PROJECT TITLE :

Short-Term Traffic Flow Forecasting Using Ensemble Approach Based on Deep Belief Networks

ABSTRACT:

Transportation services are playing an increasingly important role in people's day-to-day lives, and they bring a lot of benefits to individuals as well as to the growth of the economy. The unpredictable and unpredictable nature of traffic flows, on the other hand, makes it difficult to effectively provide transportation services to customers. This, in turn, limits the effectiveness of those services. To realize the stability of intelligent transport systems and to ensure that traffic scheduling is done in an efficient manner, accurate traffic flow forecasting has become the primary and most important task. In this paper, we investigate the use of an ensemble approach that is based on deep belief networks for forecasting short-term traffic flow. The data on traffic flow, which is collected from the real world, is then decomposed into several Intrinsic Mode Functions (IMFs), and a residue with EEMD is calculated (Ensemble Empirical Mode Decomposition). The essential feature subset is then extracted for each component using the mRMR (minimum Redundancy Maximum Relevance Feature Selection) method, taking into consideration the day's properties and the weather conditions. In addition, each component undergoes training with DBNs (Deep belief networks), and the results of each component's forecasting are finally compiled into the output of the ensemble model. According to the findings, the proposed method achieves a significant performance improvement in comparison to the use of a single DBN and the other methods that were chosen.