Research On Technology Of Heterogeneous Industrial Production Field State Data Acquisition For LED Manufacturing And Design Of Application System

With the adjustment of social economic structure and the development of modern manufacturing industry,the demands and features of manufacturing industry have changed a lot: from focusing on production towards multi-species small batches,more complex manufacturing process,faster response to the market,higher standards of quality monitoring and backtracking mechanisms.Acquisition of key manufacturing parameters of the product at the manufacturing stage on a real time,as well as tracking of product's manufacturing process is an important method of ensuring product quality.However,due to the lack of motivation to unify communication interface for manufacturing equipment suppliers and parts of the processing machine are not open,it is difficult to acquire and store those different kinds of different structural industrial heterogeneous dat,but it is of important practical significance to realize the quality monitoring of manufacturing and the product backtracking mechanism.This paper focuses on the design of heterogeneous state data acquisition system of industrial production field,takes the LED manufacturing process as the research object.A distributed and universal heterogeneous data acquisition system for industrial production was designed.The main research work of this paper is divided into the following parts:(1)The key nodes of LED production process and main quality index was studied.The distributed structure and the types of data acquisition of the heterogeneous state data acquisition system was determined.(2)On the basis of bar code and QR code technology,the material tracking technology was studied.A simple,efficient and highly automated material information collection technology scheme was put forward to complete the binding of products with raw materials and processing parameters.(3)The processing parameter collection scheme based on display signal analysis was studied and the problem of real-time acquisition of heterogeneous state data of processing machine without data interface and the communication difficulities of different manufacturers and equipment were solved.The paper paid attention to the key problems of processing machine control software in the information analysis,such as the detection of the current active window,the detection of the change of the region of interest and the analysis of the table line of the list data.(4)The key problems of OCR were studied.The character classifiers based on improved BP neural network and convolution neural network were designed and implemented to meet the requirements of heterogeneous data acquisition system for large character sets and high accuracy.After the character set training experiment,the optimal neural network structure and parameters was chosed and the accurate recognition rate was achieved.(5)The efficient and reliable data storage scheme was studied.Three kinds of communication mechanisms,such as heartbeat detection,upload response and retransmission recovery were designed for the communication protocol between single node data acquisition system and data server,to ensure the high availability of data transmission function.