Building trust by design through explainable AI for resilient and cognitive smart grids

Traditional electrical power grids have historically faced challenges with operational reliability, stability, flexibility, and efficiency. Machine learning (ML) aids in enhancing smart grids (SGs) by enabling predictive maintenance, optimizing energy distribution, and ensuring efficient and stable grid operations. Despite the ubiquitousness of ML methods, these methods lack transparency and explainability for SG applications due to their non-linear multilayer structures. Thus, explainable artificial intelligence (XAI) emerged as the missing piece toward human-centered AI that prioritizes the humanistic design perspective. This study aims to explore the potential of XAI for SGs. A background review of the explainability paradigm and its urgent need for SGs is provided. In addition, multiple XAI models are discussed while citing their advantages and drawbacks. Next, the chapter discusses a wide spectrum of SG-based XAI applications. This chapter promotes the exploration of the emerging XAI paradigms, specifically for SG applications. With this study, the authors hope that this work will serve as a touchstone for theorists and practitioners alike, promoting trust in AI systems through enhanced transparency and ultimately enhancing the resilience of SGs.