Daily throughput estimation system for semi-conductor manufacturing system

Abstract

The invention relates to a daily output prediction system of a semiconductor manufacturing system in the semiconductor manufacturing field; wherein, a GUI module is responsible for interaction with a user and other modules; a phase space reconstruction module of a daily output time sequence obtains the daily output time sequence and outputs the daily output time sequence after the reconstruction treatment to the GUI module and a neutral network module after the pre-treatment of the daily output time sequence; the neutral network module carries out the prediction treatment on the daily output time sequence and outputs the prediction treatment result to a parameter training processing module of the neutral network and the GUI module; the parameter training treatment module of the neutral network carries out the training treatment on weight parameters of a hidden layer and an output layer in the neutral network module. The daily output prediction system improves the accuracy and the precision of the daily output prediction of a semiconductor production line.

Description

technical field [0001] The invention relates to an information processing system in the technical field of semiconductor manufacturing, in particular to a daily output prediction system of a semiconductor manufacturing system. Background technique [0002] Since the semiconductor production line belongs to the mixed processing mode, its processing process is complex, there are many types and quantities of products in progress, various discrete, dynamic, and uncertain events exist in the production process, and the average processing cycle of products is long, resulting in the production of semiconductor manufacturing systems. Planning and scheduling are difficult and less feasible. By adding production prediction control means to the control system of the semiconductor manufacturing system to assist in solving the above problems, the accuracy and feasibility of the production operation planning and control of the semiconductor manufacturing system can be effectively improved...

AI Technical Summary

Problems solved by technology

The method based on artificial neural network is used for the prediction of output. The input dimension of the prediction model is selected at random, mainly through the method of enumeration test, and the adaptability of the prediction model is poor; The historical data of the day is used as the input of the prediction model, and there is no in-depth analysis of the characteristics and rules of the historical data changes of the output and other performance indicators, resulting in poor prediction accuracy of the model; at the same time, the parameter training process of the BP neural network method is easy to fall into the local optimized and therefore not suitable for the context of large-scale complex semiconductor processing yield predictions

Images