Research On Improved Model Of Performance Evaluation Index And Multi-objective Optimization Signal Timing Of Intersection

As the intersection is one of the important parts of urban road network, it is thebottleneck in urban road network control system. Improving signal control timingscheme which based on the mastering intersection traffic condition is the most directand effective way to increase intersection operating efficiency. Meanwhile, theintersection performance evaluation index can evaluate the intersection traffic conditiondirectly and effectively. Therefore, it is of important theoretic value and practicalsignificance to study the intersection performance evaluation index and the signaltiming method.In this thesis, with the urban single-point intersection as the research object, aninnovative calculating method which could improve delay time and parking ratio wasestablished as the result of the phenomenon that the existing intersection performanceevaluation index approach could not estimate the actual intersection traffic conditionprecisely. Meanwhile, considering the low operation efficiency and the bad solution setsdistribution of the algorithm for solving the multi-objective signal timing model, animproved optimizing algorithm was presented and was applied to solve the signaltiming model.Firstly, the intersection signal control parameters and performance evaluationindex are introduced. According to analyze the relationship between evaluation indexand signal cycle length, it comes to a conclusion that the delay time and parking ratioare two important and representative intersection performance evaluation indices, andwe consider the two indices as control objectives of the signal timing model. Byanalyzing the classical models of calculating the delay time and parking ratio, weconclude that the phase flow is a significant parameter in these models. Meanwhile, theimproved method based on Webster theory for computing the delay time is analyzedand the calculated value obtained by this method is bigger than the actual value.Once more, based on the improved intersection turning movement proportionsmodel, we study the phase flow in order to achieve a better performance. Afteranalyzing the two main approaches, that is, direct observation and assumptionestimation, which used to obtain the intersection phase flow, we conclude that theestimation results are greatly different from the actual values. Since the turningmovement proportions play an important role in calculating phase flow, the constraint nonlinear programming model used for estimating intersection turning movementproportions is analyzed. Moreover, as the result of the shortcomings of the abovementioned model, such as the poor accuracy, a high-accuracy model is established. Andthen, the improved model is applied to estimate the phase flow, which provides atheoretical basis for delay time and parking ratio calculating methods.Furthermore, an improved model to calculate the delay time and parking ratio isproposed Considering the problem that the results obtained by employing the delay timeand parking ratio calculating methods are larger than the actual values, the key trafficflow is introduced based on the accurate intersection phase flow and an improved modelis developed.Finally, the NSGA-II algorithm used for solving multi-objective signal timingmodel is improved and the efficiency of the signal control timing scheme proposed inthis thesis is testified. The classical genetic algorithm and pruning strategy areintroduced into the NSGA-Ⅱalgorithm due to the low operation efficiency and the poorsolution sets distribution of it. The innovative NSGA-Ⅱalgorithm which possesses highoperation efficiency and wonderful solution sets distribution has been successfullyapplied to the multi-objective signal timing model. Based on the contrasting andanalyzing of different experiments, we conclude that our proposed signal control timingscheme is better in enhancing intersection operating efficiency than other timingschemes.