Smart grids (SGs) are crucial to the efficiency and sustainability of modern energy systems. As the world’s population continues to increase, so does the need for energy, and traditional energy systems are struggling to keep up. In this context, this study reviews the possibilities of deploying machine learning (ML) on wireless sensor networks (WSNs) in smart grid systems. In several ways, SGs may gain from combining WSNs with ML, including enhance system reliability, sustainability, improve fault detection, and increase energy efficiency. This paper offers an extensive review of pertinent research emphasizing the use of supervised, unsupervised, and reinforcement learning approaches. The evaluation contains 234 peer reviewed articles from highly regarded academic journals and conferences covering the years 2017 through 2024 which depict the effectiveness of supervised techniques on WSNs in the field of SGs. In addition the paper presents set of the most usable datasets in the field of WSNs and SGs, and introduces a comparison between our paper and relevant surveys. The study also analyses the opportunities and challenges related to the application of WSNs and ML in SGs and offers possible research directions. Overall, the study makes it clear that combining WSNs with ML may significantly contribute to the creation of smart grid systems that are more effective, dependable, and sustainable.