Two-layer Network Research Articles

Neural Network Control of Green Energy Vehicles with Blended Braking Systems

A neural network-based control system is offered, which ensures high quality blended braking of the green energy vehicles in both the intensive and the gradual deceleration scenarios, with energy recovery at the changing road pavement. In this study, a neural network controller provides the torque gradient control without a tire model resulting in returning maximal energy to the hybrid energy source during the braking process. To meet the conflicting requirements of different braking modes and road surfaces, an allocation algorithm determines how to distribute the driver’s torque request between the friction and electrical brakes. Simulation demonstrates effectiveness of the proposed braking system. Model states and inputs are used as a guidance to learn a coupled two-layer neural network capable to capture various dynamic behaviours that could not be included in the simplified physics-based model. An experimental part of the research proves the model and simulation validity.

Convex Dual Theory Analysis of Two-Layer Convolutional Neural Networks With Soft-Thresholding.

Soft-thresholding has been widely used in neural networks. Its basic network structure is a two-layer convolution neural network with soft-thresholding. Due to the network's nature of nonlinear and nonconvex, the training process heavily depends on an appropriate initialization of network parameters, resulting in the difficulty of obtaining a globally optimal solution. To address this issue, a convex dual network is designed here. We theoretically analyze the network convexity and prove that the strong duality holds. Extensive results on both simulation and real-world datasets show that strong duality holds, the dual network does not depend on initialization and optimizer, and enables faster convergence than the state-of-the-art two-layer network. This work provides a new way to convexify soft-thresholding neural networks. Furthermore, the convex dual network model of a deep soft-thresholding network with a parallel structure is deduced.

Finite-time lag synchronization for two-layer complex networks with impulsive effects

<abstract><p>This paper mainly considered the finite-time lag synchronization for two-layer complex networks with impulsive effects. Different types of controllers were designed to achieve the lag synchronization of two-layer complex networks. Several sufficient conditions on lag synchronization in the sense of finite time were derived. The time for synchronization was also estimated. It is important to note that synchronization time was influenced by the initial value, as well as the impulses and impulse sequence. This implied that different impulse effects result in varying synchronization times. Additionally, desynchronizing impulses can extend the synchronization time, whereas synchronizing impulses have the opposite effect. Finally, a numerical example was presented to showcase the practicality and validity of the proposed theoretical criteria.</p></abstract>

A Reduced Order Schwarz Method for Nonlinear Multiscale Elliptic Equations Based on Two-Layer Neural Networks

Neural networks are powerful tools for approximating high dimensional data that have been used in many contexts, including solution of partial differential equations (PDEs). We describe a solver for multiscale fully nonlinear elliptic equations that makes use of domain decomposition, an accelerated Schwarz framework, and two-layer neural networks to approximate the boundary-to-boundary map for the subdomains, which is the key step in the Schwarz procedure. Conventionally, the boundary-to-boundary map requires solution of boundary-value elliptic problems on each subdomain. By leveraging the compressibility of multiscale problems, our approach trains the neural network offline to serve as a surrogate for the usual implementation of the boundary-to-boundary map. Our method is applied to a multiscale semilinear elliptic equation and a multiscale $p$-Laplace equation. In both cases we demonstrate significant improvement in efficiency as well as good accuracy and generalization performance.

Analysis of consensus time and winning rate in two-layer networks with hypocrisy of different structures

Analysis of consensus time and winning rate in two-layer networks with hypocrisy of different structures

Two-Layer Attention Feature Pyramid Network for Small Object Detection

Two-Layer Attention Feature Pyramid Network for Small Object Detection

MGCNRF: Prediction of Disease-Related miRNAs Based on Multiple Graph Convolutional Networks and Random Forest.

Increasing microRNAs (miRNAs) have been confirmed to be inextricably linked to various diseases, and the discovery of their associations has become a routine way of treating diseases. To overcome the time-consuming and laborious shortcoming of traditional experiments in verifying the associations of miRNAs and diseases (MDAs), a variety of computational methods have emerged. However, these methods still have many shortcomings in terms of predictive performance and accuracy. In this study, a model based on multiple graph convolutional networks and random forest (MGCNRF) was proposed for the prediction MDAs. Specifically, MGCNRF first mapped miRNA functional similarity and sequence similarity, disease semantic similarity and target similarity, and the known MDAs into four different two-layer heterogeneous networks. Second, MGCNRF applied four heterogeneous networks into four different layered attention graph convolutional networks (GCNs), respectively, to extract MDA embeddings. Finally, MGCNRF integrated the embeddings of every MDA into the features of the miRNA-disease pair and predicted potential MDAs through the random forest (RF). Fivefold cross-validation was applied to verify the prediction performance of MGCNRF, which outperforms the other seven state-of-the-art methods by area under curve. Furthermore, the accuracy and the case studies of different diseases further demonstrate the scientific rationale of MGCNRF. In conclusion, MGCNRF can serve as a scientific tool for predicting potential MDAs.

