Search

XGBoost is a highly effective and widely used machine learning model and its hyperparameters take an important role on the performance of the model. This paper presents a new grid search (GS) algorithm for obtaining optimal hyperparameters of the XGBoost model based on the median values of their error loss. A benchmark method used to evaluate the proposed and original GS algorithms is introduced. Datasets with measured daily electricity demand load values of Ho Chi Minh City, Vietnam and Tasmania state, Australia are analyzed for the performance of both algorithms. The error metrics, mean squared errors (MSEs), of the proposed algorithm are found to be 2,282 MW and 501 MW that are smaller than those of original algorithms, which are 2,424 MW and 537 MW in case of Ho Chi Minh City and Tasmania state, respectively. These results then verify the accuracy of the proposed algorithm.

In the past decades, the emerging concern about food safety has led to the increasing demand for monitoring food quality across the world. Aiming towards a novel solution for monitoring food, this study proposes a non-destructive method with self-powering capability for online food monitoring, which can be extendable to IoT applications. Furthermore, the study introduces a novel deep neural network model to predict different states of food quality based on the monitoring results. To monitor the variation in food quality, the paper proposes the detection of total volatile organic compounds (TVOCs) inside the food packages, which have been released during food deterioration. A low-power sensor mote comprised of a capacity humidity sensor and a metal-oxide (MOX) gas sensor was manufactured for this purpose. The self-powering capability of the mote is provided through an energy harvester module, which benefits from the far-field Radio Frequency Energy Harvesting (RFEH) technology. The operating frequency of the module was chosen at the 915-MHz ISM band. The analysis of the harvester performance showed that the harvester could generate 3.3-V dc with an RF input power of as low as -8 dBm, which was sufficient for the mote operation. To verify the proposed solutions, a demonstration to monitor the deterioration of packaged pork and fish was conducted in eight days under ambient and refrigerated storage conditions, using the self-developed RF-powered sensor mote. The raw variations in TVOCs were analyzed to evaluate the reliability of the proposed TVOC-based method. A one-dimensional (1-D) convolutional neural network (CNN) model was trained on the TVOCs dataset to predict different states of food quality. To investigate the applicability of the proposed 1-D CNN to multi-class determination of food quality, two other supervised machine learning algorithms using 2-D inputs, including Multilayer Perceptron (MLP), and Support Vector Machine (SVM), are studied. Their classification accuracies based on the confusion matrix are identified and compared.

Most radio frequency (RF) energy harvesters exhibit extremely nonlinear behavior during the harvesting phase, which renders the simulation of these circuits most challenging in design. This paper proposes a method of efficient modeling this nonlinearity based on empirically achieved characteristics of RF energy harvesting. A model using a multi-layer artificial neural network (ANN) is introduced to approximate this behavior. A harvester prototype that uses the far-field RF energy harvesting technique at 915 MHz is developed to verify the proposed model. The results highlight good agreement between the predicted and the observed outcomes of the harvester. With its flexibility, the proposed model can be further applied to characterize any nonlinear device. In addition, a self-powered system embedded with a fully-passive sensor device to monitor food storage profile is presented in the paper. An analysis of the power harvesting performance shows that the device can self-start its operation where the RF power is as low as −10.5 dBm, and the maximum achievable power conversion efficiency is 43%. To demonstrate the applicability of the proposed self-powered food monitoring system, an experiment is carried out using pork. Support vector machine (SVM) classifiers with two types of features are investigated to compare their performance of classifying three levels of food condition. As a result, the best classification accuracy of 95.7% is achieved.