| The cable-stayed bridge is of a combination structure system which is composing of cable-stayed, tower, and beam etc. three kinds of essential component. It is a multi-order statically indeterminate structure. The line shape and state of internal force of the completed bridge are closely correlation with the construction method adopted and construction order.In cable stayed bridge construction process, for the influence of various factors, the errors between actual bridge state and design bridge state must come into being. This error gradually accumulating will lead that the great error produce between the actual stress state and line shape of completed bridge and stress state and line shape of ideal design, extremely this error will endanger structure safety. Therefore, the construction monitor and control of large span cable-stayed bridge is subjected to extensive concern of the industry practitioner, more and more. This paper taking Tianxinzhou bridge project as engineering background, systemically studies on spatial structure analysis and construction control technique of large span cable-stayed bridge with plate-truss combined steel beam .The mainly works and research results as follows.(1) Base on basic principle of continuum mechanics, the geometrical nonlinear analysis theories of bar system structure are discussed in detail. The explicit FEM formula of Total Lagrangian method and Updated Lagrangian method is given. The differences between U.L. method and T.L. method are also discussed. The research results indicate that the differences of calculation results are very tiny when applied these two methods to solve some actual problem. Thereby, this can be comprehended as why two sets physical parameter are mutually used in actual calculating process in a lot of literature.(2) Taken Tianxinzhou bridge project as engineering background, the parameterized modeling method, step and implement process of structure calculating for cable-stayed bridge are discussed. The geometric nonlinear effects of steel cable-stayed bridge are analyzed by means of the model established previously. Research results reveal that the influence degree of nonlinear factor on different component. The longitudinal displacement of main tower and the internal force of main truss beam are greatly subjected to influence. Among several nonlinear factors, the effect of cable-stayed droop degree is mast prominence, the second is large displacement effect, and the pressure-bend effect of beam and column is the least.(3) Based analyzing and comparing on existing construction control methods, the thoughts of that use BP neural network method to implement construction control is put forward. The control standard, accuracy target and implement step of cable-stayed bridge construction control are discussed in detail. The construction control system is established and the measurement and monitor scheme is submitted.(4) The reasons of which result in various errors in large span steel cable-stayed bridge construction process and the influence degree on the bridge structure are researched. The results indicate that the effects of temperature can not be neglected for large span steel cable-stayed bridge with the long construction periods, the large seasonal temperature difference and good heat conduction characteristic of steel material. The influence of error of steel beam weight to the total displacement of structure and internal force are also very obvious. Cable tension error is a kind of small scope error.(5) On the foundation of detailed discussing neural network principle and application state, these several works as follows is paid attention to. The convergence property, nonlinear fit function and forecast function of BP network are discussed, so that established theoretic foundation for the applying BP neural network to implement parameter identifying and predicting in large span cable-stayed bridge construction control process. The training and error check calculation formula of BP neural network are deduced. And the detailed step how to carry out is induced. The improving measures and disposal methods are put forward for the existing shortages and blemish of BP arithmetic. A program of BP arithmetic with adaptive function is set up. And the varying property of training rate of network depending on training factorα,βand inertiaηparameter is simulated by use of the program ABPNN. The research results indicate that the influence of training factorα,βon convergence rate is mild, and the influence degree exists difference for different problem. Then the scope of training factor is wide. As a rule, a=p, and the range is 0.2-5.0. The choosing of inertial factorηis more complicated, it very easily cause to surge, in generally, it is more suitable that the value ofηis between 0.2-0.5.(6)The design parameters identifying are implemented by means of BP neural network identifying function in Tianxinzhou bridge construction process. Making use of BP neural network forecast function, carried out assembled elevation and initial cable tension predicting during cantilever construction. The results show that the predicting value of BP network is good tally with measurement value, it completely satisfy the request of engineering and control.(7) Making use of nonlinear fit function and converse control model of BP neural network for the first time, the cable force adjustment calculating are implemented for error adjusting of control parameter and after bridge completed in Tianxinzhou bridge construction control process. The application results show that the BP neural network method is effectual and rational, it own high theoretic and use value and its application is a new complement to traditional method.(8) In the end, based on summed up research experience and realizes for several years , consult ed domestic and international development trend and combine the research results of related realm, a intelligentized construction control system of large span cable-stayed bridge is put forward. And this will direct direction of further research in future.
|
PDF Full Text Request