Faraway engineers are able to sketch direct the shape of engineering components by the browser, and the recognition system will proceed with search for the component database of company by the Internet. In this paper, component patterns are stored in the database system. Component patterns with the approach of database system will be able to improve the capacity of recognition system effectively. In our approach, the recognition system adopts distributed compute, and it will raise the recognition rate of system. The system uses a recurrent neural network (RNN) with associative memory to perform the action of training and recognition. The final phase joins the technology of database match in process of the recognition except distributed compute, and it will solve the problem of spurious state. In this paper, our system will be carried out in the Yang-Fen Automation Electrical Engineering Company. The plan of experiment has gone through four months, and their engineers are also used to take advantage of the way of Web-Based pattern recognition.