Search

A multilayer perceptron (MLP) artificial neural network (ANN) model has been optimized by the multi-objective ant colony optimization (MOACO) algorithm, which uses three objective functions. A sensitivity analysis to choose MOACO parameter values is carried out by calculating hypervolume metric, and the proposed approach adopts the Vlsekriterijumska Optimizacija I Kompromisno Resenje (VIKOR) decision method to choose final compromised solution on the Pareto front obtained from MOACO. As a result, we used the MLP-MOACO developed model to estimate the value of engine emissions of NO x in a four stroke, spark ignition (SI) gasoline engine and observed acceptable correlation coefficient ( R 2) of 0.99978.

Cucurbits plants are very susceptible to fungal diseases as powdery mildew (PM) infection. Currently, the PM early detection in the field is the cutting-edge research. The objectives of our study were to assess visible and near-infrared spectroscopy of normal and infected leaves for early detection of PM in Cucurbita pepo L. plants. Samples for spectral analysis were taken three days a week (Monday, Wednesday, and Friday) after the first true leaves appeared. The reflectance spectra of leaves were collected using an HR4000CG-UV-NIR spectrometer (Ocean Optics) with a fiber optics probe. Vegetative indices were used for discrimination between infected and healthy plants. The calculated vegetation indices (green normalized difference vegetation index, triangular greenness index, single-band reflectance index, simple ratio indices, and anthocyanins reflectance index) showed the highest sensitivity to differentiate healthy, infected plants at a symptomless stage, first symptoms, and diseased plants. The best prediction on early detection in sampling days were structure-independent pigment index and red-edge reflection point with an R of 0.211 and 0.1893, respectively. Our study shows the efficacy of the identification based on reflectance spectra for an early distinction of PM disease and could be used in cucurbits plants with similar characteristics on leaves.

In photovoltaic systems, it is necessary to have DCDC converters topologies with wide transformation ranges that have the ability to provide output voltages that can be greater or less than the supply voltage by the PV modules. Other aspects that these converters must satisfy are associated to the new requirements of the load consumption. These requirements have been standardized to 48V, 24, 12V and 5V for different applications such as LED lighting, battery banks, amongst others. Also, one important requirement in these systems is that the power processing by the converter must be carried out as efficiently as possible. This paper presents an implemmentation of a quadratic buck converter based on the concept R2P2 as an alternative for photovoltaic applications, where the characteristics of the input current demanded to the panels and the reduction of EMI are improved. In this work, the description of the system formed by the quadratic buck converter R2P2 with an input filter is analyzed, where the requirements of non-pulsating input current and output voltage regulation are satisfied. In addition, the design procedure of the elements that form the converter is developed, as well as the description of the operating modes during the switching process of the converter. Finally, the linear and non-linear models of the converter are presented, as well as some comparative aspects against proposed topologies and the experimental verification.

This work presents a fault diagnosis strategy for induction motors based on multi-class classification through support vector machines (SVM), and the so-called one-against-one method. The proposed approach classifies four different motor conditions (healthy, misalignment, unbalanced rotor and bearing damage) at variable operating conditions (supply frequency and load torque). The proposed SVMs use signatures from the frequency domain characteristics related to each studied fault. These signatures combine information just from the stator condition: radial vibration and stator currents. To acquire training and validation data in steady state, different experiments were performed using a three-phase induction motor. Thirty-five data sets were obtained at different operating regimes of the induction motor for each specific fault (140 conditions including a no-fault scenario) to validate our study. The SVMs with a Gaussian radial basis function (RBF) were proposed as a kernel for the nonlinear classification process. To select the parameter value of the RBF, a bootstrap technique was used. The resulting accuracy for the fault classification process was on the range 84.8–100%.

In this work, the scaling behavior of the sleep process is evaluated by using detrended fluctuation analysis based on wavelets. The analysis is carried out from arrivals of short and recurrent cortical events called A-phases, which in turn build up the Cyclic Alternating Pattern phenomenon, and are classified in three types: A1, A2 and A3. In this study, 61 sleep recordings corresponding to healthy, nocturnal frontal lobe epilepsy patients and sleep-state misperception subjects, were analyzed. From the A-phase annotations, the onsets were extracted and a binary sequence with one second resolution was generated. An item in the sequence has a value of one if an A-phase onset occurs in the corresponding window, and a value of zero otherwise. In addition, we consider other different temporal resolutions from 2[Formula: see text]s to 256[Formula: see text]s. Furthermore, the same analysis was carried out for sequences obtained from the different types of A-phases and their combinations. The results of the numerical analysis showed a relationship between the time resolutions and the scaling exponents; specifically, for higher time resolutions a white noise behavior is observed, whereas for lower time resolutions a behavior towards to [Formula: see text]-noise is exhibited. Statistical differences among groups were observed by applying various wavelet functions from the Daubechies family and choosing the appropriate sequence of A-phase onsets. This scaling analysis allows the characterization of the free-scale dynamic of the sleep process that is specific for each sleep condition. The scaling exponent could be useful as a diagnosis parameter in clinics when sleep macrostructure does not offer enough information.

