Eloy Irigoyen

A NARX neural network model for enhancing cardiovascular rehabilitation therapies

Current medical tendencies in the rehabilitation field are trying to physically rehabilitate patients. Thus, people with cardiovascular illnesses need to exercise their injured systems in order to improve themselves. In training, each person has a different heart rate response according to the demand of physical effort. Hence, it is necessary to know the relationship between the effort (training device power/resistance) and the patients heartbeat for an optimal training configuration. This relationship has non-linear and complex dynamics, being a complicated identification problem solved by classical techniques. Soft Computing techniques based on artificial neural networks may be a way to implement more efficient control strategies in order to obtain a suitable power demand each and every time. It is necessary to be aware of the pace, length and intensity of the exercises in order to be effective and safe. In this paper, we present the results of the identification of the relationship in time, between the required exercise (machine resistance) and the heart rate of the patient in medical effort tests, using a NARX neural network model. In the experimental stage, test data have been obtained by exercising with a cyclo-ergometer in two different tests: Power Step Response (PSR) and Conconi.

Microcontroller Implementation of a Multi Objective Genetic Algorithm for Real-Time Intelligent Control

This paper presents an approach to merge three elements that are usually not thought to be combined in one application: evolutionary computing running on reasonably priced microcontrollers (μC) for real-time fast control systems. A Multi Objective Genetic Algorithm (MOGA) is implemented on a 180MHz μC.A fourth element, a Neural Network (NN) for supporting the evaluation function by predicting the response of the controlled system, is also implemented. Computational performance and the influence of a variety of factors are discussed. The results open a whole new spectrum of applications with great potential to benefit from multivariable and multiobjective intelligent control methods in which the hybridization of different soft-computing techniques could be present. The main contribution of this paper is to prove that advanced soft-computing techniques are a feasible solution to be implemented on reasonably priced μC -based embedded platforms.

Numerical bounds to assure initial local stability of NARX multilayer perceptrons and radial basis functions

In this work, local stability on the initialization phase of nonlinear autoregressive with exogenous inputs multilayer perceptrons (NARX MLP) and radial basis functions (NARX RBF) neural networks is studied. It will be shown that the selection of adequate ranges for the initial weights is related with local stability of the network in its initial stage. As a result, quantitative limits for the initial weights are established that guarantee local stability and accelerate the learning process. These theoretical developments have been tested in experiments which corroborate the improvements achieved with the proposed initialization methods.

Enhancements for a Robust Fuzzy Detection of Stress

Improving psychologically disabled people’s life quality and integration in society is strongly linked with providing them higher levels of autonomy. Occasionally, these people suffer from emotional blockages produced by situations that can be overwhelming for them. Thus, detecting whether the person is entering a mental blockage produced by stress can facilitate to mitigate the symptoms of that blockage. This work presents different enhancements and variations for an existing fuzzy logic stress detection system based on monitoring different physiological signals (heart rate and galvanic skin response). It proposes a method based on wavelet processing to improve the detection of R peaks of electrocardiograms. It also proposes to decompose the galvanic response signal into two components: the average value and the variations.

Tutor Project: An Intelligent Tutoring System to Improve Cognitive Disabled People Integration

Nowadays, the integration of people with cognitive disabilities, especially in the work environment in a growing competitive market, is a difficult task. The TUTOR project is addressed to the development and testing of an Intelligent Tutoring System (ITS). The ITS runs on a handheld device and its aim is to increase the autonomy of people with cognitive disabilities in labour and daily life activities, therefore, improving the social and labour integration of this collective and its quality of live. This paper describes the objectives of this project, the methodology followed, some preliminary achieved results and the future planned activities of this research group.

Real time direct kinematic problem computation of the 3PRS robot using neural networks

Abstract The reliable calculation of the Direct Kinematic Problem (DKP) is one of the main challenges for the implementation of Real-Time (RT) controllers in Parallel Robots. The DKP estimates the pose of the end effector of the robot in terms of the sensors placed on the actuators. However, this calculation requires the use of time-consuming numerical iterative procedures. Artificial Neural Networks have been proposed to implement the complex DKP equation mapping due to their universal approximator property. However, the proposals in this area do not consider the Real Time implementation of the ANN based solution, and no approximation error vs computational time analysis is carried out. In this work, a methodology that uses Artificial Neural Networks (ANNs) to approximate the DKP is proposed. Based on the 3PRS parallel robot, a comprehensive study is carried out in which several network configurations are proposed to approximate the DKP. Moreover, to demonstrate the effectiveness of the approach, the proposed networks are evaluated considering not only their approximation capabilities, but also their Real Time performance in comparison with the traditional iterative procedures used in robotics.

