A persuasive multi-agent simulator to improve electrical energy consumption

ABSTRACT Gathering information on electric energy consumption is important for companies and institutions aiding in making plans. Multi-agent systems can aid such companies in understanding the different behaviours presented within the context of a society constituted by autonomous entities. We propose a multi-agent simulator capable of emulating different profiles of consumers and equipment. Our approach differs from others in three fundamental aspects: i). it models parameters that are difficult to predict, which are calibrated using values provided by available records; ii). it can be applied to different simulation environments involving various degrees of scalability and heterogeneity of profiles both of individual behaviour and household appliances; iii). it simulates the complexity of the system, providing estimates through the simulation of the interaction between users and electrical appliances. The results obtained by the model were similar to the numbers provided by the literature, which suggests the validity of the proposed approach.