In this paper is presented an interfacing of a robotic arm with a motion planner and a 3D viewer (RViz). The main purpose of this work, is to manipulate and grip objects to transport them from one location to another using Robot Operating System (ROS). In order to do this task, a motion planner, who deal with constraints, i.e. collisions among the arm joints and objects, is used. Also a 3D viewer, called RViz, is used to visualize (online and offline) the motion of the arm. To easily interface the 3 degree of freedom (DOF) arm, it was designed using FreeCAD, as the output files used to print the arm, are simultaneously exported to create the Unified Robot Description Format (URDF) files required for the planner and viewer. A position-force controller for the gripper, previously presented, has been used to pick up and hold the object during the manipulation. Some experiments where carried out to asses the result of the interface, the execution of the motion planning avoiding collision and the grasp to the object. These results are shown and discussed.

This paper presents a study of the complementarity of wind and solar resources over Mexico for electricity generation. A dataset was generated using high-resolution wind and solar resource maps provided by the SIGER project to assess the possibility of a combined used of these renewable energy sources. Using image processing techniques, the values of average monthly wind speed and solar irradiation were obtained, and then the average energy production was calculated. In order to obtain new maps that show the degree of correlation between the wind and solar resources the Pearson correlation coefficient was applied. The obtained results show a good degree of complementarity between both energy sources, in specific places of the country in the different seasons of the year. The obtained maps can be useful in projects related to the installation of hybrid renewable energy generation systems that have wind turbines and photovoltaic cells.

Automatic sleep staging based on inter-beat fluctuations and motion signals recorded through a pressure bed sensor during sleep is presented. The analysis of the sleep was based on the three major divisions of the sleep time: Wake, non-rapid eye movement (nREM) and rapid eye movement (REM) sleep stages. Twelve sleep recordings, from six females working alternate shift, with their respective annotations were used in the study. Six recordings were acquired during the night and six during the day after a night shift. A Time-Variant Autoregressive Model was used to extract features from inter-beat fluctuations which later were fed to a Support Vector Machine classifier. Accuracy, Kappa index, and percentage in wake, REM and nREM were used as performance measures. Comparison between the automatic sleep staging detection and the standard clinical annotations, shows mean values of [Formula: see text]% for accuracy [Formula: see text] for kappa index, and mean errors of 5% for sleep stages. The performance measures were similar for night and day sleep recordings. In this sample of recordings, the results suggest that inter-beat fluctuations and motions acquired in non-obtrusive way carried valuable information related to the sleep macrostructure and could be used to support to the experts in extensive evaluation and monitoring of sleep.

Mobile Robot Manipulation of diverse objects is still a very difficult task. Progress in mobile robot localization, path planning and perception, have allowed to mobile robots to contend with manipulation in some domains. Parallels manipulators are commonly used, however in order to do robust grasping, as humans do, the use of independent fingers are increasing. The problem of independent fingers is to assure robust manipulation on behalf the variations between the position and control of the fingers. In this work it is presented a system for object manipulation by a robotic gripper with independent fingers (with angular motion) each of them equipped with a position/force sensor. The system deal with the problem of adjusting the position of the fingers controlled independently over the middleware ROS, in order to guaranty that object is at center position on the gripper. Once it is done, the system switch to a force controller apply the correct force on each independent finger. The proposed system uses two proportional integral derivative (PID) controller; one to control the position of the object, and the second one to control the force applied on the object. The experiments are carried out to asses, if the object is centered and lifted by the gripper without releasing. Results obtained are presented and discussed.

This paper proposes a hybrid learning of artificial neural network (ANN) with the nondominated sorting genetic algorithm-II (NSGAII) to improve accuracy in order to predict the exhaust emissions of a four stroke spark ignition (SI) engine. In the proposed approach, the genetic algorithm (GA) determines initial weights of local linear model tree (LOLIMOT) neural networks. A multi-objective optimization problem is determined. A sensitivity analysis is performed on NSGA-II parameters in order to provide better solutions along the optimal Pareto front. Then, a fuzzy decision maker and the technique for order preference by similarity to ideal solution (TOPSIS) are employed to select compromised solutions among the obtained Pareto solutions. The LOLIMOT-GA responses are compared with the provided by radial basis function (RBF) and multilayer perceptron (MLP) neural networks in terms of correlation coefficient R 2.

