Novel demand response-based stochastic optimisation in a grid-connected microgrid considering outlier detection and imputation

To address the challenges in renewable generation, the proposed work presents a risk-based stochastic optimisation approach. This method considers uncertainties by employing Monte Carlo simulations for scenario generation and scenario reduction employing K-means clustering. The unsupervised K-means clustering technique is particularly effective for outlier detection and imputation, ensuring more reliable data for decision-making. Additionally, the authors proposed a novel demand response programme called Dynamic Sustainability Pricing (DSP), designed to reduce both the environmental impact and the total cost of operating a microgrid. To evaluate the performance of the DSP, several indices such as the personalised sustainability index, the carbon intensity index, the renewable energy participation index, and the environmental impact index have been proposed. It is evident that the DSP has caused the PAR to decrease from 5.73 to 4.98 and that the load factor's percentage increase from before to after the DSP application has improved to 3.603%. The improvement in LF from 0.48 to 0.51 is the result of DSP employment. In addition, the TCPI was raised from 1 to 1.148, which helps consumers by lowering the overall consumer tariff and raising the profit index. The PSI increased from 3.3 to 3.069, indicating that RES's participation was maximized. © 2024 Informa UK Limited, trading as Taylor & Francis Group.