Research On Key Technologies Of Intelligent Transportation Based On Cloud Service Platform

Intelligent transportation system(ITS)needs not only reasonable bus network planning,but also accurate and real-time forecast of bus passenger flow,timely adjustment and optimization of bus operation dispatch,providing real-time,accurate and effective bus travel services for citizens,so as to achieve the optimization of dynamic allocation of public transport resources and save public travel time.Effectively alleviating traffic congestion helps achieve the purpose of green travel.Passenger flow prediction at short-term bus stations is an important basis for adjusting bus departure frequency and optimizing bus dispatching resources.Bus dispatching is a complex combinational optimization problem,which needs to optimize passenger flow,departure interval,arrival time,parking time and the number of buses put in.Aiming at these problems,this paper focuses on the operation cost and passenger satisfaction of the bus company as the main indicators.Through training the scheduling optimization model,it realizes the real-time dynamic bus dispatching without supervisory learning,and improves the overall service level of buses.Therefore,it is of great practical significance to study short-term passenger flow forecasting,bus arrival time prediction and bus dynamic dispatching.Aiming at the randomness,time-varying and uncertainty of urban bus passenger flow,this paper proposes a short-term bus station passenger flow forecast based on unsupervised feature learning theory and improved convolutional neural network to provide real-time,accurate and effective bus travel service for citizens.In order to prevent and reduce the occurrence of over-fitting,the improved convolution is applied.The neural network prediction training method constructs an efficient and high-reliability model prediction system.In the training process,the Adam optimizer is used to optimize the model,update the network model parameters,and set different parameters for the adaptive learning rate.The bus passenger flow algorithm model is used to predict the passenger flow of the actual bus station in Guangzhou.The prediction accuracy of the model is more accurate and reliable,which indicates that the proposed method has smaller prediction error.The example proves that the improved model and algorithm are practical and reliable.By providing real-time bus arrival time,it is conducive to travel decision-making,reasonable choice of bus ride and alleviation of bus congestion.It is an important research direction of intelligent bus arrival time prediction in the future.Based on the predicted bus arrival time,the operation dispatch of bus can be adjusted reasonably and the management level of bus operation can be improved.Compared with curve matching prediction model,curve matching prediction model based on clustering analysis has higher accuracy and better effect in predicting bus arrival time,which reduces passenger waiting time and improves the efficiency of bus dispatching and operation.The experimental results show that the model can meet the accuracy and real-time requirements of prediction.Through the evaluation and analysis of bus departure punctuality rate,bus stop time and bus arrival time interval,these three indicators are obviously reduced in the peak period.In the actual operation process of bus,due to the complexity of road traffic conditions and the random,time-varying and uncertain reasons of bus passenger flow,it is impossible for bus vehicles to be fully adjusted in accordance with the traffic plan and bus arrival time interval.Degree planning has led to bus off-duty problems.This paper presents a real-time dynamic bus dispatching method without supervised learning.It is precisely to solve the problem of bus interruption,to ensure that the bus is executed according to the traffic plan and vehicle dispatching plan,and to improve the quality and efficiency of bus dispatching.Therefore,bus departure punctuality rate,bus stop time and bus arrival time interval are the basis of bus dispatching.These three evaluation results are the important guarantee for making bus dispatching scheduling plan,and the important basis for establishing bus dispatching model and optimizing bus dispatching scheme.Aiming at the optimization problem of Guangzhou intelligent bus dispatching,a real-time dynamic bus dispatch algorithm based on unsupervised learning is proposed and applied to this problem.Combining the interests of passengers and bus companies as the goal,we can learn the expression of bus passenger outflow characteristics by unsupervised learning method,and optimize data sets and support vector machines by using attractor propagation clustering algorithm.The training sample set is combined to establish the prediction model training,and the objective function of bus network departure interval and weighting coefficient is used to optimize the scheduling model.The multi-source information fusion and multi-strategy real-time dynamic scheduling algorithm are introduced into the solving model,which realizes the real-time adjustment of the scheduling optimization model.The examples prove that the model and algorithm are practical and reliable.On the basis of intelligent transportation cloud service platform,this paper uses machine learning key technology to realize intelligent bus passenger flow prediction,dynamic bus dispatch and docking with intelligent transportation system.Cloudstack is used to build an intelligent transportation cloud computing infrastructure platform,which has good resource scalability characteristics and provides cloud hosting services and resilient block storage services.Provide cloud hosts with different resource allocation,and provide private interconnection between cloud hosts.Provide flexible block storage of various configuration sizes to meet the needs of bus passenger flow and bus scheduling data storage.Providing cloud hosting services,the upper layer can run a variety of applications;because of providing flexible block storage services,the upper layer can allocate storage services of different capacity according to demand.Through the above research,the intrinsic mechanism of bus short-term passenger flow forecasting,bus arrival time prediction and bus dynamic dispatching is complex.The model established in this paper aims to improve system performance and operating efficiency,improve bus network planning,bus dynamic dispatch design and operation.The scientific nature of the program provides theoretical guidance and data support for solving traffic congestion and difficult travel problems in major engineering applications.