Space telescope is widely used nowadays. Reflective mirror is an important part of the space telescope. Many engineers are trying their best to make the reflective mirror lighter and lighter. Carbon fiber composite material is universally used in the field of aerospace. But there is little engineering experience about making a reflective mirror from carbon fiber. This paper will make a structure optimization about reflective mirror which is made of carbon fiber. Traditional way of structure optimization will call numerical simulation software for many times before finding the best solution. This paper uses a black box model to take the place of simulation software to do the optimization. The black box in this paper is response surface. To some extent, it is a simple type of machine learning. The response surface learns some key information and characteristics from the model which is made by simulation software. Then, the black box, response surface, can work like simulation software. The paper makes a structure optimization based on a certain kind of carbon fiber reflective mirror by the response surface. The diameter of reflective mirror which needs to be optimized in the paper is 150mm. The optimization variables are the thickness of carbon fiber plate and the position of supports. The purpose of the optimization is that the reflective mirror has small value of PV, proper fundamental frequency, small value of stress. The paper finds the best structure design at last by the method stated above. This kind of method can be used to make optimization about other similar optical structure.
|