DLSS 4.5

Runyuan Guo, Qingyuan Chen, Hanqing Liu, Wenqing Wang

Despite their high prediction accuracy, deep learning-based soft sensor (DLSS) models face challenges related to adversarial robustness against malicious adversarial attacks, which hinder their widespread deployment and safe application. Although adversarial training is the primary method for enhancing adversarial robustness, existing adversarial-training-based defense methods often struggle with accurately estimating transfer gradients and avoiding adversarial robust overfitting. To address these issues, we propose a novel adversarial training approach, namely domain-adaptive adversarial training (DAAT). DAAT comprises two stages: historical gradient-based adversarial attack (HGAA) and domain-adaptive training. In the first stage, HGAA incorporates historical gradient information into the iterative process of generating adversarial samples. It considers gradient similarity between iterative steps to stabilize the updating direction, resulting in improved transfer gradient estimation and stronger adversarial samples. In the second stage, a soft sensor domain-adaptive training model is developed to learn common features from adversarial and original samples through domain-adaptive training, thereby avoiding excessive leaning toward either side and enhancing the adversarial robustness of DLSS without robust overfitting. To demonstrate the effectiveness of DAAT, a DLSS model for crystal quality variables in silicon single-crystal growth manufacturing processes is used as a case study. Through DAAT, the DLSS achieves a balance between defense against adversarial samples and prediction accuracy on normal samples to some extent, offering an effective approach for enhancing the adversarial robustness of DLSS.

A. Kļaviņa, Patricia Pérez-Fuster, J. Daems, C. N. Lyhne, E. Dervishi, Z. Pajalić, Tone Øderud, K. Fuglerud, S. Markovska-Simoska, Tomasz Przybyła, Michał Klichowski, G. Stiglic, Egija Laganovska, Soraia M. Alarcão, A. Tkaczyk, Carla Sousa

Persons with autism spectrum disorder (ASD) and/or intellectual disability (ID) have difficulties in planning, organising and coping with change, which impedes the learning of daily living skills (DLSs), social participation and self-management across different environmental settings. Assistive technologies (ATs) is a broad term encompassing devices and services designed to support individuals with disabilities, and if used in a self-controlled manner, they may contribute inclusion in all domains of participation. This comprehensive literature review aims to critically assess and unify existing research that investigates the use of assistive technology within the practical domain for individuals with ASD and/or ID. The 18 relevant studies included in this review highlighted the benefits of AT for social participation and independence in daily activities of individuals with ASD and/or ID. Professionals working with this target group should be knowledgeable of the speedy progress of AT products and the potential of persons with ASD and/or ID to use mainstream devices to meet their individual needs. This awareness provides an opportunity to advocate for the universal benefits of AT for everyone. Technologies such as virtual reality, mobile applications and interactive software have been shown to improve DLSs, communication and social interaction. These tools offer engaging, user-friendly platforms that address the specific needs of these individuals, enhancing their learning and independence.

Shuyao Zhang, Changliang Liu, Ziqi Wang, Shuai Liu

Condition monitoring (CM) based on supervisory control and data acquisition (SCADA) data has gained extensive attention for its ability to ensure the normal operation of a vast number of wind turbines (WTs). SCADA data exhibit significant nonlinear and non-Gaussian characteristics, making fault detection K-nearest neighbor (FD-KNN) an industrially valuable approach due to its adaptation to these features. To address the sensitivity of FD-KNN to redundant and noisy samples in SCADA data, a dual-layer sample selection (DLSS) method based on active learning (AL) strategies is proposed to create a high-quality training set. In the first layer, a comprehensive score of samples is calculated based on three AL strategies, and in the second layer, redundancy removal is performed on the candidates to enhance model performance. Subsequently, control limit (CL) of normal condition is set based on FD-KNN, and incoming samples are detected. For detected fault samples, a fault isolation method that does not require faulty samples is proposed, based on contribution decomposition and variable reconstruction, to identify anomalous variables. Finally, the reconstructed values of the anomalous variables are calculated using KNN regression, and the residuals between observations and reconstructed values can serve as a reference for the degree of anomaly. The experiment utilizes anomaly data from an actual WT in China. Results indicate that the sample selection method proposed in this article not only increases computational efficiency by approximately 50% but also slightly improves model accuracy and is applicable to various models. Additionally, the fault isolation method introduced can precisely identify fault variables.

Hantaek Bae, Rafael Granero-Belinch'on

Alexander Mielke, André Schlichting, Artur Stephan

We provide a derivation of the one-dimensional fourth-order DLSS equation based on an interpretation as a chemical reaction network. We consider the rate equation on the discretized circle for a process in which pairs of particles occupying the same site simultaneously jump to the two neighboring sites; the reverse process involves pairs of particles at adjacent sites simultaneously jumping back to the site located between them. Depending on the rates, in the vanishing-mesh-size limit we obtain either the classical DLSS equation or a variant with nonlinear mobility of power type. Via EDP convergence, we identify the limiting gradient structure to be driven by entropy with respect to a generalization of diffusive transport with nonlinear mobility. Interestingly, the DLSS equation with power-type mobility shares qualitative similarities with the fast diffusion and porous medium equation, since we find traveling wave solutions with algebraic tails or compactly supported polynomials, respectively.