Search

Predicting pavement surface friction during the design stage allows engineers to optimize the design of the roadway to provide the appropriate level of friction for the intended use of the road in a safe and cost-effective manner. The main goal of the study is to propose a methodology to predict pavement surface friction during the design stage. Thus, this study analyzes the role of aggregate Fineness Modulus (FM) and Hot Mix Asphalt (HMA) volumetrics including Air Voids Volume (Va) and Effective Binder Volume (Vbe) on fabricating the surface texture. Surface frictional properties were evaluated using the British Pendulum Test (BPT) and the Sand Patch Test (SPT). The data were analyzed using the analysis of variance (ANOVA) test. Several statistical modeling techniques including Multiple Linear (ML) regression, Non-Linear Stepwise (NLSW) regression with all possible interactions, Non-Linear Beta (NLB) regression, Non-Linear Curve Fitting (NLCF) regression, and multilayer neural network (MNN) were utilized. Models were evaluated using synthetical data and compared using Post-Hoc analysis. The study evaluated nine types of mixes including different gradations with different Nominal Maximum Aggregate Sizes (NMAS) and several asphalt modifiers. The results revealed that Mean Textures Depth (MTD) and British pendulum Number (BPN) values are primarily influenced by FM, followed by Va and Vbe, respectively. According to ANOVA results, the two-level interaction showed that only when FM interacts with either Va or Vbe, the interaction is significant for both MTD and BPN. MNN models had the highest Coefficient of Determination (R2) values for both MTD and BPN. However, the sensitivity analysis and the Post-Hoc analysis revealed that due to the low number of data used to generate the models, statistical regression methods had comparable results and resulted in more accurate prediction than MNN. The NLCF was found to be the most reliable model for predicting both BPN and MTD. [ABSTRACT FROM AUTHOR]

Power Quality, Equipment and Personnel safety of any distributed generation (DG) system connected to utility Grid merely depends on accurate detection of Islanding and non-islanding Power quality disturbances. The main objective of the proposed research is to detect islanding events with very narrow non-detection zone (NDZ) and classification of power quality disturbances with higher accuracy using signal processing and intelligent method together. A noise robust down sampling empirical mode decomposition (DEMD) is used to extract signature of islanding and power quality (PQ) disturbance features from the collected voltage signals and multilayer perceptron neural network (MLNN) is proposed to classify islanding and non-islanding (PQ) events. The performance of the proposed (DEMD-MLNN) technique is verified with IEEE-9 bus distributed generation system dominated by solar &wind energy penetration. The simulation work is carried out in MATLAB/Simulink platform. The efficacy of the proposed DEMD-MLNN is verified by large number of numerical experimentations with and without noise and comparing with existing competitive well-known techniques. [ABSTRACT FROM AUTHOR]

Multilayer neural networks can approximate endogenous disturbances with relatively high accuracy. However, for multilayer‐neural‐network‐based control methods of high‐order uncertain nonlinear systems, hard to handle large exogenous disturbances especially for mismatched types, complex controller scheme and so on, make them difficult to be practical. Therefore, a novel high‐performance multilayer neurocontroller which can simultaneously reject matched and mismatched disturbances will be proposed in this paper. Specially, strong endogenous and exogenous disturbances will be feedforwardly compensated. Additionally, the proposed controller not only protects from "explosion of complexity," but also owns a simple scheme. [ABSTRACT FROM AUTHOR]

In this paper, we investigate lifelong learning (LL)‐based tracking control for partially uncertain strict feedback nonlinear systems with state constraints, employing a singular value decomposition (SVD) of the multilayer neural networks (MNNs) activation function based weight tuning scheme. The novel SVD‐based approach extends the MNN weight tuning to n$$ n $$ layers. A unique online LL method, based on tracking error, is integrated into the MNN weight update laws to counteract catastrophic forgetting. To adeptly address constraints for safety assurances, taking into account the effects caused by disturbances, we utilize a time‐varying barrier Lyapunov function (TBLF) that ensures a uniformly ultimately bounded closed‐loop system. The effectiveness of the proposed safe LL MNN approach is demonstrated through a leader‐follower formation scenario involving unknown kinematics and dynamics. Supporting simulation results of mobile robot formation control are provided, confirming the theoretical findings. [ABSTRACT FROM AUTHOR]

