Abstract: To ensure the final assembly quality of military aircrafts, helicopters and other large equipment, and the products can meet performance indicators, the systems must be analyzed and calculated. But for the large complex systems, due to their high dimensionality and complexity, it is relative difficulty to analyze directly, and it is good to use reduction method to deal with the problem in this case. The paper is aimed at reduction for diesel engine controller by using Singular Value Decomposition, Krylov subspace theory and the Least Squares method. Through the comparative study,we proved that Singular Value Decomposition (SVD) algorithm can be truncated according to the size of the singular value, and maintain the structural properties of the reduction system, but the calculation process is complex. The calculation of Krylov subspace reduction algorithm is small, but the error norm is large. While using the Least Squares method can be well combined with the advantages of these two methods, from the simulation results we can get the best result when use the third method.