Optimization driven cellular automata for traffic flow prediction at signalized intersections

The traffic flow prediction using cellular automata (CA) is a trendy research domain that identified the potential of CA in modelling the traffic flow. CA is a technique, which utilizes the basic units for describing the overall behaviour of complicated systems. The CA model poses a benefit for defining the characteristics of traffic flow. This paper proposes a modified CA model to reveal the prediction of traffic flows at the signalised intersection. Based on the CA model, the traffic density and the average speed are computed for studying the characteristics and spatial evolution of traffic flow in signalised intersection. Moreover, a CA model with a self-organizing traffic signal system is devised by proposing a new optimization model for controlling the traffic rules. The Sunflower Cat Optimization (SCO) algorithm is employed for efficiently predicting traffic. The SCO is designed by integrating the Sunflower optimization algorithm (SFO) and Cat swarm optimization (CSO) algorithm. Also, the fitness function is devised, which helps to guide the control rules evaluated by traffic simulation using the CA model. Thus, the cellular automaton is optimized using the SCO algorithm for predicting the traffic flows. The proposed Sunflower Cat Optimization-based cellular automata (SCO-CA) outperformed other methods with minimal travel time, distance, average traffic density, and maximal average speed.

The Extent of Capacity Loss Caused by Rainfall at Signalised Intersections

Background: Even though their physical characteristics exert a constant influence on capacity and saturation flows, signalized intersections are fixed facilities not affected by rainfall. Whilst traffic conditions with varying effects can be regulated, rainfall conditions cannot be regulated but compensated for by warning drivers to reduce speed. Speed reduction has an impact on signalised intersection capacity, whilst signalised intersection capacity is a function of saturation flow, effective green, and cycle time. In this paper, a capacity loss is the differential percentage between ‘with and without’ rainfall scenario. Aim: The paper investigated the extent of capacity loss caused by rainfall at signalised intersections. Methods: In Durban, South Africa, rainfall data were collected, collated, and correlated with traffic data in a 'with and without' rainfall intensity study. Rainfall intensity was classified according to the rate of precipitation as follows; rainfall intensity(i): light rain (i <2.5mm/h); Moderate rain (2.5mm/h ≤ i < 10mm/h), and heavy rain (10 ≤ i ≤ 50mm/h) as prescribed by the World Meteorological Society. Results: Empirical results show that rainfall intensity has an effect on road capacity at a signalised intersection. Generally, for the vehicles going straight, light rain caused a 4.25% capacity loss; moderate rain 9.18% while heavy rain caused an 11.53% capacity reduction. With right-turning vehicles, light rain caused 7.38% capacity loss; moderate rain caused 14.3%, while heavy rain accounted for 19.15% capacity reduction. Conclusion: The paper concluded that rainfall at signalised intersections would cause an anomalous capacity reduction. Since the database for the study is small, the paper advocates for further studies based on a broader database to include yellow interval time.

The effect of lane blockage on signalised intersection performance -analysing and modelling

Unexpected stops or entry/exit manoeuvres of vehicles on the road may cause the related lane to become blocked. When this blocking happens in a signalised intersection zone, it also affects intersection performance. Determining the extent of this effect will assist traffic engineers with intersection design and performance analysis. In this study, the effects of Lane Blockage (LB) on intersection performance under various traffic conditions were analysed according to two performance criteria. ANN (Artificial Neural Network) models were also developed to enable the prediction of intersection performance. As a result of the analysis, it was clearly determined that the effect of LB on intersection performance was limited at v/c <0.5. However, it was determined that the intersection performance may decrease between 10% and 110% under the condition of 0.5 < v/c, depending on the LB frequency and duration. Additionally, the developed ANN models have R > 0.95 and will therefore be useful in LB-related intersection performance analysis.