A Distributed Neural Hybrid System Learning Framework in Modeling Complex Dynamical Systems.

In this article, a distributed neural network modeling framework including a novel neural hybrid system model is proposed for enhancing the scalability of neural network models in modeling dynamical systems. First, high-dimensional training data samples will be mapped to a low-dimensional feature space through the principal component analysis (PCA) featuring process. Following that, the feature space is bisected into multiple partitions based on the variation of the Shannon entropy under the maximum entropy (ME) bisecting process. The behavior of subsystems in the prespecified state space partitions will then be approximated using a group of shallow neural networks (SNNs) known as extreme learning machines (ELMs), and then it can further simplify the model by merging the redundant lattices based on their training error performance. The proposed modeling framework can handle high-dimensional dynamical system modeling problems with the advantages of reducing model complexity and improving model performance in training and verification. To demonstrate the effectiveness of the proposed modeling framework, examples of modeling the LASA dataset and an industrial robot are presented.

Mechanistic investigation of direct photodegradation of chloroquine phosphate under simulated sunlight.

Chloroquine phosphate (CQ) is an antiviral drug for Coronavirus Disease 2019 and an old drug for treatment of malaria, which has been detected in natural waters. Despite its prevalence, the environmental fate of CQ remains unclear. In this study, the direct photodegradation of CQ under simulated sunlight was investigated. The effect of various parameters such as pH, initial concentration and environmental matrix were examined. The photodegradation quantum yield of CQ (4.5 × 10-5-0.025) increased with the increasing pH value in the range of 6.0-10.0. The electron spin resonance (ESR) spectrometry and quenching experiments verified that the direct photodegradation of CQ was primarily associated with excited triplet states of CQ (3CQ*). The common ions had negligible effect and humic substances exhibited a negative effect on CQ photodegradation. The photoproducts were identified using high-resolution mass spectrometry and the photodegradation pathway of CQ was proposed. The direct photodegradation of CQ involved the cleavage of the C-Cl bond and substitution of the hydroxyl group, followed by further oxidation to yield carboxylic products. The photodegradation processes were further confirmed by the density functional theory (DFT) computation for the energy barrier of CQ dichlorination. The findings contribute to the assessment of the ecological risk associated with the overuse of Coronavirus drugs during global public health emergencies.

Effect of disinfection on the photoreactivity of effluent organic matter and photodegradation of organic contaminants.

Chlorine, UV254, and ozone are three typical processes commonly used for wastewater disinfection, which could change the photoreactivity of dissolved organic matter (DOM) in effluents of wastewater treatment plants (WWTPs). The photoinduced reactive species (RS) from DOM, primarily including the excited triplet state of DOM (3DOM*), singlet oxygen (1O2), and hydroxyl radical (•OH), play important roles in the attenuation of contaminants. However, the effect of disinfection processes on the photosensitized degradation of contaminants is poorly understood. This paper presents the first evidence that 3DOM*, 1O2, and •OH interaction with three typical contaminants (diphenhydramine, cimetidine, and N,N-diethyl-m-toluamide (DEET)) was largely impacted by DOM after disinfection. The results of electron spin resonance (ESR) spectrometry and laser flash photolysis (LFP) experiments demonstrated that the chlorination increased the formation rate of 3DOM* and 1O2, while UV254 irradiation and ozonation decreased the formation rate of these RS. All these three disinfection processes promoted the photoproduction of •OH and increased the photodegradation rate constants (kobs) of DEET by 26-361%. The kobs of diphenhydramine, cimetidine, and DEET correlated positively with the formation rate of 3DOM*, 1O2, and •OH, respectively. The bimolecular reaction rate constant of 3DOM* with diphenhydramine increased by ∼41% after chlorination. These findings suggest that disinfection processes altered the photogeneration of RS from DOM, which significantly impacts the fate of trace pollutants in aquatic environments.

Guaranteed approximation error estimation of neural networks and model modification.

Approximation error is a key measure in the process of model validation and verification for neural networks. In this paper, the problems of guaranteed error estimation of neural networks and applications to assured system modeling and assured neural network compression are addressed. First, a concept called guaranteed error estimation of feedforward neural networks is proposed, which intends to provide the worst-case approximation error of a trained neural network with respect to a compact input set essentially containing an infinite number of values. Given different prior information about the original system, two approaches including Lipschitz constant analysis and set-valued reachability analysis methods are developed to efficiently compute upper-bounds of approximation errors. Based on the guaranteed approximation error estimation framework, an optimization for obtaining parameter values from data set is proposed. A robotic arm and neural network compression examples are presented to illustrate the effectiveness of our approach.

