71 research outputs found
Deep learning based virtual metrology in semiconductor manufacturing processes
No full textVirtual metrology (VM) in semiconductor manufacturing is the technique of predicting critical dimensions of wafer quality characteristics without direct measurement based on process data of production equipment. VM is important in semiconductor manufacturing since it enables engineers to monitor the quality of wafers in production without physical wafer metrology thereby increasing the throughput of the process. As the process information consists of a large number of process variables in the form of raw sensor signals, learning new useful features in a low dimensional space is a key to build accurate VM prediction models. Earlier efforts in VM modeling were carried out by employing linear dimensionality reduction techniques such as PCA. Autoencoder is a deep learning based feature extraction method that has the capability to explore the non-linearity in the modeling and to represent high dimensional input into a low dimensional space. In this thesis, we propose a new VM model by incorporating the autoencoder based feature learning. We apply the proposed model to the prediction of critical dimensions of wafers at a plasma etching process in semiconductor manufacturing and compare the predictive performance of the proposed model with conventional VM models. The experimental results show that the proposed model outperforms the existing models thus showing that autoencoder based feature learning is helpful in VM modeling with raw sensor signals.M.S.Includes bibliographical referencesby Harshit Bokadi
Automated drug dispenser based on pressure ejection of medications
No full textVarious types of automated drug dispensers exist in the market. However, they usually involve extraction of medications from their packaging and their temporary storage in internal bins. In this paper, we propose a different approach which can bypass this step through pressure ejection of medications (especially capsules) from their packaging strips. Further, it is proposed that a relevant consensus between various pharmaceutical manufacturers for standardization of the size and packaging of medications can allow for increased automation in the dispensation of medications to patients without altering the logistics of the existing manual dispensation of medications
On the Turbulent Viscosity Parameter Cμ in the K–ε Model
No full textThe Reynolds-averaged Navier–Stokes (RANS) models depend on empirical constants to close the Reynolds stress terms. The empirical constants were obtained using experiments conducted at low Reynolds numbers several decades ago. In this paper, we revisit the turbulent viscosity parameter Cμ, based on the stress–intensity ratio c2 = |uw|/k. Here, |uw| and k are the absolute values of the Reynolds stress and turbulent kinetic energy, respectively. Through a priori comparisons, we find that the currently accepted value of Cμ = 0.09 does not agree with the latest direct numerical simulation (DNS) and experimental datasets of wall-bounded turbulent planar flows. Therefore, a new value is suggested by averaging c2 in the equilibrium region, where the production (P) of k is within 10 % of the dissipation rate (ε), and consequently, c4 ≈ Cμ. We evaluate flows up to friction Reynolds number ReΓ ≈ 10 000 and find that with increasing ReΓ, Cμ approaches a value of 0.06, which is almost 50 % lower than the prevalent value of 0.09. Finally, we perform an a priori test with the new (proposed) value of Cμ = 0.06 to show that the estimated turbulent viscosity νT for wall-bounded flows is in much closer agreement with the exact (DNS) values than when νT is estimated using Cμ = 0.09
New Method to Calculate Friction Velocity in Smooth Channel Flows using Direct Numerical Simulation Data
No full textIn this paper, we leverage the direct numerical simulation (DNS) data for closed-channel flow for a range of friction Reynolds number (Reτ∼180-5,000) to develop a new one-point friction velocity method (OPFVM) to calculate friction velocity U∗ in terms of free-surface velocity Um, flow depth h, and kinematic viscosity ν. In contrast to prevalent methods that require several cumbersome near boundary measurements to obtain friction velocity, the OPFVM relies on a single easy-to-measure free-surface velocity measurement. The formulation is used to obtain friction velocity for a closed-channel flow (CCF) DNS regime with Reτ=10,049 and on four open-channel flow (OCF) DNS regimes with Reτ∼180-2,000. The same formulation was then experimentally verified in our laboratory. To avoid being prescriptive, a sensitivity analysis was performed to determine the permissible variation in Um to restrict the error in estimated U∗ to 2%. The relationship between the depth-averaged velocity Ub and the maximum free-stream velocity Um is also explored using the DNS data sets and an approximate relationship between Ub and Um is proposed. With advances in remote-sensing technology that enables free-stream velocity measurements, this method extends the potential to measure even the friction velocity remotely
Effect of perforation on exhaust performance of a turbo pipe type muffler using methanol and gasoline blended fuel: A step to NOx control
No full textUnderstanding facilitators and challenges in the new product development process : the case of a small Australian firm
No full textThis thesis was scanned from the print manuscript for digital preservation and is copyright the author.
Researchers can access this thesis by asking their local university, institution or public library to
make a request on their behalf. Monash staff and postgraduate students can use the link in the References field
Modeling for combined effect of muffler geometry modification and blended fuel use on exhaust performance of a four stroke engine: A computational fluid dynamics approach
No full textA sui generis QA approach using RoBERTa for adverse drug event identification
No full textAbstract Background Extraction of adverse drug events from biomedical literature and other textual data is an important component to monitor drug-safety and this has attracted attention of many researchers in healthcare. Existing works are more pivoted around entity-relation extraction using bidirectional long short term memory networks (Bi-LSTM) which does not attain the best feature representations. Results In this paper, we introduce a question answering framework that exploits the robustness, masking and dynamic attention capabilities of RoBERTa by a technique of domain adaptation and attempt to overcome the aforementioned limitations. With formulation of an end-to-end pipeline, our model outperforms the prior work by 9.53% F1-Score. Conclusion An end-to-end pipeline that leverages state of the art transformer architecture in conjunction with QA approach can bolster the performances of entity-relation extraction tasks in the biomedical domain. In particular, we believe our research would be helpful in identification of potential adverse drug reactions in mono as well as combination therapy related textual data
- …
