With the progress of science and technology, data fusion has shown great application potential in problem solving in many fields, especially in the application of partial differential equations in physical science and engineering technology. In this article, a new type of partial differential equation data fusion algorithm is studied. The algorithm is based on D-S (Dempster-Shafer) evidence theory and fuzzy mathematics framework. By introducing fuzzy logic to expand and optimize evidence theory, it effectively handles the uncertainty and ambiguity in data fusion. In addition, the algorithm combines the mathematical characteristics of partial differential equations, and significantly improves the accuracy and stability of fusion results by constructing an optimized weight allocation mechanism and iterative solution strategy. In order to evaluate the effect of PDE data fusion algorithm based on D-S evidence theory and fuzzy mathematics in practical applications, the article selects a specific physical or engineering problem, that is, a fault diagnosis prediction model for multi-sensor data fusion. During t1 to t4, the device is in normal condition. At t5 and t6, the device enters a warning state and some parameters approach or exceed the safety threshold. This research not only enriches the theoretical system of data fusion algorithm, but also provides a new tool for the accurate solution of complex systems, which has important theoretical and practical application value.