Methodology based on spiking neural networks for univariate time-series forecasting.

Spiking Neural Networks (SNN) are recognised as well-suited for processing spatiotemporal information with ultra-low energy consumption. However, proposals based on SNN for classification tasks are more common than for forecasting problems. In this sense, this paper presents a new general training methodology for univariate time-series forecasting based on SNN. The methodology is focused on one-step ahead forecasting problems and combines a PulseWidth Modulation based encoding-decoding algorithm with a Surrogate Gradient method as supervised training algorithm. In order to validate the generality of the presented methodology sine-wave, 3 UCI and 1 available real-world datasets are used. The results show very satisfactory forecasting results (MAE∈[0.0094,0.2891]) regardless of the characteristics of the dataset or the application field. In addition, weights can be initialised just once to achieve robust results, boosting the advantages of computational and energy cost of SNN.

Assessment of personal care product use and perceptions of use in a sample of US adults affiliated with a university in the Northeast.

Evidence supports unequal burdens of chemical exposures from personal care products (PCPs) among some groups, namely femme-identifying and racial and ethnic minorities. In this study, we implemented an online questionnaire to assess PCP purchasing and usage behaviors and perceptions of use among a sample of US adults recruited at a Northeastern university. We collected PCP use across seven product categories (hair, beauty, skincare, perfumes/colognes, feminine hygiene, oral care, other), and behaviors, attitudes, and perceptions of use and safety across sociodemographic factors to evaluate relationships between sociodemographic factors and the total number of products used within the prior 24-48 h using multivariable models. We also summarized participants' perceptions and attitudes. Among 591 adults (20.0% Asian American/Pacific Islander [AAPI], 5.9% Hispanic, 9.6% non-Hispanic Black [NHB], 54.6% non-Hispanic White [NHW], and 9.9% multiracial or other), the average number of PCPs used within the prior 24-48 h was 15.6 ± 7.7. PCP use was greater among females than males (19.0 vs. 7.9, P < 0.01) and varied by race and ethnicity among females. Relative to NHWs, AAPI females used fewer hair products (2.5 vs. 3.1) and more feminine hygiene products (1.5 vs. 1.1), NHB females used more hair products (3.8 vs. 3.1), perfumes (1.0 vs. 0.6), oral care (2.3 vs. 1.9), and feminine hygiene products (1.8 vs. 1.1), and multiracial or other females used more oral care (2.2 vs. 1.9) and feminine hygiene products (1.5 vs. 1.1) (P-values <0.05). Generally, study participants reported moderate concern about exposures and health effects from using PCPs, with few differences by gender, race, and ethnicity. These findings add to the extant literature on PCP use across sociodemographic characteristics. Improving the understanding of patterns of use for specific products and their chemical ingredients is critical for developing interventions to reduce these exposures, especially in vulnerable groups with an unequal burden of exposure.

Soft-Sensor for Class Prediction of the Percentage of Pentanes in Butane at a Debutanizer Column.

Refineries are complex industrial systems that transform crude oil into more valuable subproducts. Due to the advances in sensors, easily measurable variables are continuously monitored and several data-driven soft-sensors are proposed to control the distillation process and the quality of the resultant subproducts. However, data preprocessing and soft-sensor modelling are still complex and time-consuming tasks that are expected to be automatised in the context of Industry 4.0. Although recently several automated learning (autoML) approaches have been proposed, these rely on model configuration and hyper-parameters optimisation. This paper advances the state-of-the-art by proposing an autoML approach that selects, among different normalisation and feature weighting preprocessing techniques and various well-known Machine Learning (ML) algorithms, the best configuration to create a reliable soft-sensor for the problem at hand. As proven in this research, each normalisation method transforms a given dataset differently, which ultimately affects the ML algorithm performance. The presented autoML approach considers the features preprocessing importance, including it, and the algorithm selection and configuration, as a fundamental stage of the methodology. The proposed autoML approach is applied to real data from a refinery in the Basque Country to create a soft-sensor in order to complement the operators' decision-making that, based on the operational variables of a distillation process, detects 400 min in advance with 98.925% precision if the resultant product does not reach the quality standards.

Pulsewidth Modulation-Based Algorithm for Spike Phase Encoding and Decoding of Time-Dependent Analog Data.

This article proposes a new spike encoding and decoding algorithm for analog data. The algorithm uses the pulsewidth modulation principles to achieve a high reconstruction accuracy of the signal, along with a high level of data compression. Two benchmark data sets are used to illustrate the method: stock index time series and human voice data. Applications of the method for spiking neural network (SNN) modeling and neuromorphic implementations are discussed. The proposed method would allow the development of new applications of SNNs as regression techniques for predictive time-series modeling.