Research on control strategy of pneumatic soft bionic robot based on improved CPG.

To achieve the accuracy and anti-interference of the motion control of the soft robot more effectively, the motion control strategy of the pneumatic soft bionic robot based on the improved Central Pattern Generator (CPG) is proposed. According to the structure and motion characteristics of the robot, a two-layer neural network topology model for the robot is constructed by coupling 22 Hopfield neuron nonlinear oscillators. Then, based on the Adaptive Neuro-Fuzzy Inference System (ANFIS), the membership functions are offline learned and trained to construct the CPG-ANFIS-PID motion control strategy for the robot. Through simulation research on the impact of CPG-ANFIS-PID input parameters on the swimming performance of the robot, it is verified that the control strategy can quickly respond to input parameter changes between different swimming modes, and stably output smooth and continuous dynamic position signals, which has certain advantages. Then, the motion performance of the robot prototype is analyzed experimentally and compared with the simulation results. The results show that the CPG-ANFIS-PID motion control strategy can output coupled waveform signals stably, and control the executing mechanisms of the pneumatic soft bionic robot to achieve biological rhythms motion propulsion waveforms, confirming that the control strategy has accuracy and anti-interference characteristics, and enable the robot have certain maneuverability, flexibility, and environmental adaptability. The significance of this work lies in establishing a CPG-ANFIS-PID control strategy applicable to pneumatic soft bionic robot and proposing a rhythmic motion control method applicable to pneumatic soft bionic robot.

Development of a route setting model in the dispatching centralization system

These days, automation and remote control technologies provide the foundation for the organization of train movement across several categories. At railroad sections and stations, dispatcher centralization systems are utilized to guarantee the security of train movement. However, in the railroad portions where high-speed train movements are organized, time limitations based on the human factor are being implemented. Consequently, this results in longer wait times for technological processes at stations and longer train intervals on some railway segments. Developing an automated route preparation procedure at stations in dispatching centralization systems will allow for the resolution of current issues. A dispatching centralized system model for route establishment has been created in this scientific study. Using a two-layer neural network, a mathematical model of route setting in the dispatcher centralization system was determined. Neural networks are employed in rail dispatch rooms for intelligent route comparison and in dispatcher centralized systems for tracking train routes at stations.

Exploring soil multi-parameter stacking measurement through Raman and NIR dual-spectroscopy.

The excessive use of fertilizers can lead to increased production costs, degraded soil quality, diminished product excellence, and environmental contamination. To address this issue, a solution involving soil testing and customizing fertilizer application has been proposed. The current standard methodology for soil parameter assessment relies on chemical analysis performed by trained laboratory technicians, which only allows for the measurement of one indicator at a time. Hence, a novel approach utilizing the fusion of near-infrared (NIR) and Raman dual-spectral features has been suggested to simultaneously determine five crucial indicators (hydrolyzed N, available P, quick-release K, OM, and pH) in soil with a single scan. In this research, seven preprocessing techniques and four feature extraction methods were initially explored to optimize the composite NIR and Raman feature variables. Subsequently, a regressor with a two-layer network structure (RF, LR, SVR; ELM, and PLS) was developed using the stacking algorithm. This methodology synergizes the strengths of the five base learners, minimizes the risk of overfitting, and demonstrates high computational efficiency for linear data correlations and robust fitting capabilities for nonlinear data correlations. Additionally, it showcases strong generalization capabilities, noise resilience, and robustness. The model produced relevant results for hydrolyzed N, available P, quick release K, OM, and pH measurements, with Rp2 values of 0.9966, 0.9722, 0.9855, 0.9557, and 0.9951, RMSEP values of 2.9547, 2.9972, 7.6550, 0.0765, and 0.0313, and RPD values of 6.0855, 2.4655, 3.0511, 8.3084, and 10.6977. This work delivers a twofold contribution by presenting a swift method for simultaneous measurement of multiple soil parameters, enabling concurrent ploughing, soil surveying, and fertilizer application. Furthermore, it introduces a stacking measurement model based on dual fusion features, showcasing detailed model parameters. The stacking model outperformed mono-spectral models (NIR and Raman) and the dual PLS model in terms of Rp2, RPD, and RMSEP values, and fluctuation ranges, demonstrating enhanced stability, predictive prowess, and reliable observations. Overall, the stacking model offers a cost-effective, rapid, and precise solution for online evaluation of soil physicochemical conditions, catering to the requirements of modern agricultural production well. This innovative approach signifies a significant leap forward and provides a solid theoretical foundation for the enhancement of associated online monitoring systems and tools.

