Reservoir Porosity Prediction Model based on Improved Shuffled Frog Leaping Algorithm and BP Neural Network

Abstract

A reservoir porosity prediction model based on an improved shuffled frog leaping algorithm (SFLA) and BP neural network (BPNN) is proposed. Aiming at the problem that the BPNN is sensitive to the initial weight and easy to fall into local optimum, an algorithm called SFLA_RGC is proposed. Firstly, the roulette selection mechanism is introduced in the process of dividing subgroups to improve the selection probability of elite individuals; secondly, genetic coding is carried out by making full use of the effective information such as the global and local optimum of the population. Then, the SFLA_RGC algorithm is verified on 8 benchmark functions and compared with 3 optimization algorithms, and experimental results show its good performance. Finally, the SFLA_RGC algorithm is applied to the optimization of the initial weights and thresholds of the BPNN, and a new model (SFLA_RGC_BP) is proposed. The porosity prediction performance of the model was compared with five methods and results show the SFLA_RGC_BP model has higher training accuracy, stability and faster convergence speed.