Research And Development Of Simulation Optimization Acceleration And Solution Module

Simulation optimization has become an important method for complex system optimal design. However, it holds some flaws in the practical application. On the one hand, every time the new generated design parameters need to be evaluated by simulation system, so it costs quite a lot of time in the simulation course. If the response curved surface of design parameters is very complicated and large-scale, the cost of time for simulation optimization will increase geometrically; On the other hand, in manufacturing cell control, there is a kind of problem called reverse solution, which means that every design parameter of simulation models is solved according to the user's design objective, but traditional simulation optimization can't realize this function. In order to solve the above-mentioned problems, based on the predecessor's achievements, the simulation optimization acceleration and solution module oriented to manufacturing system is researched and developed in this paper. The module increases the computing speed of simulation optimization and realizes the function of reverse solution for manufacturing system simulation models. Specific research contents are as follows:Firstly, the theoretical frame and architecture of simulation optimization acceleration and solution module are put forward. The core of this acceleration and solution module is to get the target output of design parameters by a completely trained neural network substituting for actual simulation models. As a result of the transient response of neural network, the cost of time for simulation optimization reduces greatly, and the function of acceleration for simulation optimization is realized; at the same time, a reverse solution mechanism combining Case-Based Reasoning with Rule-Based Reasoning is designed for realizing the function of reverse solution in manufacturing system simulation models. Secondly, several key technologies are researched, including the design of neural network engine and reverse solution mechanism, the pretreatment of sample data, and the design of sample characteristics in manufacture cell. Finally, simulation optimization acceleration and solution module based on VC++6.0 and QUEST 5.0 manufacturing system simulation platform of Delmia Company is developed, and it is validated by specific examples.