Adaptive Neural Consensus Observer Networks Design for a Class of Semilinear Parabolic PDE Systems.

This article concerns the investigation on the consensus problem for the joint state-uncertainty estimation of a class of parabolic partial differential equation (PDE) systems with parametric and nonparametric uncertainties. We propose a two-layer network consisting of informed and uninformed boundary observers where novel adaptation laws are developed for the identification of uncertainties. Particularly, all observer agents in the network transmit their information with each other across the entire network. The proposed adaptation laws include a penalty term of the mismatch between the parameter estimates generated by the other observer agents. Moreover, for the nonparametric uncertainties, radial basis function (RBF) neural networks are employed for the universal approximation of unknown nonlinear functions. Given the persistently exciting condition, it is shown that the proposed network of adaptive observers can achieve exponential joint state-uncertainty estimation in the presence of parametric uncertainties and ultimate bounded estimation in the presence of nonparametric uncertainties based on the Lyapunov stability theory. The effects of the proposed consensus method are demonstrated through a typical reaction-diffusion system example, which implies convincing numerical findings.

Modeling urban rail transit system resilience under natural disasters: A two-layer network framework based on link flow

Modeling urban rail transit system resilience under natural disasters: A two-layer network framework based on link flow

Artificial neural network-based prediction model of elastic floor response spectra incorporating dynamic primary-secondary structure interaction

The evaluation of the Floor Response Spectrum (FRS) holds paramount significance in assessing the seismic behavior of secondary structures. Precise FRS prediction empowers engineers to make informed decisions concerning structural design, retrofitting, and safety precautions. This study aims to scrutinize the impact of dynamic interaction between primary and secondary structures on FRS. Both the elastic primary structure (PS) and elastic secondary structure (SS) employ a single-degree-of-freedom (SDOF) system. Governing motion equations for both coupled (with dynamic interaction) and uncoupled (without dynamic interaction) systems are formulated and solved numerically. The study investigates how variations in the vibration period of PS (Tp), tuning ratio (Tr), mass ratio (μ), and damping ratio (ξs) of SS influence FRS. The FRS impact remains minimal at μ = 0.001 (0.1%); however, with increasing mass ratio, PS-SS dynamic interaction significantly affects SS's spectral acceleration response. Coupled analysis is crucial only for secondary structures tuned to the primary structure's vibration period (0.8≤Tr≤1.2). This study utilizes two-layer feed-forward Artificial Neural Networks (ANNs) for FRS prediction. The Levenberg-Marquardt (LM) backpropagation (BP) algorithm trains the network using a comprehensive dataset. In summary, it is evident that the ANNs, once trained, enable accurate prediction of the FRS, exhibiting a R2 of 99%. Additionally, a design expression is formulated utilizing the ANN model and subsequently compared with the existing formulation.

LSTM-Based Virtual Load Sensor for Heavy-Duty Vehicles.

In this paper, a special recurrent neural network (RNN) called Long Short-Term Memory (LSTM) is used to design a virtual load sensor that estimates the mass of heavy vehicles. The estimation algorithm consists of a two-layer LSTM network. The network estimates vehicle mass based on vehicle speed, longitudinal acceleration, engine speed, engine torque, and accelerator pedal position. The network is trained and tested with a data set collected in a high-fidelity simulation environment called Truckmaker. The training data are generated in acceleration maneuvers across a range of speeds, while the test data are obtained by simulating the vehicle in the Worldwide harmonized Light vehicles Test Cycle (WLTC). Preliminary results show that, with the proposed approach, heavy-vehicle mass can be estimated as accurately as commercial load sensors across a range of load mass as wide as four tons.