Instrumented Crutch Tip for Monitoring Force and Crutch Pitch Angle.

In rehabilitation procedures related to the lower limbs, gait monitoring is an important source of information for the therapist. However, many of the approaches proposed in the literature require the use of uncomfortable and invasive devices. In this work, an instrumented tip is developed and detailed, which can be connected to any crutch. The instrumented tip provides objective data of the crutch motion, which, combined with patient movement data, might be used to monitor the daily activities or assess the recovery status of the patient. For that purpose, the tip integrates a two-axis inclinometer, a tri-axial gyroscope, and a force sensor to measure the force exerted on the crutch. In addition, a novel algorithm to estimate the pitch angle of the crutch is developed. The proposed approach is tested experimentally, obtaining acceptable accuracies and demonstrating the validity of the proposed lightweight, portable solution for gait monitoring.

Virtual Sensors for Advanced Controllers in Rehabilitation Robotics.

In order to properly control rehabilitation robotic devices, the measurement of interaction force and motion between patient and robot is an essential part. Usually, however, this is a complex task that requires the use of accurate sensors which increase the cost and the complexity of the robotic device. In this work, we address the development of virtual sensors that can be used as an alternative of actual force and motion sensors for the Universal Haptic Pantograph (UHP) rehabilitation robot for upper limbs training. These virtual sensors estimate the force and motion at the contact point where the patient interacts with the robot using the mathematical model of the robotic device and measurement through low cost position sensors. To demonstrate the performance of the proposed virtual sensors, they have been implemented in an advanced position/force controller of the UHP rehabilitation robot and experimentally evaluated. The experimental results reveal that the controller based on the virtual sensors has similar performance to the one using direct measurement (less than 0.005 m and 1.5 N difference in mean error). Hence, the developed virtual sensors to estimate interaction force and motion can be adopted to replace actual precise but normally high-priced sensors which are fundamental components for advanced control of rehabilitation robotic devices.

Prediction of performance by heart rate-derived parameters in recreational runners.

We investigated whether heart rate (HR)-derived parameters are accurate performance predictors in endurance recreational runners. One hundred thirty recreational athletes completed an incremental running test (4´running + 1´rest). After each stage, we recorded HR, % of maximum HR (%HRmax), and blood lactate. We also assessed HR after each recovery period, and calculated lactate and HR recovery thresholds and HR deflection point. We tested these parameters for associations with running performance, as measured by peak treadmill speed (PTS) and personal best International Association of Athletics Federations (IAAF) score. The %HRmax at 14.5 km·h-1 correlated with PTS (r = -0.92), and IAAF score (rho = -0.80). The magnitudes of the correlations of lactate-related parameters with PTS (|r| = 0.84 to 0.86) or IAAF score (|rho| = 0.70 to 0.77) in absolute values were slightly lower. The correlations detected between other HR-derived parameters and running performance were weaker (|r or rho| = 0.24 to 0.70). Regression models identified %HRmax at 14.5 km·h-1 as the strongest predictor of both PTS (ß = -0.72) and IAAF score (ß = -0.72). Consequently, tests based on %HRmax may provide a non-invasive and inexpensive alternate method for predicting the performance of these athletes.

Virtual sensors for on-line wheel wear and part roughness measurement in the grinding process.

Grinding is an advanced machining process for the manufacturing of valuable complex and accurate parts for high added value sectors such as aerospace, wind generation, etc. Due to the extremely severe conditions inside grinding machines, critical process variables such as part surface finish or grinding wheel wear cannot be easily and cheaply measured on-line. In this paper a virtual sensor for on-line monitoring of those variables is presented. The sensor is based on the modelling ability of Artificial Neural Networks (ANNs) for stochastic and non-linear processes such as grinding; the selected architecture is the Layer-Recurrent neural network. The sensor makes use of the relation between the variables to be measured and power consumption in the wheel spindle, which can be easily measured. A sensor calibration methodology is presented, and the levels of error that can be expected are discussed. Validation of the new sensor is carried out by comparing the sensor's results with actual measurements carried out in an industrial grinding machine. Results show excellent estimation performance for both wheel wear and surface roughness. In the case of wheel wear, the absolute error is within the range of microns (average value 32 µm). In the case of surface finish, the absolute error is well below Ra 1 µm (average value 0.32 µm). The present approach can be easily generalized to other grinding operations.