Abstract In this study, a hybrid traffic signal control (HTSC) system based on phase and time optimization was developed. The Flower Pollination Algorithm (FPA) approach was used for phase optimization, while Type-2 Fuzzy Logic, optimized with the Crow Search Algorithm (CSA), was utilized for time optimization. The hybrid system's performance was investigated using nine different traffic conditions and four different intersection geometries. The hybrid system was compared with three controller systems which are a fixed-time signal controller, a signal controller based on the FPA approach (FPA_TSC), and the optimized Type-1 fuzzy logic signal controller (Type-1 FL-TSC). The HTSC approach achieved the best performance with about 32% improvement over the fixed-time traffic controller and it showed 5% and 6% better performance than the FPA_TSC and Type-1 FL-TSC, respectively. Considering the performance of the new hybrid system, it is effective in minimizing delays and driver dissatisfaction occurring from signalization. It also contributes to the reduction of emissions and fuel consumption. The HTSC approach can be used as an alternative signal control method in the control of intersections with high traffic volume due to its fast and effective performance.

Abstract This article presents a new methodology for estimating delay parameter k, and proposes analytical models which are used artificial intelligence technique for signalized intersections that considers the variation in traffic flow with under‐saturated and over‐saturated conditions. The delay parameter k has been expressed as a function of the degree of saturation in the proposed analytical models. Using the Bezier Search Differential Evolution algorithm (BeSD) algorithm, four different model forms were developed separately for under‐saturated conditions (x <1) and over‐saturated conditions (x > =1). Among the model forms developed as linear, quadratic, power, and logarithmic, the quadratic model presented the best results in both traffic conditions. In the validation of the models, a total of 140 different traffic conditions were determined, 56 of which cover the under‐saturated and 84 the over‐saturated traffic conditions. According to the statistical results, using k values depending on the proposed model instead of using a constant k value (0.5) provides 1.5 and 4.3 improvements for RMSE and MAPE values in under‐saturated traffic conditions respectively, while these improvements in over‐saturated traffic conditions have reached 9.5 and 6, respectively. As a result, using k values depending on the proposed model will be effective in obtaining a more accurate delay value. This effect is more evident, especially in over‐saturated traffic conditions.

Abstract A fixed‐time traffic control system is widely used to manage the traffic flow at intersections. Cycle length has an important effect on the performance of the fixed‐time control system and the Webster model is widely used in the literature to determine the cycle length. However, when the traffic flow ratio (Y) approaches 1, the Webster model loses its effectiveness and cannot determine the cycle length when Y is above 1. In the scope of this study, it is aimed to develop models that can predict cycle length for all traffic conditions in which are Y ≤ 1 and Y ≥ 1. Two different heuristic algorithms, atom search optimization (ASO) and grasshopper optimization algorithm (GOA) were used in the development of models and the performances of these algorithms were demonstrated. The efficiency of the developed models has been demonstrated by comparing them with both Webster's model and VISTRO optimization program. Models developed in exponential, power, and quadratic forms have been able to predict cycle lengths with lower delay values by showing better performance than Webster and VISTRO. In addition, statistical results show that the ASO approach is more successful than the GOA approach.

Abstract The imbalance between modes of transport in our country appears as the most important problem. Therefore, in air transportation, which has a significant increasing trend, estimating the passenger demand with directly related parameters and novel algorithms is important for Turkey. In this study, different prediction models were developed applying for the first time with five different meta‐heuristic algorithms which are Flower Pollination Algorithm (FPA), Artificial Bee Colony Algorithm (ABC), Crow Search Algorithm (CSA), Krill Herd Algorithm (KH), and the Butterfly Optimization Algorithm (BOA) to estimate Turkey's air transport demand. While developing the models, Fuel Price, Gross Domestic Product per Capita, Seat Capacity, and Annual Fuel Consumption were selected as the model parameters. Although each model developed using different approaches is applicable, quadratic and power models developed using CSA showed the highest performance. For this reason, future projections were based on these models. Air transport passenger demand was examined using two scenarios in a process until 2035. In the first scenario, according to model forms, Turkey's future air transport passenger demand will reach about 460 and 490 million passengers, respectively. In the second scenario, the number of passengers will reach approximately 375 and 660 million for quadratic and power models, respectively. The results of this study will contribute to the evaluation of the current investment plans and the development of strategic plans that will meet the demands. Additionally, they will help take necessary measures and introduce some necessary regulations to ensure the income and expense balance so that the efficiency of airline companies can be improved.