Wafer Edge Yield Prediction Using a Combined Long Short-Term Memory and Feed- Forward Neural Network Model for Semiconductor Manufacturing

Kim, Dasol; Kim, Mintae; Kim, Wooju

doi:10.1109/access.2020.3040426

Cited by 26 publications

(11 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Among the common neural network, it can be divided into feedforward neural networks (FFNN) [26] and recurrent neural networks (RNN) [27] [28]. Feedforward neural networks, also called fully connected neural networks (FNN), use a single multilayer structure with each layer containing multiple neurons.…”

Section: B Long Short Term Memory Neural Networkmentioning

confidence: 99%

Ship Attitude Prediction Model Based on Cross-Parallel Algorithm Optimized Neural Network

et al. 2022

View full text Add to dashboard Cite

In recent years, with the development of modern technology and industrialization, the shipbuilding industry has become an essential field in a country. Offshore operations, such as helicopter takeoff and landing, and ship-to-ship cargo replenishment, need an accurate prediction of ship motion attitude, not only to improve efficiency but also to protect crew members' lives. However, using the single model has limited influence on prediction. Therefore, this paper proposed a Cross-Parallel optimization algorithm framework, which made full use of the development ability of the Harris Hawk optimization algorithm (HHO) and the searching ability of the sine-cosine algorithm (SCA). In this paper, we use Rosenbrock and Rastrigin functions for testing. The results show that the cross-parallel optimization algorithm has faster search speed and stronger optimization capability. The optimization algorithm is used to optimize the hyperparameters of long short-term memory (LSTM) neural networks. It improves the network's training ability to predict the ship's attitude. The proposed method in this paper is compared with 9 models such as BP, LSTM, GRU, BILSTM and TCN. It is tested with three degrees of freedom under different sea states. MAE and RMSE are used as the evaluation indexes of this paper. In the six sets of experimental results, compared with the other nine models, the method in this paper can reduce the MAE by up to 9.2%, 7.2%, 4.7%, 5.4%, 6.4%, and 6.8%, respectively. When using RMSE as the evaluation index, the decrease was at most 10.1%, 8.7%, 5.6%, 7.1%, 7.5%, and 8.2%. The experimental results show that the method in this paper has the highest prediction accuracy.

show abstract

Section: B Long Short Term Memory Neural Networkmentioning

confidence: 99%

Ship Attitude Prediction Model Based on Cross-Parallel Algorithm Optimized Neural Network

et al. 2022

View full text Add to dashboard Cite

show abstract

“…The integration is executed over the set X f , which contains all designs satisfying the performance specifications (e.g., conditions ( 1 ) through ( 3 ) for the considered coupling structure). We have 39 …”

Section: Yield Optimization Of Microwave Passives Using Multi-fidelit...mentioning

confidence: 99%

Rapid yield optimization of miniaturized microwave passives by response features and variable-fidelity EM simulations

Pietrenko‐Dabrowska

Kozieł

2022

Sci Rep

View full text Add to dashboard Cite

The operation of high-frequency devices, including microwave passive components, can be impaired by fabrication tolerances but also incomplete knowledge concerning operating conditions (temperature, input power levels) and material parameters (e.g., substrate permittivity). Although the accuracy of manufacturing processes is always limited, the effects of parameter deviations can be accounted for in advance at the design phase through optimization of suitably selected statistical performance figures. Perhaps the most popular one is the yield, which provides a straightforward assessment of the likelihood of fulfilling performance conditions imposed upon the system given the assumed deviations of designable parameters. The latter are typically quantified by means of probability distributions pertinent to the fabrication process. The fundamental obstacle of the yield-driven design is its high computational cost. The primary mitigation approach nowadays is the employment of surrogate modeling methods. Yet, a construction of reliable metamodels becomes problematic for systems featuring a large number of degrees of freedom. Our work proposes a technique for fast yield optimization of microwave passives, which relies on response feature technology as well as variable-fidelity simulation models. Utilization of response features enables efficient handling of issues related to the system response nonlinearities. Meanwhile, the incorporation of variable-resolution simulations allows for accelerating the yield estimation process, which translates into remarkably low overall cost of the optimizing the yield. Our approach is verified with the use of three microstrip couplers. Comprehensive benchmarking demonstrates its superiority in terms of computational efficiency over the state-of-the-art algorithms, whereas reliability is corroborated by electromagnetic-driven Monte Carlo simulations.

show abstract

“…For the purpose of design centering, we will denote by dx the vector of parameter deviations from nominal values, resulting from fabrication tolerances characterized by probability distributions, specific to a given manufacturing technology, e.g., joint Gaussian G (0,σ), or uniform of maximum deviation d max . As mentioned before, one of the most popular statistical performance metrics is yield Y [37], described as…”

Section: Formulation Of Design Centering Problemmentioning

confidence: 99%

Design Centering of Compact Microwave Components Using Response Features and Trust Regions

Pietrenko‐Dabrowska

Koziel

2021

Energies

View full text Add to dashboard Cite

Fabrication tolerances, as well as uncertainties of other kinds, e.g., concerning material parameters or operating conditions, are detrimental to the performance of microwave circuits. Mitigating their impact requires accounting for possible parameter deviations already at the design stage. This involves optimization of appropriately defined statistical figures of merit such as yield. Although important, robust (or tolerance-aware) design is an intricate endeavor because manufacturing inaccuracies are normally described using probability distributions, and their quantification has to be based on statistical analysis. The major bottleneck here is high computational cost: for reliability reasons, miniaturized microwave components are evaluated using full-wave electromagnetic (EM) models, whereas conventionally utilized analysis methods (e.g., Monte Carlo simulation) are associated with massive circuit evaluations. A practical approach that allows for circumventing the aforementioned obstacles offers surrogate modeling techniques, which have been a dominant trend over the recent years. Notwithstanding, a construction of accurate metamodels may require considerable computational investments, especially for higher-dimensional cases. This paper brings in a novel design-centering approach, which assembles forward surrogates founded at the level of response features and trust-region framework for direct optimization of the system yield. Formulating the problem with the use of characteristic points of the system response alleviates the issues related to response nonlinearities. At the same time, as the surrogate is a linear regression model, a rapid yield estimation is possible through numerical integration of the input probability distributions. As a result, expenditures related to design centering equal merely few dozens of EM analyses. The introduced technique is demonstrated using three microstrip couplers. It is compared to recently reported techniques, and its reliability is corroborated using EM-based Monte Carlo analysis.

show abstract

Wafer Edge Yield Prediction Using a Combined Long Short-Term Memory and Feed- Forward Neural Network Model for Semiconductor Manufacturing

Cited by 26 publications

References 27 publications

Ship Attitude Prediction Model Based on Cross-Parallel Algorithm Optimized Neural Network

Ship Attitude Prediction Model Based on Cross-Parallel Algorithm Optimized Neural Network

Rapid yield optimization of miniaturized microwave passives by response features and variable-fidelity EM simulations

Design Centering of Compact Microwave Components Using Response Features and Trust Regions

Contact Info

Product

Resources

About