Research On Digital Flight Control System Modelling And Realization Of Auxiliary Training For Flight Simulator

Flight simulator which is the indispensable simulation equipment for the development of new aircraft and flight training in aviation industry can safely and realistically reproduce the behaviour of aircraft flying in the air. At present, the research and technology of flight simulator in our country are far behind the oversea level. Thus, it is significant to research key technology in flight simulator under the background of rapid development of civil aviation and the large civil aircraft develop program. As a key subsystem of civil aeroplane, the digital flight control system plays a significant role in auto-control of flight trajectory and assisting control of flight path. Thus, the high fidelity simulation for the digital flight control system is an essentially requirement in the development of high level simulator. Furthermore, the digital flight control system's function of assisting the control flight trajectory can be extended to assist instructor to demonstrate the diversified flight training, improve the training efficiency, and solve the problems discovered in training using flight simulators.In order to achieve the above aim, the digital flight control system was modeled firstly. Based on the analysis of the function, constitution and operational principle, the whole system was divided into four parts, which are system interface, logic management, roll control and pitch control. Due to the extreme complexity of system, a method that can fast list the expression of system logic by logically setting variables was given. And this method enhances efficiency in the system development. On the basis of the characteristics of digital flight control system and its typical configuration, a control configuration to satisfy the requirement of various flight mode was built using attitude command calculation. Under this configuration, the training requirements of flight simulator is satisfied by providing corresponding link rules and control algorithm according to each flight mode.In order to reposition and slew operations rapidly and exactly in training, enhance training efficiency and effect, provide exact initialized condition for controller in auxiliary training, this paper studied on the solving problem of steady flight state. The steady flight state, as a most perfect initialized simulation condition, is widely used. However, there are several problems in the solving algorithm for the steady flight state presently, that is, it easy to get local optimization and its precision is not enough, which may affect flight training seriously. Thus, a new mixed genetic algorithm was given for resolving steady flight state. Lamarckian learning and Baldwinina learning are combined together organically in the algorithm. The algorithm could distribute the number of the local search in the population reasonably, make the advantage of the learning into full play and make the disadvantage into inhibitory. Actual test shows that, using the proposed algorithm the local optimum is avoided, the number of iteration is reduced and accuracy of solution is improved.Although the digital flight control system possesses the ability of flight control under normal conditions, it is insufficient to meet the auxiliary training requirements for some particular cases. In order to assist or even take the place of instructors in flight training, the flight decision making method in auxiliary training is investigated. Decision knowledge is converted into a set of flight mode transformational rules according to flight information through the analysis of flight decision. Then, decision-making in autonomous of auxiliary training in flight simulator is achieved by the method of rule based reasoning. In order to ensure validity of rule reasoning, flight mode should be identified exactly. Thus, the two hidden-layer wavelet process neural networks is built to identify flight mode. The convergence of network is improved by wavelet excited functions and the ability of the non-linear mapping and knowledge storage is enhanced by introducing the time-invariant hidden neurons. The digital flight control system possesses a certain degree of autonomous flight ability by the flight decision making method based on the rule based reasoning and flight mode identification makes. Thus, the requirement of auxiliary training is satisfied.The software framework of flight simulator was investigated by analyzing the principles of composition, hardware constitution and computer system configuration. Based on module division, class-hierarchy plan and operating mechanism design, a new flight simulator software structure is developed. The structure can greatly increase the software system of flight simulator flexibility, expansibility and maintainability and reusability. In the software architecture,digital flight control system model, pattern recognition algorithm and a full demonstration flight process is simulated. Results show that the digital flight control system model is correct and rational, the flight pattern recognition is accurate and the auxiliary training is valid. Moreover, the applicability and superiority of the object-oriented simulator software architecture is also verified.