The PSO-SVR-based Prediction Method for Electric Energy Substitution in Fishery

Abstract

Following the proposal of "carbon neutrality and peak carbon emissions" goals, electrical energy substitution has emerged as a key strategy for sustainable development across various industries. As a crucial component of agricultural development globally, this paper introduces a particle swarm optimization vector regression method to predict the potential for electrical energy substitution in the fisheries sector. Initially, this article analyzes the key factors influencing the electrical energy substitution in fisheries from technological, economic, and policy perspectives. To explore how these various factors influence the potential for energy substitution, a Support Vector Regression (SVR) algorithm is applied. This SVR model's performance is subsequently improved by optimizing its parameters using Particle Swarm Optimization (PSO). Compared to a BP neural algorithm, the enhanced particle swarm optimization (PSO)-SVR model demonstrated markedly superior prediction accuracy and goodness-of-fit. Consequently, this model was effectively utilized to generate a forecast of the electrical energy substitution potential within China's fisheries in recent years. This study provides theoretical and data support for promoting electrical energy substitution in the fisheries sector and offers guidance for analyzing the potential for energy substitution.