Multistage energy management system using autoregressive moving average and artificial neural network for day‐ahead peak shaving

Abstract

In this Letter, a multistage energy management system is developed and analysed incorporating the coordination between the tertiary stage and the primary stage. In the tertiary level, a day-ahead peak shaving strategy is developed using the autoregressive moving average and artificial neural network technique. These two techniques are used to predict customer's power demand and photovoltaic power generation, which are fed to a tertiary controller for the day-ahead power demand management. As the peak-power-demand management system is highly dependent on the demand-generation values, any fluctuations and errors in predicted values impact the performance of the peak shaving. The reference power generated from the energy management layer at the tertiary stage is communicated to the local inverter controller at the primary stage. The inverter implements a dq -current controller to track the reference power efficiently.