Improved Garson algorithm based on neural network model

The evaluation of input factors of complex system is a hot and difficult point in the sensitivity analysis. In this paper, the Garson algorithm based on artificial intelligence is studied and the original Garson algorithm accuracy is not high. Therefore, an improved Garson algorithm is proposed and the input factors are introduced into the Garson algorithm. At the same time, the original local sensitivity analysis algorithm is improved as the global sensitivity analysis algorithm and it increases the accuracy and stability of the Garson algorithm. Through the typical benchmark test function simulation, the experimental results show that the improved Garson algorithm has higher accuracy and stability in the evaluation of sensitivity coefficient. Finally, the improved Garson algorithm is applied to evaluate the input factors of the plate-fin heat exchangers. It shows that the IGarson algorithm is more feasibility and effectiveness.