class="info-alert">『Your web browser does not support JavaScript, but it does not affect browsing through the rest of the web site.』
jump to main content

Industrial Technology Research Institute

:::

Quality Indicator Prediction Technology

Technology Overview

Quality Indicator Prediction Technology.
Quality Indicator Prediction Technology.

The developed product quality prediction techniques can analyze the sensor data generated by process tool and real-time estimate/predict the product quality precisely. It can help to achieve fully-inspection as we as dynamically-adjustment of labors and process to increase the profit.

Applications & Benefits

Because of the high cost of real-time fully-inspection, sampling approach is often used. However, this approach easily lead to the outflow of defective products. Also, in the multi-stage process, because the final product quality cannot be known immediately, it is necessary to wait until the complete process is completed and quality can be obtained, which may take time and raw materials are wasted. The proposed product quality prediction techniques are based on ensemble prediction method, which combines multiple advanced AI algorithms. We analyzes the sensor data to predict the product quality. We also developed the model adaptation technique that helps to fan out the AI models to different machines. These techniques can fulfill the real-time fully-inspection. The two major applications are

  • Real-time quality estimation: product quality is simultaneously estimated after the process is finished without actual metrology so as to achieve indirectly full-inspection and avoid releasing defect products.
  • Cross-process quality prediction: predict final product quality in advance after the critical process is finished to avoid unnecessary post-process and dynamically modify the configuration of post-process, which can effectively reduce the unnecessary production cost as well as improve the profit.

The developed product quality prediction techniques can analyze the sensor data generated by process tool and quality metrology data to obtain the data correlation to real-time predict the product quality precisely. The two major applications are

  • Real-time quality estimation: product quality is simultaneously estimated after the process is finished without actual metrology to achieve indirectly full-inspection and avoid releasing defect products.
  • Cross-process quality prediction: predict final product quality in advance after the critical process is finished to avoid unnecessary post-process and dynamically modify the configuration of post-process, which can effectively reduce the unnecessary production cost as well as improve the profit.


Related Links