Traffic Flow Prediction Model Using an Integrated Framework of Improved Intelligent Optimization Algorithms and Deep Learning Models

Abstract

Whale Optimization Algorithm (WOA) is a meta-heuristic optimization algorithm with fast convergence, which can well simulate the search behavior of humpback whales. In traditional WOA, local search and global search are carried out randomly, which leads to poor global search ability of the algorithm in the initial iteration and easy to fall into local optimum. In order to solve the shortcomings of traditional WOA, we propose an adaptive adjustment search option Probabilistic WOA algorithm (PWOA). And PWOA has carried on the function test and analysis of typical functions. In order to predict short-term traffic flow, the data is subjected to empirical mode decomposition (EMD) to obtain multiple eigenmode functions, and PWOA and deep belief network (DBN) are combined to predict each mode function sequence separately, and the prediction results are superimposed to obtain the final Predictive value. Experimental data shows that the EMD-PWOA-DBN model has higher prediction accuracy than the DBN and PWOA-DBN models.KeywordsWhale optimization algorithmEmpirical mode decompositionDeep belief networkTraffic flow prediction