This article is designed to help in the contribution Dynamic Multi-Objective Evolutionary Algorithms (DMOEAs) can effectively solve Dynamic Multi-Objective Optimization Problems (DMOPs) by extracting knowledge from historical environments to predict new solutions. Currently, most existing prediction-based DMOEAs rely on information from the previous two historical environments and lack dynamic maintenance and selective reuse of the extracted historical knowledge. To address these issues, this paper proposes a dynamic multi-objective evolutionary algorithm based on adaptive fusion prediction of multi-source knowledge, named MSFP-DMOEA. The framework captures nonlinear change trends in the decision space through a Dynamic Feature Projection Network (DFPN) and systematically stores and retrieves optimization patterns in historical environments using a Multi-Modal Knowledge Base (MultiModalKB). Finally, a multi-source knowledge fusion prediction mechanism is adopted to weighted fuse the recent environment trend prediction and historical similar pattern prediction, generating high-quality initial populations. Meanwhile, dynamic maintenance of historical knowledge is achieved through modal update, merging and elimination mechanisms. Experimental results on multiple benchmark test problems demonstrate that MSFP-DMOEA significantly outperforms several state-of-the-art dynamic multi-objective optimization algorithms in terms of convergence, diversity and stability.