Because of the dynamic and dense nature of contemporary network traffic, congestion control presents a significant difficulty in wireless sensor networks. Current approaches, which mostly rely on heuristic-based & conventional routing techniques, frequently lack flexibility and effectiveness, which results in higher latency and worse service quality. In order to overcome these constraints, this study presents a novel, integrated method that combines the Bacterial Foraging Optimiser (BFO), ensemble classification using Multi-Layer Perceptron (MLP) and Logistic Regression (LR), and Q-Learning. Together, these methods improve path optimisation and congestion detection. The suggested approach mimics the natural foraging behaviours of bacteria by leveraging the Bacterial Foraging Optimiser for path finding and its bio-inspired mechanics. Consequently finding less crowded paths by effectively traversing the solution space. An ensemble classification approach that combines MLP and LR is used in addition to this. While LR offers insightful probabilistic outputs that improve overall accuracy in forecasting crowded nodes and pathways, MLP is the best at capturing complicated nonlinear relationships inside data. These classifiers’ integration aids in accurately detecting congestion under a range of network conditions. Additionally, Q-Learning is used to dynamically optimise routing choices according to current network conditions, enabling scalable and adaptive congestion control. Using a reward structure adjusted for network performance measures like throughput and packet loss, the Q-Learning algorithm adjusts its policy to favour routes that reduce anticipated congestion in the future. When compared to traditional approaches, this integrated strategy has a considerable impact on throughput and latency reduction. The suggested approach exhibits notable gains in network stability and service quality through thorough performance tests, in addition to greater scalability and robustness across a range of network situations. This model represents a critical breakthrough in network management technology and a major step forward in the creation of strong, effective, and adaptive congestion control techniques for wireless networks. WFQ initially uses characteristics such as protocol type, source/destination IP addresses, and port numbers to classify traffic flows according to priority. Standing in line Following classification, the flows are allocated to various queues or sub-queues. The process of weighting A weight that corresponds to each queue's priority level is assigned. Queues with higher priority will be served more frequently because they are given higher weights. Scheduling: The scheduler determines the sequence in which packets from various queues are sent out during transmission by using the set weights. Bandwidth Allocation: WFQ distributes bandwidth to each queue based on its weight during times of congestion, granting higher-priority flows more bandwidth while guaranteeing that low-priority flows continue to get a part.