This research is concerned with finding the optimal solution of a job shop scheduling problem (JSSP) in as short time as possible. A JSSP can be formulated as a 0-1 mixed integer programming (0-1 MIP), and it is expected that the optimal schedule can be obtained in a shorter time by predicting a good solution based on data of scheduling performed previously and using the solution as the initial solution for the optimization algorithm. In this concept, it is an important point to predict a solution which satisfy constraints of the 0-1 MIP. This paper provides an improved method based on this concept where the prediction is carried out so that the constraints are always satisfied. Numerical experiments showed that solution time of the proposed method is shorter than that of the previous method, which does not assure the constraints are satisfied, and learning time reduces about half. In addition, it turned out that there is a reason of the shorter solution time other than that the predicted solution always satisfies the constraints. One possible reason is that the number of circular sequences tends to be smaller than the previous method. Further analysis and additional evaluations from this point of view will be performed in a future work. [ABSTRACT FROM AUTHOR]

This paper concerns the problem of adaptive compensation tracking control for a class of time-varying delay nonlinear systems with unknown structures and unknown actuator dead zones where time-varying delays are unknown. First, a variable separation approach is used to overcome the difficulty in dealing with a nonstrict-feedback structure, and multilayer neural networks are used to approximate the unknown nonlinear structures with time-varying delays. On this basis, we designed an adaptive multilayer neural-network compensation controller to reduce the error of multilayer neural networks. Furthermore, for unknown actuator dead zones, this paper separates the controller and adopts multilayer neural networks to deal with unknown actuator dead zones. In order to reduce the error of the dead-zone controller, wer designed an adaptive compensation controller for the dead zones. Lastly, this paper proves the stability of the systems with the Lyapunov method, and simulation results demonstrate the effectiveness of the scheme. [ABSTRACT FROM AUTHOR]

Lithium batteries are secondary batteries used as power sources in various applications, such as electric vehicles, portable devices, and energy storage devices. However, because explosions frequently occur during their operation, improving battery safety by developing battery management systems with excellent reliability and efficiency has become a recent research focus. The performance of the battery management system varies depending on the estimated accuracy of the state of charge (SOC) and state of health (SOH). Therefore, we propose a SOH and SOC estimation method for lithium–ion batteries in this study. The proposed method includes four neural network models—one is used to estimate the SOH, and the other three are configured as normal, caution, and fault neural network model banks for estimating the SOC. The experimental results demonstrate that the proposed method using the long short-term memory model outperforms its counterparts. [ABSTRACT FROM AUTHOR]

In this brief, we propose a new neuro-fuzzy reinforcement learning-based control (NFRLC) structure that allows autonomous underwater vehicles (AUVs) to follow a desired trajectory in large-scale complex environments precisely. The accurate tracking control problem is solved by a unique online NFRLC method designed based on actor-critic (AC) structure. Integrating the NFRLC framework including an adaptive multilayer neural network (MNN) and interval type-2 fuzzy neural network (IT2FNN) with a high-gain observer (HGO), a robust smart observer-based system is set up to estimate the velocities of the AUVs, unknown dynamic parameters containing unmodeled dynamics, nonlinearities, uncertainties and external disturbances. By employing a saturation function in the design procedure and transforming the input limitations into input saturation nonlinearities, the risk of the actuator saturation is effectively reduced together with nonlinear input saturation compensation by the NFRLC strategy. A predefined funnel-shaped performance function is designed to attain certain prescribed output performance. Finally, stability study reveals that the entire closed-loop system signals are semi-globally uniformly ultimately bounded (SGUUB) and can provide prescribed convergence rate for the tracking errors so that the tracking errors approach to the origin evolving inside the funnel-shaped performance bound at the prescribed time.

Composite influences coming from signal measurement noises, unknown nonlinear dynamics, external disturbances and input saturation nonlinearity make it challenging to synthesize high-performance closed-loop control algorithms for uncertain robotic manipulators. In the face of these challenges, we employ the nonlinear multilayer neural networks to approach uncertain nonlinear dynamics and exploit the robust adaptive control to deal with external disturbances without knowing their bounds in advance. More importantly, robust adaptive based auxiliary functions are creatively introduced to offset the possible input saturation nonlinearity. Furthermore, the desired trajectory based model compensation technology is integrated into the control scheme to reduce measurement noises as much as possible. In theory, the global closed-loop stability of the dynamical uncertain system is testified and significant asymptotic tracking result can be acquired. The application verification under different working conditions including severe high-frequency working conditions is implemented to indicate the high-performance effect of the synthesized intelligent controller. [ABSTRACT FROM AUTHOR]

