CNN-RNN Algorithm-based Traffic Congestion Prediction System using Tri-Stage Attention

Abstract

Most people consider traffic congestion to be a major issue since it increases noise, pollution, and time wastage. Traffic congestion is caused by dynamic traffic flow, which is a serious concern. The current normal traffic light system is not enough to handle the traffic congestion problems since it functions with a fixed-time length strategy. Methodology: Despite the massive amount of traffic surveillance videos and images collected in daily monitoring, deep learning techniques for traffic intelligence management and control have been underutilized. Hence, in this paper, we propose a novel traffic congestion prediction system using a deep learning approach. Initially, the traffic data from the sensors is obtained and pre-processed using normalization. The features are extracted using Multi-Linear Discriminant Analysis (M-LDA). We propose Tri-stage Attention-based Convolutional Neural Network- Recurrent Neural Network (TA- CNN-RNN) for predicting traffic congestion. Results: To evaluate the effectiveness of the proposed model, the Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) were used as the evaluation metrics. Conclusion: The experimental trial could extend its successful application to the traffic surveillance system and has the potential to enhancement an intelligent transport system in the future.

A Novel Fault Diagnosis Method for a Power Transformer Based on Multi-Scale Approximate Entropy and Optimized Convolutional Networks

“…CNNs, using convolutional and pooling layers, automatically learn the local and global features from the input data. It can provide effective representation of images, sequences, and so on [ 20 ]. The strength of CNNs lies in their efficient processing of complex data and feature learning capabilities.…”