Securing cyber-physical systems (CPS) that incorporate Internet of Things (IoT) capabilities presents challenges due to the potential incompatibility of security mechanisms built for conventional information/operational technology (IT/OT) technologies within the context of CPS environments. This paper presents an approach for detecting and attributing ensemble attacks in Cyber-Physical Systems (CPS), with a particular emphasis on Industrial Control Systems (ICS). At the initial level, a decision tree is employed in conjunction using a unique collective deep representation-learning model to identify assaults in unbalanced ICS situations. A second-level attack identification framework is developed using a collective deep neural network. The suggested model is estimate by means of real-world data sets gathered from wastewater treatment facilities and gas pipelines. The findings specify with the purpose of the projected model demonstrate better presentation evaluate to added competing strategies that possess similar computational challenges.