A Novel Fault Detection and Classification Approach in Semiconductor Manufacturing Using Time Series Alignment Kernel

A Novel Fault Detection and Classification Approach in Semiconductor Manufacturing Using Time Series Alignment Kernel

Bibliographic Details
Main Author: Zhu, Feng
Language:English
Published: University of Cincinnati / OhioLINK 2020
Subjects:
Engineering
Semiconductor
Time Series Alignment Kernel
Sensor Selection
Fault Detection and Classification
Online Access:http://rave.ohiolink.edu/etdc/view?acc_num=ucin1592135306729513
id ndltd-OhioLink-oai-etd.ohiolink.edu-ucin1592135306729513
record_format oai_dc
spelling ndltd-OhioLink-oai-etd.ohiolink.edu-ucin15921353067295132021-08-03T07:15:21Z A Novel Fault Detection and Classification Approach in Semiconductor Manufacturing Using Time Series Alignment Kernel Zhu, Feng Engineering Semiconductor Time Series Alignment Kernel Sensor Selection Fault Detection and Classification Fault Detection and Classification (FDC) is an integral part of Advanced Process Control (APC) in semiconductor manufacturing. The main objective of FDC is to monitor and analyze the variations in the process data to detect anomalies and to identify the potential root causes. With the rapid advancement of sensor and communication technology, a huge amount of data in semiconductor manufacturing is generated. However, designing useful and reliable features from massive process measurements is not an easy task. Moreover, feature extraction must lose some information of the raw data, which might inadvertently eliminate important sensory information and significantly lower the efficiency of model development.This study proposes a novel approach for FDC system in semiconductor manufacturing process. The proposed method can directly process the multivariate raw trace signal data without the feature extraction in order to reduce the efforts in feature designing, by using Time Series Alignment Kernels (TSAKs). The proposed methodology contains two parts: first is the important sensor screening to prioritize the useful sensor channels for FDC model development in semiconductor manufacturing. This method can be used as a pre-processing step prior to modeling, and the selected sensor channels can be leveraged by process engineers for finer modeling. It combines TSAKs with a feature selection algorithm, minimum Redundancy Maximum Relevance (mRMR), to identify the important sensor channels. Furthermore, a TSAK-Kernel Principal Component Analysis (KPCA) algorithm is proposed to visualize the results. Secondly, TSAK-SVM is employed for FDC modeling. Based on TSAKs, the Gram matrix can be generated directly by using the multivariate raw trace signal, which can be plugged into SVM model for modeling. In this study, validation of the proposed method is based on both open-source datasets and the proprietary datasets from a real production line. 2020-06-15 English text University of Cincinnati / OhioLINK http://rave.ohiolink.edu/etdc/view?acc_num=ucin1592135306729513 http://rave.ohiolink.edu/etdc/view?acc_num=ucin1592135306729513 unrestricted This thesis or dissertation is protected by copyright: all rights reserved. It may not be copied or redistributed beyond the terms of applicable copyright laws.
collection NDLTD
language English
sources NDLTD
topic Engineering
Semiconductor
Time Series Alignment Kernel
Sensor Selection
Fault Detection and Classification
spellingShingle Engineering
Semiconductor
Time Series Alignment Kernel
Sensor Selection
Fault Detection and Classification
Zhu, Feng
A Novel Fault Detection and Classification Approach in Semiconductor Manufacturing Using Time Series Alignment Kernel
author Zhu, Feng
author_facet Zhu, Feng
author_sort Zhu, Feng
title A Novel Fault Detection and Classification Approach in Semiconductor Manufacturing Using Time Series Alignment Kernel
title_short A Novel Fault Detection and Classification Approach in Semiconductor Manufacturing Using Time Series Alignment Kernel
title_full A Novel Fault Detection and Classification Approach in Semiconductor Manufacturing Using Time Series Alignment Kernel
title_fullStr A Novel Fault Detection and Classification Approach in Semiconductor Manufacturing Using Time Series Alignment Kernel
title_full_unstemmed A Novel Fault Detection and Classification Approach in Semiconductor Manufacturing Using Time Series Alignment Kernel
title_sort novel fault detection and classification approach in semiconductor manufacturing using time series alignment kernel
publisher University of Cincinnati / OhioLINK
publishDate 2020
url http://rave.ohiolink.edu/etdc/view?acc_num=ucin1592135306729513
work_keys_str_mv AT zhufeng anovelfaultdetectionandclassificationapproachinsemiconductormanufacturingusingtimeseriesalignmentkernel
AT zhufeng novelfaultdetectionandclassificationapproachinsemiconductormanufacturingusingtimeseriesalignmentkernel
_version_ 1719457659615182848

Similar Items