Abstract: The invention discloses a process quality prediction system and a method thereof. When a processing apparatus performs a process on a target, the process is measured by a measurement apparatus to receive a process value. The process value and several previous quality data collected from the measurement apparatus are used to predict the quality of the product which is processing inline. The method is composed of a moving window, a stepwise regression scheme and an analysis of covariance (ANCOVA). The drift and shift of process are overcome by the moving window. A key variable set is selected by the stepwise regression scheme and a virtual model is identified by the analysis of covariance.

Abstract: The invention discloses a method for building adaptive soft sensor. The method comprises the following steps. The input and schedule vectors are constructed, and a novel learning algorithm that uses online subtractive clustering is used to recursively update the structure and parameters of a local model network. Three rules are proposed for updating centers and local model coefficients of existing clusters, for generating new clusters and new models as well as for merging existing clusters and their corresponding models. Once verified, the online inferential model can be created to generate the predicted value of process. Thus, it does not need much memory space to process the method and can be easily applied to any other machine.

Abstract: The present invention discloses a virtual measuring system and a method thereof for predicting the quality of thin film transistor liquid crystal display processes. The virtual measuring method comprises the steps of: capturing a plurality of process parameter data from at least one process machine by an advanced process control unit; normalizing the process parameter data by an original data processing unit; picking a plurality of key process parameter data from the process parameter data by a key parameter choosing unit; establishing a virtual measuring model by a predicting unit according to the key process parameter data, and generating a virtual measuring data by the virtual measuring model. The virtual measuring model is established after a disturbing coefficient is generated through a time sequence regression algorithm by the predicting unit.

Abstract: The invention discloses a method for building adaptive soft sensor. The method comprises the following steps. The input and schedule vectors are constructed, and a novel learning algorithm that uses online subtractive clustering is used to recursively update the structure and parameters of a local model network. Three rules are proposed for updating centers and local model coefficients of existing clusters, for generating new clusters and new models as well as for merging existing clusters and their corresponding models. Once verified, the online inferential model can be created to generate the predicted value of process. Thus, it does not need much memory space to process the method and can be easily applied to any other machine.

Abstract: The invention discloses a process quality prediction system and a method thereof. When a processing apparatus performs a process on a target, the process is measured by a measurement apparatus to receive a process value. The process value and several previous quality data collected from the measurement apparatus are used to predict the quality of the product which is processing inline. The method is composed of a moving window, a stepwise regression scheme and an analysis of covariance (ANCOVA). The drift and shift of process are overcome by the moving window. A key variable set is selected by the stepwise regression scheme and a virtual model is identified by the analysis of covariance.