Browsing by Author "Chueh, Chuang-Hua"

Now showing 1 - 3 of 3

Forecasting Rare Faults of Critical Components in LED Epitaxy Plants Using a Hybrid Grey Forecasting and Harmony Search Approach
(2016-12)
Lin, Chun-Cheng
;
Deng, Der-Jiunn
;
Kang, Jia-Rong
;
Chang, Sen-Chia
;
Chueh, Chuang-Hua
In the light emitting diode (LED) manufacturing industry, the most expensive and crucial facilities are manufacturing machines. Condition-based maintenance (CBM) for crucial components of a manufacturing machine aims to forecast in advance the precise time when some aging component will be broken and replace it in time, to avoid performing abnormally to manufacture defect products. This study focuses on the CBM for a crucial component called particle filter of a pneumatic conveyor machine in the LED epitaxy plant. Conventional forecasting methods were based on the theory of statistics, which requests a large number of data samples and assumes some probability distribution. With advance of machine technology, however, the data samples of broken particle filters to be collected are very few, such that those conventional methods cannot be applied. As a result, this study proposes a novel hybrid grey forecasting and harmony search approach, in which grey forecasting was shown to perform well for small data samples. In the proposed method, operating conditions of particle filters are monitored and collected by industrial sensors. Then, those data are preprocessed by data filtering and clustering. Finally, a hybrid grey forecasting and harmony search approach is used to fit the curve of the aging condition of a particle filter. Numerical analysis of a real example in an LED epitaxy plant shows that the proposed method performs better than conventional methods.
Forecasting Rare Faults of Critical Components in LED Epitaxy Plants Using a Hybrid Grey Forecasting and Harmony Search Approach
(2016-12)
Lin, Chun-Cheng
;
Deng, Der-Jiunn
;
Kang, Jia-Rong
;
Chang, Sen-Chia
;
Chueh, Chuang-Hua
In the light emitting diode (LED) manufacturing industry, the most expensive and crucial facilities are manufacturing machines. Condition-based maintenance (CBM) for crucial components of a manufacturing machine aims to forecast in advance the precise time when some aging component will be broken and replace it in time, to avoid performing abnormally to manufacture defect products. This study focuses on the CBM for a crucial component called particle filter of a pneumatic conveyor machine in the LED epitaxy plant. Conventional forecasting methods were based on the theory of statistics, which requests a large number of data samples and assumes some probability distribution. With advance of machine technology, however, the data samples of broken particle filters to be collected are very few, such that those conventional methods cannot be applied. As a result, this study proposes a novel hybrid grey forecasting and harmony search approach, in which grey forecasting was shown to perform well for small data samples. In the proposed method, operating conditions of particle filters are monitored and collected by industrial sensors. Then, those data are preprocessed by data filtering and clustering. Finally, a hybrid grey forecasting and harmony search approach is used to fit the curve of the aging condition of a particle filter. Numerical analysis of a real example in an LED epitaxy plant shows that the proposed method performs better than conventional methods.
LATENT DIRICHLET LEARNING FOR HIERARCHICAL SEGMENTATION
(2012)
Chien, Jen-Tzung
;
Chueh, Chuang-Hua
;
電機資訊學士班
;
Undergraduate Honors Program of Electrical Engineering and Computer Science
Topic model can be established by using Dirichlet distributions as the prior model to characterize latent topics in natural language. However, topics in real-world stream data are non-stationary. Training a reliable topic model is a challenging study. Further, the usage of words in different paragraphs within a document is varied due to different composition styles. This study presents a hierarchical segmentation model by compensating the heterogeneous topics in stream level and the heterogeneous words in document level. The topic similarity between sentences is calculated to form a beta prior for stream-level segmentation. This segmentation prior is adopted to group topic-coherent sentences into a document. For each pseudo-document, we incorporate a Markov chain to detect stylistic segments within a document. The words in a segment are generated by identical composition style. This new model is inferred by a variational Bayesian EM procedure. Experimental results show benefits by using the proposed model in terms of perplexity and F measure.