Hybrid Time Series Forecasting for Real-Time Electricity Market Demand Using ARIMA-LSTM and Scalable Cloud-Native Architecture

Abstract

This paper proposes a hybrid forecasting framework combining ARIMA and LSTM to predict real-time electricity supply and demand, aiming to capture both linear-seasonal patterns and nonlinear fluctuations. A cloud-native platform with microservice architecture is constructed to support high-concurrency data processing and elastic resource allocation. Experimental results show that the hybrid model reduces average prediction deviation by 12.5% compared to traditional methods, with 92.3% accuracy. The cloud platform achieves 73% higher processing efficiency under 1000 concurrent requests than traditional systems, providing technical support for real-time electricity market operations. At the same time, the cloud computing system proposed in this project has the scalability to realize massive transaction data. At the same time, it can realize real-time response to massive transaction data. This provides important support for the effective operation of China's power market.