Cost-Effectiveness Evaluation of Standby Power Configuration for Cluster Data Centers Based on Multi-Objective Optimization Algorithm Combined with Deep Learning

Abstract

Cluster data centers serve as the core carriers of the digital industry, and their standby power systems directly determine the power supply reliability and overall operational costs of data centers. Traditional standby power configuration methods mostly rely on empirical selection, which fail to simultaneously address multiple indicators including construction costs, operational energy consumption and power supply efficiency, nor adapt to the dynamic operational characteristics of clustered and large-scale data centers. This paper combines multi-objective optimization algorithms with deep learning technologies to establish an evaluation system for standby power configuration of cluster data centers, and conducts quantitative analysis centered on two core dimensions: cost and efficiency. First, this paper sorts out the operational characteristics and configuration requirements of standby power for cluster data centers. Second, a multi-objective optimization model is constructed, and the calculation rules for cost and efficiency indicators are defined. Third, deep learning algorithms are adopted to complete load forecasting and parameter fitting, and the optimization algorithm is combined to solve the optimal configuration scheme. Simulation verification and comparative analysis are carried out based on measured data, with four statistical tables presenting detailed data of different configuration schemes intuitively. The research results indicate that the evaluation method integrating the two technologies can effectively balance the investment cost and operational efficiency of standby power, providing practical references for the planning, type selection and operation & maintenance of standby power systems in cluster data centers.