This paper presents an adaptive neural network (NN) control of a two-degree-of-freedom manipulator driven by an electrohydraulic actuator. To restrict the system output in a prescribed performance constraint, a weighted performance function is designed to guarantee the dynamic and steady tracking errors of joint angle in a required accuracy. Then, a radial-basis-function NN is constructed to train the unknown model dynamics of a manipulator by traditional backstepping control (TBC) and obtain the preliminary estimated model, which can replace the preknown dynamics in the backstepping iteration. Furthermore, an adaptive estimation law is adopted to self-tune every trained-node weight, and the estimated model is online optimized to enhance the robustness of the NN controller. The effectiveness of the proposed control is verified by comparative simulation and experimental results with Proportional-integral-derivative and TBC methods.

When dealing with patients with psychological or emotional symptoms, medical practitioners are often faced with the problem of objectively recognizing their patients' emotional state. In this paper, we approach this problem using a computer program that automatically extracts emotions from EEG signals. We extend the finding of Koelstra et. al [IEEE trans. affective comput., vol. 3, no. 1, pp. 18-31, 2012] using the same dataset (i.e. the DEAP: dataset for emotion analysis using electroencephalogram, physiological and video signals), where we observed that the accuracy can be further improved using wavelet features extracted from shorter time segments. More precisely, we achieved accuracy of 65% for both valence and arousal using the wavelet entropy of 3 to 12 seconds signal segments. This improvement in accuracy entails an important discovery that information on emotions contained in the EEG signal may be better described in term of wavelets and in shorter time segments.

The first objective of this paper is to introduce a nonlinear system to model the heart rate (HR) response during and after treadmill walking exercise. The model is a feedback interconnected system that has components to describe the central and peripheral local responses to exercise and their interactions. The parameters of the model were experimentally identified from subjects walking on a treadmill at different speeds. The stability of the obtained nonlinear model was mathematically proven. The modeling results demonstrate that the proposed model can be useful in examining the cardiovascular response to exercise. Based on the nonlinear model, the second objective is to present a computer-controlled treadmill system for the regulation of HR during treadmill exercise. The proposed nonlinear controller consists of feedforward and feedback components. The designed control system was experimentally verified and the results demonstrated that the proposed computer-controlled treadmill system regulated the HR of the experimental subjects according to two different exercising HR profiles, indicating that it can play an important role in the design of exercise protocols for individuals.

This paper proposes a novel integrated approach for the identification and control of Hammerstein systems to achieve desired heart rate profile tracking performance for an automated treadmill system. For the identification of Hammerstein systems, the pseudorandom binary sequence input is employed to decouple the identification of dynamic linear part from input nonlinearity. The powerful epsiv-insensitivity support vector regression method is adopted to obtain sparse representations of the inverse of static nonlinearity in order to obtain an approximate linear model of the Hammerstein system. An H_infin controller is designed for the approximated linear model to achieve robust tracking performance. This new approach is successfully applied to the design of a computer-controlled treadmill system for the regulation of heart rate during treadmill exercise. Minimizing deviations of heart rate from a preset profile is achieved by controlling the speed of the treadmill. Both conventional proportional-integral-derivative (PID) control and the proposed approaches have been employed for the controller design. The proposed algorithm achieves much better heart rate tracking performance.

Recently, the smart grid has been considered as a next-generation power system to modernize the traditional grid to improve its security, connectivity, efficiency and sustainability. Unfortunately, the smart grid is susceptible to malicious cyber attacks, which can create serious technical, economical, social and control problems in power network operations. In contrast to the traditional cyber attack minimization techniques, this paper proposes a recursive systematic convolutional U+0028 RSC U+0029 code and Kalman filter U+0028 KF U+0029 based method in the context of smart grids. Specifically, the proposed RSC code is used to add redundancy in the microgrid states, and the log maximum a-posterior is used to recover the state information, which is affected by random noises and cyber attacks. Once the estimated states are obtained by KF algorithm, a semidefinite programming based optimal feedback controller is proposed to regulate the system states, so that the power system can operate properly. Test results show that the proposed approach can accurately mitigate the cyber attacks and properly estimate and control the system states.

This article investigates deep neural networks (DNNs)-based lung nodule classification with hyperparameter optimization. Hyperparameter optimization in DNNs is a computationally expensive problem, and a surrogate-assisted evolutionary algorithm has been recently introduced to automatically search for optimal hyperparameter configurations of DNNs, by applying computationally efficient surrogate models to approximate the validation error function of hyperparameter configurations. Different from existing surrogate models adopting stationary covariance functions (kernels) to measure the difference between hyperparameter points, this article proposes a nonstationary kernel that allows the surrogate model to adapt to functions whose smoothness varies with the spatial location of inputs. A multilevel convolutional neural network (ML-CNN) is built for lung nodule classification, and the hyperparameter configuration is optimized by the proposed nonstationary kernel-based Gaussian surrogate model. Our algorithm searches with a surrogate for optimal setting via a hyperparameter importance-based evolutionary strategy, and the experiments demonstrate our algorithm outperforms manual tuning and several well-established hyperparameter optimization methods, including random search, grid search, the tree-structured parzen estimator (TPE) approach, Gaussian processes (GP) with stationary kernels, and the recently proposed hyperparameter optimization via RBF and dynamic (HORD) coordinate search.