A criteria for selection of parameters of proportional and integral (PI) controller for a DC–DC quadratic buck converter is proposed here. This procedure was developed from a large-signal model of converter in which a multiloop control scheme is added. This scheme has two control loops: an inner current loop, which improves the dynamic behaviour using a single gain, and a conventional PI controller as outer loop, which provides regulation. For the resulting system, the necessary conditions are derived for the existence of equilibrium point and stability in closed-loop. In comparison with the traditional method of loopshaping, which uses small-signal model, analytic expressions are found that simplifies the selection of controller parameters. At the end, the procedure is applied to 28 W quadratic buck regulator prototype and experimental results are given to verify the effectiveness.

In this paper, a multi-objective particle swarm optimization (MOPSO) algorithm and a nondominated sorting genetic algorithm II (NSGA-II) are used to optimize the operating parameters of a 1.6 L, spark ignition (SI) gasoline engine. The aim of this optimization is to reduce engine emissions in terms of carbon monoxide (CO), hydrocarbons (HC), and nitrogen oxides (NOx), which are the causes of diverse environmental problems such as air pollution and global warming. Stationary engine tests were performed for data generation, covering 60 operating conditions. Artificial neural networks (ANNs) were used to predict exhaust emissions, whose inputs were from six engine operating parameters, and the outputs were three resulting exhaust emissions. The outputs of ANNs were used to evaluate objective functions within the optimization algorithms: NSGA-II and MOPSO. Then a decision-making process was conducted, using a fuzzy method to select a Pareto solution with which the best emission reductions can be achieved. The NSGA-II algorithm achieved reductions of at least 9.84%, 82.44%, and 13.78% for CO, HC, and NOx, respectively. With a MOPSO algorithm the reached reductions were at least 13.68%, 83.80%, and 7.67% for CO, HC, and NOx, respectively.

Today, the manipulation of objects by mobile robots is still a challenging task. This task is commonly decomposed on three stages: a) approaching to the objects, b) path planning and trajectory execution of the manipulator arm and finally c) fine tuning and grasping. In this work is presented an implementation of a 3D pose visual servoing for an autonomous mobile manipulator dealing with the last stage of the manipulation task (fine tuning and grasping). The methodology proposed consists of three steps: a) beginning with a fast monocular image segmentation, followed by b) 3D model reconstruction and finally c) pose estimation to feedback fine tuning manipulation control loop. Objects and end-effector are colored in different colors and their models are supposed to be known. Our mobile manipulator prototype consists of a stereo camera under a binocular stand-alone configuration and an anthropomorphic 7DoF arm with a parallel end-effector (gripper). Our methodology runs in real time and is suitable to perform continuous visual servoing. Experimental results are reported.

This paper presents a novel solution for the classical leader-follower formation problem considering the case of nonholonomic mobile robots. A formation control strategy is proposed in a discrete-time context by considering the exact discrete-time discretization of the non-linear continuous-time kinematic model of the vehicle. The geometric formation of the robots allows us to derive an alternative model that describes the time evolution of the relative distance and angle between the robots. These variables are obtained in real-time by a vision-based localization system on board, in which the follower robot is equipped with a Kinect device, together with a recognition board mounted on the leader robot. The boundedness of the relative position error is formally proven by considering a feedback law that is delayed by one sampling period of time. Numerical simulations and real-time experiments are presented to verify the performance of the control strategy.

In this article, we propose a software architecture for controlling a mobile robot that works as a guide on human environments. The architecture is divided into three broad action modules lead by a supervisor. The decision is based on current user behavior in order to provide better service to people.

In this paper, are presented the characterization of two force sensors, in bias to incorporate them to the fingers of a robotic hand. The main goal of this work is to determine as main characteristic, the response of each sensor, in order to decide, which of them is the most effective for manipulation tasks in a mobile robotic platform. Manipulation mobile robot tasks, require to adapt its behavior and fingers configuration depending of the type and form of diverse objects. The first evaluated sensor is a cheap resistive force sensor and second is a piezoelectric force sensor. As robotic hand and then sensors will be incorporate on a mobile platform, sensor signals are sent to the a computer through a data card acquisition; an Arduino card has been used. We show the evaluation, results and conclusions of the experiments done.