The increasing integration of distributed renewable energy sources into the grid has made it more challenging to design a cyber security plan for active distribution networks. This article outlines a method for locating and identifying cyberattacks in distributed active distribution systems using electrical waveform analysis. A sequential deep learning model is the foundation for cyber attack detection since it can identify even the smallest incursions. In a two-stage approach, the targeted cyberattack is first localized inside the estimated cyberattack sub-region. We introduce a network splitting strategy for "coarse" localization of hierarchical cyberattacks, based on a modified version of spectral clustering. It is recommended to create several waveform parameters and apply a normalised effect score based on statistical metrics of the waves themselves in order to further localise the origin of a cyberattack. Lastly, a comprehensive quantitative evaluation based on two case studies shows that, in comparison to both traditional and cutting-edge approaches, the suggested framework produces reliable estimations. Suggested Search Terms: Support Vector Machine, Random Forest, Logistic Regression, Gradient Boosting, and Cyber Attack Detection.