Perylene diimide supermolecule (PDI) as a novel and highly efficient cocatalyst for photocatalytic degradation of tetracycline in water: A case study of PDI decorated graphitic carbon nitride/bismuth tungstate composite.

Employing perylene diimide supermolecule (PDI) as metal-free cocatalyst, a novel PDI/g-C3N4/Bi2WO6 (PCB) photocatalyst was constructed for the effective degradation of antibiotics. Both the photocatalytic activity and photostability of g-C3N4/Bi2WO6 (gCB) were further improved after loading PDI. Under simulated sunlight illumination, the apparent rate constant of tetracycline (TC) degradation by PCB reached 2.6 times that of gCB. The photocatalytic activity of PCB still kept over 80% after 4 cycle experiments, while gCB only remained around 21%. The superior activity of PCB was ascribed to the synergism between the extended visible light absorption range through the participation of PDI cocatalyst and facilitated gCB-to-PDI photoelectron transfer. TC would finally be transformed into non-toxic ring opening products and mineralized. This work demonstrated that PDI was an excellent metal-free cocatalyst and exhibited great potential to boost the activity of photocatalysts.

Runtime Safety Monitoring of Neural-Network-Enabled Dynamical Systems.

Complex dynamical systems rely on the correct deployment and operation of numerous components, with state-of-the-art methods relying on learning-enabled components in various stages of modeling, sensing, and control at both offline and online levels. This article addresses the runtime safety monitoring problem of dynamical systems embedded with neural-network components. A runtime safety state estimator in the form of an interval observer is developed to construct the lower bound and upper bound of system state trajectories in runtime. The developed runtime safety state estimator consists of two auxiliary neural networks derived from the neural network embedded in dynamical systems, and observer gains to ensure the positivity, namely, the ability of the estimator to bound the system state in runtime, and the convergence of the corresponding error dynamics. The design procedure is formulated in terms of a family of linear programming feasibility problems. The developed method is illustrated by a numerical example and is validated with evaluations on an adaptive cruise control system.

Review on application of perylene diimide (PDI)-based materials in environment: Pollutant detection and degradation.

Environment pollution is getting serious and various poisonous contaminants with chemical durability, biotoxicity and bioaccumulation have been widespreadly discovered in municipal wastewaters and surface water. The detection and removal of pollutants show great significance for the protection of human health and other organisms. Due to its distinctive physical and chemical properties, perylene diimide (PDI) has received widespread attention from different research fields, especially in the area of environment. In this review, a comprehensive summary of the development of PDI-based materials in fluorescence detection and advanced oxidation technology for environment was introduced. Firstly, we chiefly presented the recent progress about the synthesis of PDI and PDI-based nanomaterials. Then, their application in fluorescence detection for environment was presented and categorized, principally including the detection of heavy metal ions, harmful anions and organic contaminants in the environment. In addition, the application of PDI and PDI-based materials in different advanced oxidation technologies for environment, such as photocatalysis, photoelectrocatalysis, Fenton and Fenton-like reaction and persulfate activation, was also summarized. At last, the challenges and future prospects of PDI-based materials in environmental applications were discussed. This review focuses on presenting the practical applications of PDI and PDI-based materials as fluorescent probes or catalysts (especially photocatalysts) in the detection of hazardous substances or catalytic elimination of organic contaminants. The contents are aimed at supplying the researchers with a deeper understanding of PDI and PDI-based materials and encouraging their further development in environmental applications.

Reachable Set Estimation for Neural Network Control Systems: A Simulation-Guided Approach.

The vulnerability of artificial intelligence (AI) and machine learning (ML) against adversarial disturbances and attacks significantly restricts their applicability in safety-critical systems including cyber-physical systems (CPS) equipped with neural network components at various stages of sensing and control. This article addresses the reachable set estimation and safety verification problems for dynamical systems embedded with neural network components serving as feedback controllers. The closed-loop system can be abstracted in the form of a continuous-time sampled-data system under the control of a neural network controller. First, a novel reachable set computation method in adaptation to simulations generated out of neural networks is developed. The reachability analysis of a class of feedforward neural networks called multilayer perceptrons (MLPs) with general activation functions is performed in the framework of interval arithmetic. Then, in combination with reachability methods developed for various dynamical system classes modeled by ordinary differential equations, a recursive algorithm is developed for over-approximating the reachable set of the closed-loop system. The safety verification for neural network control systems can be performed by examining the emptiness of the intersection between the over-approximation of reachable sets and unsafe sets. The effectiveness of the proposed approach has been validated with evaluations on a robotic arm model and an adaptive cruise control system.