Prediction of carbon content in the metal of final blow period in BOF using neural network

Prediction and control of the carbon content after the end of oxygen blow in BOF converter are key points of steel production efficiency. One of the most accurate methods is the dynamic predicting method based on the use of intermediate sublance measurement (TSC probe) when about 85 – 90 % of total oxygen is consumed and on the final period model. Models of the final period are traditionally based on exponential or cubic functions, currently there are developments based on neural network technologies. We investigated the possibility of using a neural network to predict the final carbon content using the results of intermediate sublance measurement (TSO probe) when about 95 % of total oxygen is consumed. As a model of the final period, a two-layer neural network with one hidden layer and an activation function of the Softplus type for all neurons was implemented in software. The input vectors contain initial carbon content and oxygen consumption for the second blow values. The output vector contains the predicted final carbon content, the output training vector - actual final carbon content values. The network was trained on 700 heats data of the training set. The model trained in this way was tested on 232 heats data of the testing set. The prediction errors distribution and values of the mean absolute error and root mean square error for the training and testing sets are correspondingly close. They are also comparable with similar indicators of the heats, the final period of which was carried out without oxygen blow (only flux additions and/or nitrogen blow), and this indicates a high accuracy of the prediction.

The diffusion of platform self-operation on reputation-based two-layer network

PurposeDue to the cross-network effect, two-sided users communicate with each other, producing a coupling network. To study the spread of platform self-operation in two-sided users' marketing and purchasing tactics, this paper considers the differences in reputation acquired by platform-owned and third-party operating channels.Design/methodology/approachThis study proposes a two-layer network with cross-network links: one layer represents the social network of consumers, while the other layer represents the competitive network of buyers. A closed system of differential equations, based on the binary dynamics of the stochastic network, is developed to study the trend and stability points of the platform self-operation dissemination. Then the overall benefits of platform are analyzed to unify the platform diffusion and pricing strategies.FindingsThe degree of difference in social influence and cross-network effects affect diffusion synergistically. Cross-network effects hinder diffusion when there is a significant difference of social influence between consumers and sellers but promote diffusion when there is little difference of social influence between consumers and sellers. Additionally, the network weights and reputation gap exhibit a nonlinear correlation with diffusion. For pricing strategy of the platform, it can achieve maximum profit when the pricing of self-operated goods and third-party-operated goods is equal.Originality/valueThis study considers the complex network architecture created by bilateral markets and the dynamic influence of group interactions on product. Additionally, this study takes reputation into account when considering the price and dissemination tactics of various operating channels, offering guidelines for platforms to control merchants and mediate disputes between various operating channels.

A two-layer model with partial mapping: Unveiling the interplay between information dissemination and disease diffusion

This study delves into the pivotal role of information dissemination in public health, particularly how it influences the spread of diseases. By implementing a sophisticated two-layer partial mapping network model (UAU-SIRS), we investigate the dynamic relationship between information flow and disease transmission. Our approach utilizes extensive multiplexed network data, processed through a micro Markov chain (MMC) model, to simulate the interplay between information spread and disease dynamics. The findings reveal a noteworthy positive correlation between the rates of information dissemination, recovery in the network, and the epidemic threshold. Conversely, the conversion rate is inversely related to this threshold. A critical observation is that Scale-free (SF) networks, characterized by their uneven node distribution, are more susceptible to the impacts of information spread on their outbreak thresholds compared to Erdős-Rényi (ER) networks. This research offers crucial insights for epidemic prevention strategies and provides valuable guidance for managing the dissemination of disease-related information within complex network structures.

Development and validation of BVB-UIE: an underwater image-enhancement model with two-layer deep neural network

Due to the absorption and scattering of light in water, underwater visual visibility is poor, and underwater images usually suffer from color distortion and feature loss, which poses a great challenge for underwater vision tasks. To handle these issues, we present an underwater image enhancement (UIE) method. A Gaussian pyramid is constructed for the degraded underwater images based on an improved visual saliency model, with the characteristics of luminance, orientation, texture, and color. By combining channel and spatial attention mechanisms, a deep asymmetric feature enhancement network is built to preserve the high-dimensional features of the image. In addition, a polynomial loss function is used for the depth hopping supervising constraints during the enhancement process, and the gating signals are employed to control the importance of the features at different spatial locations. These innovations maximize the efficiency of the feature extraction while simplifying the network complexity. Experiments on an open benchmark dataset for UIE show that our method effectively eliminates the color bias and contrast distortion problems while preserving the complex image details compared to the existing UIE algorithms. Objective metrics show a significant improvement in the algorithm, with a 15% increment in the peak signal-to-noise ratio metric compared to the closest competing algorithm.

