The Research And Practice Of Project Modeling And Calculation On The Basis Of The Artificial Neural Network

At present, Project Modeling is mostly to build up dummy models according to the requirements of practical tasks in computers, then to analyse the models with the corresponding softwares, and to get the necessary results finally. On the one hand, the modeling course is complicated and it takes much more analytical time; On the other hand, when the actual parameters change, the Project Modeling is repeated according to the actual conditions, and the analysis repeated too. Not only the work is tediously long and repeated, but also it delays much more working time repeatedly and reduces the working efficiency.Based on the summary to a lot of domestic literatures and foreign literatures, this paper reviews and summarizes the forefathers' research of the application of Artificial Neural Network(ANN) in machinery design, and raises the frame thought of establishing the project models and caculating based on the ANN. In addition, this paper also relatively lucubrates the problems of ANN upon the modern machinery design, such as the application of ANN and the software realization, and draws a series of valuable conclusions finally. The paper takes the Standard Involute Spur Gear for example, and according to the fact that deformation of gear tooth can't be obtained by the formula, this paper stimulates the deformation of gear tooth in use of the ANN with the characters of speediness and non-linearity, and gets better results. Besides that, the paper also utilizes Pro/ENGINEER to build up its parameter model at first, then educes the model into ANSYS to analyse according to the assumed force state, and gains the database of the model finally. This database provides the data for the study and training of BP network. After that, this paper optimizes the BP network structure in MATLAB according to the training situation, and gets the optimum network structure of 4-10-7-4. And it also utilizes the LabVIEW software to develop the Simulation and Emulation System of BP neural network. By running the system, satisfying results are gotten and the frame thought is proved right and feasible. At last, this paper raises some questions need to be resolved and gives some opinions.