The Research Of Public Bus Timetable Plan Based On Passenger Flow's Data Mining

The bus timetable, which plays a fundamental role in bus agency management, is the concrete working plan of bus agency to operate the lines. Bus passenger flow and related IC card data are necessary basics for analyzing city public transport planning, operation and dispatchment and bus organization and can help the deparment of bus planning to produce better timetable so as to reduce the cost and improve the service quality.Firstly three models are established which respectively calculate on-board passenger number by IC card, passenger number alight using IC card only when boarding and improve the Fisher sequential clustering model dividing passenger peak hour interval by selecting, screening, calculating and data mining analysis of bus ID card data. Specifically the on-board passenger number using IC card is calculated by matching bus arrival time and using IC card time clustering group which is produced by clustering bus ID card using time. Then the alighting passenger numbers at each stop and interval are given by combining the on-board passenger number of upward or downward direction, bus arrival time and alighting passenger proability matrix, which is produced by analyzing stop attraction power and stop number alighting probabilities. After that, the time interval division during the passenger flow peak hour is produced by calculating line passenger class diameter and error function according to the actual operations.Secondly, the time interval division during the passenger flow peak hour is scientifically given by using the actual operations of line 2 in Beijing Bus Company as an example in matlab.Thirdly, the model of public bus timetable plan based on multi-objective evolutionary is established with the weighted minimum value of bus agency operation cost and passenger waiting cost as the objective function. Based on the features of different multi-objective evolutionary algorithms, three multi-objective evolutionary algorithms which are GA, PSO and GAPSO are chooseen to slove the models calcuting the bus depature time interval. Then the object function value of bus timetable which refers to the first and last bus depature time and interval is used to compare with the current one.In the end, a case study is given. Result 1:The fitness function curve almost stops changing and the optimum solution is gained at the 57th,27th and 38th generation by using GA, PSO and GAPSO algorithms respectively. Compared to GA and PSO, the precision and efficiency have been improved greatly by using GAPSP algorithm proving the superiority of GAPSO. Result 2:The cost has been saved by 4.41%,4.9% and 6.03% respectively by using GA, PSO and GAPSO proving the feasibility and applicability of the model and optimum algorithm.