<p>Wireless Sensor Systems (WSNs) comprise dispersed sensors that monitor and hold physical boundaries, delivering valuable information. The distributed nature of WSNs eliminates the need for consistent data transmission to a centralized node, resulting in significant energy savings. However, the sustainability of WSNs is a primary concern due to the limited energy sources of battery-powered sensor nodes. Traditional centralized data mining techniques are unsuitable for WSNs due to the limited battery life and computational capabilities of sensor nodes. To solve this issue, a Distributed Data Mining (DDM) approach is developed by using deep learning techniques to distribute the data mining process across the network, reducing the burden on individual nodes and minimizing energy expenditure. This research provides a deep learning-based DDM approach to improve energy efficiency and load balancing at the WSN fusion center. The DDM model is implemented using Adaptive Cascaded Residual Long Short Term Memory (ACas-ResLSTM), which helps to integrate the network into multiple layers, in turn, it places the nodes. The parameters of ResLSTM are effectively optimized using the hybrid algorithm named Defined Random Number-based Coati with Dolphin Swarm Optimization (DRN-CDSO), which incorporates the Coati Optimization Algorithm (COA) and Dolphin Swarm Optimization (DSO). Due to this optimization, the proposed model helps to reduce the signaling overhead and average delay and maximizes the overall throughput. This deep learning-based approach is subjected to experiments to confirm its efficacy in deep learning-based DDM models for energy-efficient WSN assessment. The energy efficiency of the proposed technique is 14.5, which is enhanced than the other conventional approaches. Thus, the outcome demonstrates the promising and robust solution for maximizing the lifetime of sensor nodes and allowing for a scalable and reliable network in the field of WSN.</p>