<p>Software-defined networks (SDNs) improve network flexibility and dynamic management by separating the control and data layers. However, the centralized nature of this architecture has generated challenges with load balancing among controllers. Improper load distribution can increase response time, reduce network efficiency, and elevate processing overhead. This study presents a new SDN controller design that improves load balancing in these networks. The suggested controller structure contains four key layers: data gathering, analysis and prediction, decision-making, and execution and control. In this architecture, the analysis and prediction layer uses a bidirectional long short-term memory-based prediction module that can improve future load prediction accuracy through future–past and past–future analyses. Simulation findings in the OMNeT++ environment show that this methodology significantly outperforms network performance compared to earlier methods. An average of 30.9% reduction occurred in response time, and processing overhead was improved by 5.34%. Furthermore, load distribution between controllers was improved by an average of 6.26%. These results show that the proposed strategy is highly effective in improving load balancing and enhancing the efficiency of SDNs.</p>