Search

The pivotal factor to regulate and enhance within an operating batch is the quality process, a challenging variable to monitor online. Soft sensors offer an immediate alternative for providing real-time insights into process quality, yet persisting issues include the imbalance between process and quality measurements, noisy measurements, and the intricate 3-dimensional dynamic batch data structure. This paper introduces the Robust Semi-Supervised Dual-Attention Latent Dynamic Conditional State-Space Model (RS2DA-LDCSSM) to address these challenges for within-batch quality prediction. Given the frequent absence of quality data due to measurement inconveniences, an imputation network is embedded within the RS2DA-LDCSSM to facilitate the prediction of future quality. To prevent information distortion during data unfolding, the Attentional Sequence-to-Sequence RNN Encoder-Decoder (AS2S-RNNED) is employed to process the 3-dimensional batch data. The proposed method integrates AS2S-RNNED with a probability state-space model to filter out noisy process data and stabilize the probability prediction of quality data, extracting spatial and temporal latent data from past process and quality data while minimizing the loss in multi-step prediction. This work represents a novel approach to dynamic nonlinear batch processes through probabilistic semi-supervised learning. The RS2DA-LDCSSM is adaptable to any batch process for within-batch quality prediction, as evidenced by numerical cases demonstrating its reliability, achieving an R2 value of 0.87, surpassing comparison methods. In an industrial penicillin fermentation batch process, the RS2DA-LDCSSM exhibits remarkable robustness with a high R2 index value of 0.99 in practical quality prediction scenarios, showcasing its efficacy. • Semi-supervised model training utilizes unlabeled process data. • Separate networks are crafted for process and quality feature extraction. • The state-space model with the encoder-decoder is used to extract latent features. • A multi-step-ahead prediction learning scheme yields favorable prediction results. [ABSTRACT FROM AUTHOR]

The pivotal factor to regulate and enhance within an operating batch is the quality process, a challenging variable to monitor online. Soft sensors offer an immediate alternative for providing real-time insights into process quality, yet persisting issues include the imbalance between process and quality measurements, noisy measurements, and the intricate 3-dimensional dynamic batch data structure. This paper introduces the Robust Semi-Supervised Dual-Attention Latent Dynamic Conditional State-Space Model (RS2DA-LDCSSM) to address these challenges for within-batch quality prediction. Given the frequent absence of quality data due to measurement inconveniences, an imputation network is embedded within the RS2DA-LDCSSM to facilitate the prediction of future quality. To prevent information distortion during data unfolding, the Attentional Sequence-to-Sequence RNN Encoder-Decoder (AS2S-RNNED) is employed to process the 3-dimensional batch data. The proposed method integrates AS2S-RNNED with a probability state-space model to filter out noisy process data and stabilize the probability prediction of quality data, extracting spatial and temporal latent data from past process and quality data while minimizing the loss in multi-step prediction. This work represents a novel approach to dynamic nonlinear batch processes through probabilistic semi-supervised learning. The RS2DA-LDCSSM is adaptable to any batch process for within-batch quality prediction, as evidenced by numerical cases demonstrating its reliability, achieving an R2 value of 0.87, surpassing comparison methods. In an industrial penicillin fermentation batch process, the RS2DA-LDCSSM exhibits remarkable robustness with a high R2 index value of 0.99 in practical quality prediction scenarios, showcasing its efficacy. • Semi-supervised model training utilizes unlabeled process data. • Separate networks are crafted for process and quality feature extraction. • The state-space model with the encoder-decoder is used to extract latent features. • A multi-step-ahead prediction learning scheme yields favorable prediction results. [ABSTRACT FROM AUTHOR]

• Dynamic features of process and quality data are learned for quality prediction. • Bi-directional RNN is trained by past and future data to prevent over-fitting. • The unlabeled process data are used to enhance the quality prediction performance. • Improved loss functions for different data types enables parallel parameter update. The online quality variables of soft sensors contribute greatly to obtaining immediate process information. The complex correlations between a large number of process variables inherited from the dynamic and nonlinear characteristics of chemical processes put more challenges on constructing soft-sensor models. Past developed steady-state soft sensors are not reliable for dynamic operating systems. Unequal sampling rates for the process and quality data cause missing values of quality data at some time points. This paper proposes a semi-supervised latent dynamic variational autoencoder to learn features between the process and quality data. A prediction network is constructed to generate artificial quality values for model training. Then the process and quality data are compressed into the latent space and the temporal relation is modeled in the clean latent space. The proposed method is compared with the conventional method for quality prediction in a numerical case and an industrial case. [ABSTRACT FROM AUTHOR]

We demonstrate a novel avalanche photodiode (APD) design which fundamentally relaxes the trade-off between responsivity and saturation-current performance at receiver end in coherent system. Our triple In0.52Al0.48As based multiplication (M-) layers with a stepped electric (E-) field inside has more pronounced avalanche process with significantly less effective critical-field than the dual M-layer. Reduced E-field in active M-layers ensures stronger E-field allocation to the thick absorption-layer with a smaller breakdown voltage (Vbr) resulting in less serious space-charge screening effect, less device heating at high output photocurrent. Compared to the dual M-layer reference sample, the demonstrated APD exhibits lower punch-through (− 9 vs. − 24 V)/breakdown voltages (− 43 vs. − 51 V), higher responsivity (19.6 vs. 13.5 A/W), higher maximum gain (230 vs. 130), and higher 1-dB saturation-current (> 5.6 vs. 2.5 mA) under 0.95 Vbr operation. Extremely high saturation-current (> 14.6 mA), high responsivity (7.3 A/W), and decent O-E bandwidth (1.4 GHz) can be simultaneously achieved using the demonstrated APD with a 200 µm active window diameter. In coherent FMCW LiDAR test bed, this novel APD exhibits a larger signal-to-noise ratio and high-quality 3-D images than the reference dual M-layer and high-performance commercial p-i-n PD modules, while requiring significantly less optical local-oscillator (LO) power (0.5 vs 4 mW). [ABSTRACT FROM AUTHOR]

Recessive variants of the SLC26A4 gene are an important cause of hereditary hearing impairment. Several transgenic mice with different Slc26a4 variants have been generated. However, none have recapitulated the auditory phenotypes in humans. Of the SLC26A4 variants identified thus far, the p.T721M variant is of interest, as it appears to confer a more severe pathogenicity than most of the other missense variants, but milder pathogenicity than non-sense and frameshift variants. Using a genotype-driven approach, we established a knock-in mouse model homozygous for p.T721M. To verify the pathogenicity of p.T721M, we generated mice with compound heterozygous variants by intercrossing Slc26a4+/T721M mice with Slc26a4919-2A>G/919-2A>G mice, which segregated the c.919-2A > G variant with abolished Slc26a4 function. We then performed serial audiological assessments, vestibular evaluations, and inner ear morphological studies. Surprisingly, both Slc26a4T721M/T721M and Slc26a4919-2A>G/T721M showed normal audiovestibular functions and inner ear morphology, indicating that p.T721M is non-pathogenic in mice and a single p.T721M allele is sufficient to maintain normal inner ear physiology. The evidence together with previous reports on mouse models with Slc26a4 p.C565Y and p.H723R variants, support our speculation that the absence of audiovestibular phenotypes in these mouse models could be attributed to different protein structures at the C-terminus of human and mouse pendrin. [ABSTRACT FROM AUTHOR]