The paper observes a similarity between the stochastic optimal control of discrete dynamical systems and the learning multilayer neural networks. It focuses on contemporary deep networks with nonconvex nonsmooth loss and activation functions. The machine learning problems are treated as nonconvex nonsmooth stochastic optimization problems. As a model of nonsmooth nonconvex dependences, the so-called generalized-differentiable functions are used. The backpropagation method for calculating stochastic generalized gradients of the learning quality functional for such systems is substantiated basing on Hamilton–Pontryagin formalism. Stochastic generalized gradient learning algorithms are extended for training nonconvex nonsmooth neural networks. The performance of a stochastic generalized gradient algorithm is illustrated by the linear multiclass classification problem. [ABSTRACT FROM AUTHOR]

In this paper, parametric optimization problems are investigated. In a parametric optimization problem we assume ∈ Rn is the vector of the parameters and x* is the optimal answer corresponding to it. The purpose of this paper is to determine a function such as so that we have ψ = x*. To do this, first for each, the corresponding optimal answer is calculated. In this way, a set of data bases consisting of parameters and the corresponding optimal values are obtained. A multilayer network of data base is trained to obtain the function ψ in a domain. In fact, the function ψ for each value of the parameter specifies the corresponding answer by the trained multilayer network. Finally, we conduct several numerical examples to test our method. [ABSTRACT FROM AUTHOR]

This paper concerns the problem of adaptive compensation tracking control for a class of time-varying delay nonlinear systems with unknown structures and unknown actuator dead zones where time-varying delays are unknown. First, a variable separation approach is used to overcome the difficulty in dealing with a nonstrict-feedback structure, and multilayer neural networks are used to approximate the unknown nonlinear structures with time-varying delays. On this basis, we designed an adaptive multilayer neural-network compensation controller to reduce the error of multilayer neural networks. Furthermore, for unknown actuator dead zones, this paper separates the controller and adopts multilayer neural networks to deal with unknown actuator dead zones. In order to reduce the error of the dead-zone controller, wer designed an adaptive compensation controller for the dead zones. Lastly, this paper proves the stability of the systems with the Lyapunov method, and simulation results demonstrate the effectiveness of the scheme.

Abstract Machine learning (ML) techniques play a crucial role in today's modern world. Over the last years, road traffic safety is one of the applications where ML‐methods have been successfully employed to prevent road users from being killed or seriously injured. A reliable data‐driven predictive model is essential for this purpose. This could be achieved by successfully applying an intelligent transportation system to identify a driver at a higher risk of crashes. This study investigates the capabilities of different deep learning techniques to predict motorcycle crash severity. This study is based on 2,430 motorcycle crashes in a mountainous area in the United States over a 10‐year period. Different deep networks (DNNs), including deep belief network, standard recurrent neural network (RNN), multilayer neural network, and single‐layer neural network, were considered and compared in terms of prediction accuracy of motorcycle crash severity. Before conducting any analysis, feature reduction was performed to identify the optimal number of variables to include in the models by minimizing the error rate. Different metrics including the area under the curve and confusion matrix were used to compare the different models. Although the analyses were conducted on a relatively small dataset, the results indicate that almost all the DNN models better perform in predicting the severity of motorcycle crashes, compared with the single layer neural network. Finally, the RNN outperforms the other three neural network models. A comprehensive discussion has been made about the methodological approach implemented in this study.

It is assumed that by 2021, about half of the companies will use digital counterparts of different levels. The simplest digital twin models may not use machine learning, but the models using machine learning algorithms will have the greatest advantage. In this article, we offer our approach to the construction of digital twins for real objects. We rely on our unified process of constructing approximate solutions of boundary value problems for equations of mathematical physics and accumulated experience in solving numerous specific problems of this type. In this paper, we present five approaches to the construction of digital twin models based on the evolutionary algorithms developed and tested by us. The peculiarity of our approach to evolutionary algorithms is the use of genetic procedures for constructing the structure of the model and nonlinear optimization algorithms for adjusting its parameters. Also, we propose our approach to the construction of multilayer models upon differential equations, which allows doing without the time-consuming procedure of neural networks training. We are confident that the proposed approaches can significantly simplify and unify the creation and adaptation (keeping up to date) of digital twins for real objects of various kinds – technical, biological, socio-economic, etc.

