Abstract
With the deployment of smart grid technologies and Advanced Metering Infrastructure, demand side management via feedback is a subject of interest to utility companies and researchers for modeling consumer behavior. Home area devices, such as in-home displays and smart appliances, are being developed and implemented to achieve real-time feedback. Near real-time feedback provided via Advanced Metering Infrastructure is used to change the consumers’ electricity consumption behavior and to conserve scarce resources. Analyses on the effects of real-time feedback show improvements in energy conservation, but the improvements are not as expected. This is due to the fact that altering human behavior is not an easy task, and not all react in the same way to a similar feedback. Modeling consumer behavior helps to understand their reactions toward different types of feedback, and hence allows for the categorization of the feedback and its relevance to a specific consumer category. This research focuses on two main tasks: (1) modeling human behavior for residential electricity consumption, and (2) identifying different consumer behavioral categories and feedback sets for each consumer behavioral category. The first task addresses load profile (electricity consumption) data needed by electric utility companies for expansion planning, in view of the evolution of smart grid and distributed energy resource concepts. Conventional methods of collecting the load profile data, such as surveys and metering, are tedious and time-consuming activities. Consumer demand, as well as continuous technological evolution, contributes to rendering data obsolete in a short period of time. For these reasons, the conventional methods pose barriers for electric utility companies. In response, this research presents an innovative behavior model for generating electricity consumption load profiles. The model will be narrowed to focus on consumption-based decision-making that is related to human comfort and to the consumer’s state of mind. Based on fuzzy logic and activity graphs, the method requires minimum consumer data and can be easily updated to adapt to changes in technology. We demonstrate the accuracy of our model against real world data. The goal of the second task is to analyze behavior change by using a multi-agent-based system, in order to eventually improve energy consumption. Analyses of behavior change have been based on surveys or face-to-face interviews. These methods are inefficient and time consuming and do not always measure the impact on energy consumption, besides suffering from difficulties related to the sample size. Therefore, we propose a multi-agent-based system to study the effects of feedback on different types of consumers. An evaluation is performed on the factors that influence the changes in consumer behavior positively and negatively. Consumer categories are generated based on their behavioral responses to given feedback. The feedback methods that are most effective for each category are evaluated and identified.