Research On Advanced-surface Movement Guidance And Control System And Simulation Platform

Airport plays an important role in the Airlins operation. With the development of civil aviationindustry, the operations on the airport surface have been increasingly busy and crowded. Therefore,how to improve the capacity of surface operation via new generation technique, without updatingcurrent airport layout and keeping the safty level at the same time, it is an important question in thesustainable development of civil aviation. In this thesis, the airport surface guidance and control is thestudy object, driving decision-making modeling, distributed micro-scopic traffic simulation platformand prototype system of advanced surface movement guidance and control system are also studied.The main content and innovation in this thesis may be summarized as follows:1. Research on fuzzy based driving decision-making modeling. Based on the measured data,firstly identify fuzzy model from sample data set, according to drving habit hypothesis, divid thedecision-making process into some sub-process, training each sub-process as linear model via leastsquare method, then combine them as a complete TSK model. Genetic algorithm is used to optimizethe parameters of TSK model, taking special care on the shortage of local data quantity. Based onbinary logic production rules system, as supplement of TSK model, proposes a hybrid rule system ofdriving decision-making model, which enhanced the robustness and scope of application by addinguncommon driving condition rules.2. Research on support vector machine based driving decision-making modeling. Aiming atbuilding a model without enough prior knowledge, takes use of transduction instead of inductionmethod, to construct the model meets the accuracy requirements. To this end, discussion of the basicproblems of machine learning, introduces the principle of empirical risk minimization and structuralrisk minimization principle, describes the support vector machine (SVM) classification and regressionalgorithms. Taking the measured experimental data as training sample data, with a set of parameterswhich can uniquely identify driving state, builds SVM based drving decision-making model. In orderto improve the accuracy of the model, particle swarm algorithm is applied to the optimization problemof the penalty coefficient and the kernel parameter of the model.3. Research on multi-agent system (MAS) based distributed microscopic traffic simulationplatform. Microscopic traffic simulation software for the lack of support for the status of the airport scene, which is based on MAS and designed as a distributed framework for open simulation platform,is proposed. In order not to let the complexity of the underlying traffic facilities interfere withsimulation system design and realization of other Agent, designs an extensible way for environmentalinformation perception, which solves the problem that decision-making Agent how to obtainenvironment information. The method abstracts a realization independent querying interface, whichmake the system compatible with other MAS systems. Application applying mobile agent techinologyto effectively reduce the traffic between the hosts improves the performance of the system. Adoptingflexible system configuration mode, according to specific airport distribution allocation host load,makes full use of distributed system processing ability.4. Research on hardware design of A-SMGCS prototype system. Based on the premise of A-SMGCS requirements, the main function of prototype system is decomposed, and the main framework of the system is proposed. The on-board terminal hardware platform is designed, using ARM asthe main chip, combined with external chip and I/O interface for connection to GPS and WLANmodules. On this basis, the bootloader, Linux kernel and drivers, together with BusyBox, SQLite, QtE,and other commonly used third-party library are ported to the board, which laid a solid foundation forsubsequent application layer software development.5. Research on softwareware design of A-SMGCS prototype system. Firstly, the key points ofelectrical map are measured with JSCORS, and then the raw data is projected into right-angledcoordinates, which are used in the satellite raster map registration, finally, a specific digital mapformat is proposed, which is dedicated for on-board navigation. Considering the sisufficient of simplealarm algorithm, and the fact that vehicle runs along the specified path, taking the distance along thepath as the alarm judgment basis, a promising alarm algorithm is proposed. Applying C/S mode as thecommunication between monitoring center and on-board terminal, using UDP multicast and multi-thread technology to realize the efficient UDP server module. Using cross-platform-ability of Pythonlanguage and PyQt library, realized the cross-platform monitoring center, and completed the mainfunctions, such as network transceiver, path planning, regional designation, dynamic informationdisplay.6. Research on simulation validation of A-SMGCS prototype system. Based on college digitalmap, and handly added traffic instruments, such as stop line, exclusion, road node, etc., a virtualaerodrome environment used for simulation is founded, which provide a controllable scene forsubsequent system validation. Several typical A-SMGCS operational scenes are designed to test themain functions of the prototype system; simulation results show that the main function of theprototype system meets the design requirements for the operation of the aerodrome to provide a safe and efficient support.