An Effective Financial Crisis Early Warning Model Based on an IFOA-BP Neural Network

Abstract

Existing financial crisis models have weak effects and low accuracy in prediction. We propose a financial crisis early warning model based on the improved fruit fly optimization algorithm-back-propagation neural network (IFOA-BP). First, we select listed enterprises in the manufacturing industry as the research object and use quantum computing to select financial crisis indicators. In view of the shortcomings of the FOA, which has low convergence accuracy and easily falls into the local optimum, we propose the IFOA, which introduces orthogonal experimental arrays in the initialization of the population to improve the diversity of the solutions and optimizes the individual crossing to avoid the algorithm falling into the local optimum. Finally, the IFOA is used for the optimization of the BP neural network’s weights and thresholds to improve the model prediction performance. In the simulation experiments, we verify the IFOA’s performance with the benchmark function. In the simulation experiments, we verify that the IFOA has good performance, and the IFOA-BP model has a good advantage over the ACO-BP, PSO-BP, FOA-BP, WOA-BP, and CSO-BP models in terms of MSE, MAE, and MAPE indices.