A real-time stress classification system based on arousal analysis of the nervous system by an F-state machine

BACKGROUND AND OBJECTIVE Detection and labelling of an increment in the human stress level is a contribution focused principally on improving the quality of life of people. This work is aimed to develop a biophysical real-time stress identification and classification system, analysing two noninvasive signals, the galvanic skin response and the heart rate variability. METHODS An experimental procedure was designed and configured in order to elicit a stressful situation that is similar to those found in real cases. A total of 166 subjects participated in this experimental stage. The set of registered signals of each subject was considered as one experiment. A preliminary qualitative analysis of the signals collected was made, based on previous counselling received from neurophysiologists and psychologists. This study revealed a relationship between changes in the temporal signals and the induced stress states in each subject. To identify and classify such states, a subsequent quantitative analysis was performed in order to determine specific numerical information related to the above mentioned relationship. This second analysis gives the particular details to design the finally proposed classification algorithm, based on a Finite State Machine. RESULTS The proposed system is able to classify the detected stress stages at three levels: low, medium, and high. Furthermore, the system identifies persistent stress situations or momentary alerts, depending on the subjects arousal. The system reaches an F1 score of 0.984 in the case of high level, an F1 score of 0.970 for medium level, and an F1 score of 0.943 for low level. CONCLUSION The resulting system is able to detect and classify different stress stages only based on two non invasive signals. These signals can be collected in people during their monitoring and be processed in a real-time sense, as the system can be previously preconfigured. Therefore, it could easily be implemented in a wearable prototype that could be worn by end users without feeling to be monitored. Besides, due to its low computational, the computation of the signals slopes is easy to do and its deployment in real-time applications is feasible.

Artificial neural network modelling of the bioethanol-to-olefins process on a HZSM-5 catalyst treated with alkali

Abstract In this work the kinetic modelling of the transformation of bioethanol-to-olefins (BTO) process over a HZSM-5 catalyst treated with alkali using artificial neural networks (ANN) is presented. The main goal has been to obtain a BTO process neuronal model with the desired accuracy that allows the simplification and reduction of the computational cost with respect to a mechanistic knowledge model. To check the goodness of ANN base model structures, during the study a comparison with other alternative modelling techniques such as support vector machines was performed. Following a parameters optimization procedure and testing different training methods, the optimal ANN structure results to be a feed-forward 3–5–1 network with the Bayesian regularization training method. Using a set of experimental data obtained in a laboratory scale fixed bed reactor, we have obtained a similar fit to the knowledge model but with the advantage of being up to 43 times faster. These results are important for moving forward real time automatic control strategies in the biorefinery context.

Neuro-fuzzy models for hand movements induced by functional electrical stimulation in able-bodied and hemiplegic subjects

Functional Electrical Stimulation (FES) may be effective as a therapeutic treatment for improving functional reaching and grasping. Upper-limb FES models for predicting joint torques/angles from stimulation parameters can be useful to support the iterative design and development of neuroprostheses. Most such models focused on shoulder or elbow joints and were defined for fixed electrode configurations. This work proposes the use of a Recurrent Fuzzy Neural Network (RFNN) for modeling FES induced wrist, thumb, and finger movements based on surface multi-field electrodes and kinematic data from able-bodied and neurologically impaired subjects. Different combinations of structure parameters comprising fuzzy term numbers and feedback approaches were tested and analyzed in order to see their effect on the model performance for six subjects. The results showed mean success rates in the range from 60% to 99% and best success rates in the range from 78% to 100% on test data for all subjects. No common trend was found across subjects regarding structure parameters. The model showed the ability to successfully reproduce the response to FES for both able-bodied and hemiplegic subjects at least with one of the tested combinations.

Implementation and testing of a soft computing based model predictive control on an industrial controller

This work presents a real time testing approach of an Intelligent Multiobjective Nonlinear-Model Predictive Control Strategy (iMO-NMPC). The goal is the testing and analysis of the feasibility and reliability of some Soft Computing (SC) techniques running on a real time industrial controller. In this predictive control strategy, a Multiobjective Genetic Algorithm is used together with a Recurrent Artificial Neural Network in order to obtain the control action at each sampling time. The entire development process, from the numeric simulation of the control scheme to its implementation and testing on a PC-based industrial controller, is also presented in this paper. The computational time requirements are discussed as well. The obtained results show that the SC techniques can be considered also to tackle highly nonlinear and coupled complex control problems in real time, thus optimising and enhancing the response of the control loop. Therefore this work is a contribution to spread the SC techniques in on-line control applications, where currently they are relegated mainly to be used off-line, as is the case of optimal tuning of control strategies.