Behavioral Modeling and Response Mechanism of Residential Multi-Dimensional Electricity Use

To address the multi-dimensional electricity demand and bounded rationality of residential users, this paper proposes a behavioral modeling framework that integrates mental accounting theory with cumulative prospect theory, systematically characterizing electricity usage responses under different living scenarios. First, based on behavioral economics, a multi-dimensional demand structure is established, and corresponding load characteristic and comfort level models are developed. Subsequently, a prospect value evaluation method based on multi-dimensional mental accounts is proposed, along with value and weighting functions that reflect users’ subjective trade-offs between cost and comfort. For parameter identification, a hybrid approach combining the mixed Logit model and deep fuzzy clustering is employed to capture individual preferences. To enhance robustness and generalization in small-sample conditions, a Wasserstein Generative Adversarial Network (WGAN) and collaborative regression algorithm are further introduced. Empirical results demonstrate that the proposed model can effectively assess residential demand response potential, with strong predictive performance and practical applicability.