Highly efficient degradation of iohexol on a heterostructured graphene-analogue boron nitride coupled Bi2MoO6 photocatalyst under simulated sunlight.

Iohexol (IOH), as a typical iodinated X-ray contrast media (ICMs) with potential threat to human health, is difficult to be removed with the conventional wastewater treatment methods. In this work, new boron nitride coupled Bi2MoO6 layered microspheres (BN/Bi2MoO6) composites were applied to remove IOH from water via photocatalytic degradation. The degradation constant kapp of IOH over 3.5 wt% BN/Bi2MoO6 was 0.016 min-1, which was 3.2 times that of Bi2MoO6 (0.005 min-1). The degradation rate of IOH on 3.5 wt% BN/Bi2MoO6 reached 92% in 150 min. The enhanced photocatalytic activity of BN/Bi2MoO6 can be attributed to the heterojunction between BN and Bi2MoO6. The matched type-I band alignment heterojunction of two semiconductors prominently improved the charge separation. Based on the trapping experiments, holes and superoxide radicals were proved to be the main active species for photocatalytic IOH degradation. Besides, the degradation products of IOH were analyzed by LC-HRMS and the possible degradation mechanism of IOH was also proposed in this work.

Enhancing the performance of pollution degradation through secondary self-assembled composite supramolecular heterojunction photocatalyst BiOCl/PDI under visible light irradiation.

A novel n-n type inorganic/organic heterojunction of flaky-like BiOCl/PDI photocatalyst was constructed by water bath heating method. Meanwhile, a simple method - secondary self-assembly was used to prepare the BiOCl/PDI with a special band structure. The photocatalytic activities were evaluated by degrading aqueous organic pollutants under visible light (λ > 420 nm). The removal rates of 5 mg L-1 phenol (non-ionic type), methyl orange (MO, anionic type), rhodamine B (RhB, cationic type) and 10 mg L-1 RhB by secondary self-assembly BiOCl/PDI (BiOCl/PDI-2) were 8.0%, 3.4%, 27.8% and 78.9% higher than self-assembly BiOCl/PDI (BiOCl/PDI-1) under visible light (λ > 420 nm). The better photocatalytic activity for BiOCl/PDI-2 was attributed to the optimization of energy-band structures, which arose from different exposed surfaces, narrower interplanar spacing and stronger visible light absorption performance. Under acidic condition, BiOCl/PDI-2 showed a good photocatalytic activity, which was not affected by neutral ionic intensity and had good recycling properties. Moreover, the photocatalytic mechanism was explored by free radical capture test and electron paramagnetic resonance (EPR), and contribution of active species was calculated. The main active species of BiOCl/PDI-2 were ·O2-, 1O2 and h+. Our work may provide a route to design efficient inorganic/organic heterojunctions for organic pollutants degradation.

Reachable set estimation and decentralized control synthesis for a class of large-scale switched systems.

The reachable set estimation and stabilization problems of large-scale switched systems have been investigated in this paper. To start with, a class of specific Lyapunov functions for the reachable set estimation of large-scale switched systems is proposed. Secondly, by the method of the proposed specific Lyapunov functions of the large-scale switched systems, the boundary of system state response is calculated under arbitrary switching, which might be used for the safety verification. Then, the concept of decentralized control is introduced and the decentralized state feedback controllers are developed to ensure the state trajectories of the closed-loop large-scale switched systems are limited within the estimated set and guarantee the system globally uniformly asymptotically stable. Finally, several numerical and practical examples are established to confirm the effectiveness and correctness of our results.

Output Reachable Set Estimation and Verification for Multilayer Neural Networks.

In this brief, the output reachable estimation and safety verification problems for multilayer perceptron (MLP) neural networks are addressed. First, a conception called maximum sensitivity is introduced, and for a class of MLPs whose activation functions are monotonic functions, the maximum sensitivity can be computed via solving convex optimization problems. Then, using a simulation-based method, the output reachable set estimation problem for neural networks is formulated into a chain of optimization problems. Finally, an automated safety verification is developed based on the output reachable set estimation result. An application to the safety verification for a robotic arm model with two joints is presented to show the effectiveness of the proposed approaches.