Design Of Multi-terminal Semiconductor Packaging Equipment Running State Monitoring System Based On Decision Tree

Recently,with the rapidly development and strong industrial demand of semiconductor industry,the semiconductor packaging equipment plays a vital role in the development of semiconductor industry.Ensuring the safe and stable operation of semiconductor packaging equipment on the production line is the primary key.However,the main fault maintenance of such equipment in China is manual maintenance and it cannot meet the needs of industry development.Therefore,this paper designs a multi-terminal semiconductor packaging equipment monitoring system based on decision tree to realize the remote real-time monitoring and fault prediction of semiconductor packaging equipment.The main contents and conclusions of this paper are as follows:1.System related theory and development technology.The decision tree algorithm used in this system is introduced in detail,and a series of related technologies such as JAVA programming language,Modbus communication protocol and MVC design pattern are described.2.Monitoring system design.Select the semiconductor packaging equipment involved in the system,and then elaborate the design of the system user login module and interface,and carry out three-dimensional modeling of the computer room,restore the internal real scene,and display to the front interface,then elaborate the design of the four functional modules of the system including real-time monitoring of equipment,equipment communication,system alarm and system log.Finally,design the database table that used to store system data.3.Equipment fault data collection and construction of decision tree.Taking the solid crystal machine as an example,briefly introduce the possible fault types of the equipment,use the Modbus communication protocol to write code to establish a connection with the equipment and obtain fault data for processing,then use the obtained fault data set to describe the generation steps of the fault decision tree,including fault attribute division and establishment.Finally,the compiler software is used to run the code of the decision tree algorithm to generate the decision tree model,and the algorithm is optimized to improve the model generation efficiency by 40%.4.System test.Detailed tests were carried out on the functional modules of the developed system,and the test results showed that the system could obtain real-time device data and display it to the front page,and the alarm and other functions could be used normally.Finally,the sample data collected were used to test the fault prediction function of the decision tree model,and the results showed that the fault prediction accuracy was more than92%,in line with the system design expected to achieve the goal.