Synchronization in the two time-scale multiplex network with symplectic interactions

Mounting experimental evidence suggests a significant role for glial cells, particularly astrocytes, in shaping neuronal activity in both cell cultures and the brain, driving interest in exploring multi-component network dynamics [1]. These systems are modeled at varying levels of physiological detail, ranging from the Hodgkin-Huxley and Ullah models for neurons and astrocytes, respectively, to simple integrate-and-fire models or phase oscillators. Multiplex networks are commonly used to model generic dynamical phenomena, with an emphasis on network connectivity effects rather than the complexity of an individual cell’s behavior. This model is particularly advantageous due to its use of a two-layer network structure [2]. This structure accurately describes neural connectivity through the long-range random coupling, while glial cells interact locally through the diffusion of mediatory molecules. Furthermore, the oscillatory timescales between neural and glial cells differ by at least one order of magnitude. In this context, the analysis of synchronization is a focal point as it is a crucial process that underpins information processing, decision making, and movement control in living neural systems [3]. Network topology is known to be essential, where all-to-all coupling and random Erdos–Renyi connectivity support the second-order phase transition to global phase coherence. In contrast, regular local coupling such as a 2D lattice only permits frequency and phase locking. Additionally, symplectic networks are capable of demonstrating a first-order phase transition, which is commonly referred to as “explosive” synchronization [4]. Recently, studies have shown that combining model neural (random) and glial (regular lattice) oscillatory layers can lead to a variety of outcomes [5]. Specifically, this combination was found to induce a second-order phase transition in the regular glial layer. In addition, synchronization in both layers may be preceded by desynchronization as the coupling between the layers becomes stronger. Here we investigate synchronization in the multiplex neural-glial network, where the neural layer is symplectic and contains triadic interactions. We demonstrate that symplectic coupling in the neural layer can induce the first order transition in the regular (glial) layer. Additionally, we show that such a transition can be induced by strengthening the glial and interlayer coupling, even if the symplectic neural layer alone is below the synchronization threshold. Desynchronization in the neuronal layer, resulting from the moderated coupling to the glial layer, never occurs abruptly.

Data mining and spatio-temporal characteristics of urban road traffic emissions: A case study in Shijiazhuang, China

Accurate estimation of traffic emissions and analysis of spatio-temporal distribution on urban roads play a crucial role in the development of low-carbon transportation system. Traditionally, a region’s emission characteristics have been studied using numerous emission models with GPS-based spatio-temporal data. Due to the heavy data processing needs of GPS-based data, emission characteristics for a large region have been studied by dividing the region into a limited number of smaller areas or units. Additionally, GPS data are based on a few vehicles in the traffic which does not fully reflect road conditions. This paper proposed an approach that can be used to study and calculate the spatio-temporal emission pattern of a region at a roadway section level by using Baidu’s online traffic data and COPERT model. The proposed method can be used to estimate road-level emission patterns while avoiding the impact of redundant data in large datasets, making the dataset more reliable, applicable, and scalable. The proposed approach has been demonstrated through a study of spatio-temporal emission patterns in the Qiaoxi district within city of Shijiazhuang, China. Online data crawling technology was used to obtain data on urban road traffic speed and driving distance. The linear reference technology was used to construct a two-layer road network model to conduct the coupling and matching of traffic data with the road network data. The COPERT model was implemented to calculate the average traffic emissions on each road in the road network, and a traffic emission intensity index was proposed to quantify the CO, VOC, NOx and CO2 emissions on urban roads in the study area. The analysis results show that the traffic emission intensity of the expressway, trunk road, secondary road, and branch road is high during the morning peak (7 AM-9 AM) and evening peak (5 PM—7 PM). The sections with higher traffic emission intensity are mainly concentrated on the main roads and secondary roads such as Jiefang South Street, Shitong Road and Xinhua Road. Nearly one-third of 2nd Ring and 3rd Ring roads also have relatively high emission intensity. The research results provide new ideas for estimating traffic emissions in urban road networks and analyzing the spatio-temporal distribution of traffic emissions. The research results can also provide a decision-making basis for traffic management departments to formulate energy-saving and emission-reduction measures and promote the development of urban green and low-carbon transportation.