Machine learning (ML) techniques play a crucial role in today's modern world. Over the last years, road traffic safety is one of the applications where ML‐methods have been successfully employed to prevent road users from being killed or seriously injured. A reliable data‐driven predictive model is essential for this purpose. This could be achieved by successfully applying an intelligent transportation system to identify a driver at a higher risk of crashes. This study investigates the capabilities of different deep learning techniques to predict motorcycle crash severity. This study is based on 2,430 motorcycle crashes in a mountainous area in the United States over a 10‐year period. Different deep networks (DNNs), including deep belief network, standard recurrent neural network (RNN), multilayer neural network, and single‐layer neural network, were considered and compared in terms of prediction accuracy of motorcycle crash severity. Before conducting any analysis, feature reduction was performed to identify the optimal number of variables to include in the models by minimizing the error rate. Different metrics including the area under the curve and confusion matrix were used to compare the different models. Although the analyses were conducted on a relatively small dataset, the results indicate that almost all the DNN models better perform in predicting the severity of motorcycle crashes, compared with the single layer neural network. Finally, the RNN outperforms the other three neural network models. A comprehensive discussion has been made about the methodological approach implemented in this study. [ABSTRACT FROM AUTHOR]

In the presence of uncertain dynamic terms and external disturbances, the problem of trajectory tracking with application to an underactuated underwater vehicle is addressed in this paper. Based on Lyapunov theory and properties of neural networks, a nonlinear neural controller is designed, where multilayer neural networks are adopted to approximate the unmodeled dynamic terms and external disturbances. In order to confine the values of estimated weights within predefined bounds, smooth projection functions are employed. Moreover, measurement noises are considered so as to simulate realistic operation scenario, while filters are designed to get cleaner states. From the stability analysis, it is proven that the tracking errors are globally uniformly ultimately bounded. Numerical examples are provided to demonstrate the robustness of the controller in the presence of unmodeled terms, disturbances and measurement noises. [ABSTRACT FROM AUTHOR]

Problems of nonsmooth nonconvex dynamic optimization, optimal control (in discrete time), including feedback control, and machine learning are considered from a common point of view. An analogy between controlling discrete dynamical systems and multilayer neural network learning problems with nonsmooth objective functionals and connections is traced. Methods for computing generalized gradients for such systems based on the Hamilton–Pontryagin functions are developed. Gradient (stochastic) algorithms for optimal control and learning are extended to nonconvex nonsmooth dynamic systems. [ABSTRACT FROM AUTHOR]

Copyright of Cybernetics & Systems Analysis / Kibernetiki i Sistemnyj Analiz is the property of V.M. Glushkov Institute of Cybernetics of NAS of Ukraine and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Summary: In this paper, a neural network (NN)-based tracking controller is proposed for a general class of type (m , s) wheeled mobile manipulators (WMMs) subjected to model uncertainties with prescribed transient and steady-state performance specifications. First, an input–output model of WMMs is derived by introducing proper output equations. Then, the prescribed performance technique is employed to propose a proportional integral derivative trajectory tracking controller for WMMs to ensure that the tracking errors converge to a smaller, arbitrary ultimate bound with a predefined maximum overshoot/undershoot and convergence speed. The learning capabilities of multilayer NNs are incorporated into the controller to approximate the uncertain nonlinear dynamics of the robot. An adaptive saturation-type controller is utilized to compensate NN estimation errors and external disturbances. A Lyapunov-based stability analysis is used to demonstrate that the tracking errors are uniformly ultimately bounded and converge to a small neighborhood of zero with a guaranteed prescribed performance. Numerical computer simulations are presented to show the effectiveness of the proposed controller. [ABSTRACT FROM AUTHOR]

Biology‐inspired neural computing is a potential candidate for the implementation of next‐generation intelligent systems. Memristor is a passive electrical element with resistance‐switching dynamics. Owing to its natural advantages of non‐volatility, nanoscale geometries, and variable conductance, memristor can effectively simulate the synaptic connecting strength between the neurones in the multilayer neural networks. This study presents a kind of memristor synapse‐based multilayer neural network hardware architecture with a suitable training methodology. Specifically, a novel dual‐complementary metal–oxide–semiconductor/memristor synaptic circuit is presented, which is capable of performing the negative, zero, and positive synaptic weights via controlling the direction of current passing through the memristors. Then, the neurone circuit synthesised with multiple synaptic circuits and an activation unit is further designed, which can be utilised to constitute a compact multilayer neural network with fully connected configuration. Also, a hardware‐friendly chip‐in‐the‐loop training method is provided during the network training phase. For the verification purpose, the presented neural network is applied for the realisation of single image super‐resolution reconstruction. [ABSTRACT FROM AUTHOR]

Multilayer neural networks (MNNs) have achieved excellent performance in machine‐learning domain. Memristors are a possible device for implementing MNNs in hardware with efficiency and limited area. In this work, a simple model of stochastic memristors was presented first. Then, an MNN architecture based on proposed memristor model was presented. The simulation processes on stochastic memristors were elaborated. The simulation demonstrates that the MNN classification accuracy based on stochastic memristors is usually higher than that based on deterministic memristors when the dataset noise is low. The results have significant meaning to develop analogue memristive devices or memristive chips for MNN applications. [ABSTRACT FROM AUTHOR]

