Abstract: To solve the problem that the incremental PID search algorithm (PSA) suffers from poor global search ability and is prone to falling into local optima, we propose an elite group-guided PID search algorithm with dynamic perception and bidirectional search (DPSA). The proposed algorithm introduces three key improvements. Firstly, it incorporates a dynamic sine factor-controlled elite group-guided mechanism that extracts information from high-quality individuals, enhancing global search capability. Secondly, it integrate a fitness landscape-aware dynamic perception-based reward and penalty factor \varepsilon to adaptively balance between exploration and exploitation, which improves its adaptability to different optimization problems. Finally, it employs a bidirectional search strategy for weak subgroups based on cosine similarity to reduce the number of ineffective individuals while enhancing the algorithm’s ability to escape local optima and conduct refined searches. Simulation experiments on the CEC2017 benchmark functions demonstrate that DPSA achieves an average fitness superiority rate of 79.31% in the 100-dimensional CEC2017 test set. Additionally, statistical analyses, including Friedman tests, Wilcoxon rank-sum tests, and stability analysis, show that DPSA significantly improves both convergence speed and accuracy, outperforming other optimization algorithms in solution quality and stability. Furthermore, its application to 19 real-world engineering problems and the node coverage problem in 3D wireless sensor networks (WSN) further validates its effectiveness and applicability.