Unidirectional and Bidirectional LSTM Models for Short-Term Traffic Prediction

Abstract

This paper presents the development and evaluation of short-term traffic prediction models using unidirectional and bidirectional deep learning long short-term memory (LSTM) neural networks. The unidirectional LSTM (Uni-LSTM) model provides high performance through its ability to recognize longer sequences of traffic time series data. In this work, Uni-LSTM is extended to bidirectional LSTM (BiLSTM) networks which train the input data twice through forward and backward directions. The paper presents a comparative evaluation of the two models for short-term speed and traffic flow prediction using a common dataset of field observations collected from multiple freeways in Australia. The results showed BiLSTM performed better for variable prediction horizons for both speed and flow. Stacked and mixed Uni-LSTM and BiLSTM models were also investigated for 15-minute prediction horizons resulting in improved accuracy when using 4-layer BiLSTM networks. The optimized 4-layer BiLSTM model was then calibrated and validated for multiple prediction horizons using data from three different freeways. The validation results showed a high degree of prediction accuracy exceeding 90% for speeds up to 60-minute prediction horizons. For flow, the model achieved accuracies above 90% for 5- and 10-minute prediction horizons and more than 80% accuracy for 15- and 30-minute prediction horizons. These findings extend the set of AI models available for road operators and provide them with confidence in applying robust models that have been tested and evaluated on different freeways in Australia.

Development and Evaluation of Bidirectional LSTM Freeway Traffic Forecasting Models Using Simulation Data

“…It can be noticed that there is more focus in these studies on exploring the spatial and temporal traffic features when predicting traffic conditions using the BiLSTM model [33,34,35,36,37]. However, few studies have explored the feasibility of this type of model to be validated or transferred (without retraining) to an independent dataset from a different freeway [38] or in the case of this paper, validate the model against future traffic scenarios where the demand is expected to increased to up to 80% in the future. Also, this paper tests the model on multiple prediction horizons on multiple traffic variables such as speed, flow and occupancy using data generated from a calibrated freeway model which hasn't been established in any previous literature on the topic.…”

“…Unidirectional LSTM received considerable attention in recent years for its superior performance compared to the state-of-art Recurrent Neural Networks (RNNs). Even though RNNs provide good accuracy, they have been found to underperform for long-term memory as RNNs are unable In these models, the following formulae are used to calculate the predicted values [38,50]:…”

“…The "input gate" specifies new input to the cell state, the "forget gate" determines how much of the prior memory values should be removed from the "cell state" [38,50] and the "cell state" and "output gate" of the LSTM at time t is calculated as follows: Ct = ft⊙ct−1+it⊙gt (7) Ht = ot⊙σc(ct) (8) where ⊙ denotes the Hadamard product (element-wise multiplication of vectors).…”

“…The models were trained on the first 60% of the sequence and tested on the last 40%. To prevent model overfitting, the training/testing data were standardised to have zero mean and unit variance [38]. The LSTM networks were created using four layers: Sequence Input Layer (number of Features =1), Uni-LSTM/ BiLSTM Layers (number of Hidden Units =300), fully Connected Layer (number of Responses=1) and a Regression Layer.…”

“…The model hyper parameter settings are presented in Table II. The same parameters were optimised by the authors for [38] and they achieved high prediction accuracies. The tanh and sigmoid functions were used for state and gate activation functions, respectively.…”