The paper deals with a parameter identification problem for creep and fracture model. The system of ordinary differential equations of kinetic creep theory is applied for describing this model. As for solving the parameter identification problem, we proposed to use the technique of neural network modeling, as well as the multilayer approach. The procedures of neural network modeling and multilayer approximation constructing application is demonstrated by the example of finding parameters for uniaxial tension model for isotropic steel 45 specimens at creep conditions. The solution corresponding to the obtained parameters agrees well with theoretical strain-damage characteristics, experimental data, and results of other authors. [ABSTRACT FROM AUTHOR]

For servo electromechanical systems, the existing modeling uncertainties, signal measurement noises and so on always make it difficult to design high-performance closed-loop controllers. In this paper, a novel intelligent controller will be designed to deal with these uncertainties. Specifically, two multilayer neuroadaptive disturbance observers will be proposed to estimate the uncertain nonlinear dynamics and exogenous disturbances simultaneously. And these uncertainties will be compensated feedforwardly. Moreover, in order to reduce the influence of signal measurement noises, the desired-command compensation technique will be incorporated. Additionally, different validation examples will be proposed to demonstrate the advantages of the designed controller. • The algorithm can compensate matched and mismatched nonlinear dynamics and external disturbances. • The approximation ability of the neural networks is further improved. • It can be easily transplanted to other servo systems without much change. [ABSTRACT FROM AUTHOR]

One of the major problems in budgeting is to predict the various kinds of future income precisely as possible. Since tax revenue is very important component in the combination of state income, the present paper considers the forecasting of VAT on gasoline consumption. The main purpose is to achieve an efficient method to forecast gasoline consumption and VAT on it in Iran. Hence, a Hybrid ARIMA- Neural Network model is used to forecast gasoline consumption. After confirming the good performance of this method compared with autoregressive integrated moving average processes(ARIMA), VAT on gasoline consumption is calculated by applying its tax rate. Results indicate that during the years 2013 to 2016, VAT on gasoline consumption will grow by 31.6 percent on average.

A novel learning approach for a composite function that can be written in the form of a matrix system of linear equations is introduced in this paper. This learning approach, which is gradient-free, is grounded upon the observation that solving the system of linear equations by manipulating the kernel and the range projection spaces using the Moore–Penrose inversion boils down to an approximation in the least squares error sense. In view of the heavy dependence on computation of the pseudoinverse, a simplification method is proposed. The learning approach is applied to learn a multilayer feedforward neural network with full weight connections. The numerical experiments on learning both synthetic and benchmark data sets not only validate the feasibility but also depict the performance of the proposed formulation.

Highlights • An efficient machine learning algorithm for emotional mapping based upon EDA is proposed. • The model is based on emotion dimensions of Arousal, Valence and Dominance. • A set of features from different signal representations is extracted. • Optimal features subset and network configuration is identified. Abstract In this paper, a machine learning algorithm is proposed for emotional pattern recognition during audio-visual stimuli (music videos) using Electrodermal Activity (EDA). For emotion prediction apart from conventional time domain features of EDA signal, various features in different signal representation i.e. frequency and wavelet were analysed. The comparative result indicated that the wavelet features subset outperformed the conventional time domain features in term of classification accuracy. For identification of optimal network configuration, various combination of optimization algorithms (i.e. backpropagation algorithms) and error function were explored. The best performance of 79% for arousal, 69.8% for valence and 71.2% for dominance were obtained for emotion recognition respectively. [ABSTRACT FROM AUTHOR]

Deep neural networks (DNNs) could help to identify the internal sources of volcano-seismic events. However, direct applications of DNNs are challenging, given the multiple seismic sources and the small size of available datasets. In this paper, we propose a novel approach in the field of volcano seismology to classify volcano-seismic events based on fully connected DNNs. Two DNN architectures with different weights scheme initialization are studied: stacked denoising autoencoders and deep belief networks. Using a combined feature vector of linear prediction coefficients and statistical properties, we evaluate classification performance on seven different classes of isolated seismic events. These proposed architectures are compared to multilayer perceptron, support vector machine, and random forest. Experimental results show that DNNs can efficiently capture complex relationships of volcano-seismic data and achieve better classification performance with faster convergence when compared to classical models. [ABSTRACT FROM PUBLISHER]