Networks and Genetic Algorithms

This paper proposes a novel framework for an in-depth examination of carbon storage, integrating a genetic algorithm-backpropagation (GA-BP) approach with a greedy algorithm, based on empirical research into the optimization methods that combine genetic algorithms, neural networks, and greedy algorithms. A carbon prediction model utilizing a BP neural network is developed, focusing on domestic and international indicators, technical indicators, and exchange rates, alongside lagging macro indicators. Additionally, an optimization model for carbon storage prediction is established using a GA-BP neural network, enhanced and advanced through the LSTM method. The findings indicate that the model exhibits substantial robustness and attains relatively stable performance across varied market conditions, thereby enhancing efficiency and accuracy, fostering further advancements in carbon storage, offering strategic guidance for producers, and advancing traditional carbon storage towards a more efficient, transparent, and intelligent future.