Load Classification Research Articles

An interval prediction method for quantifying the uncertainties of cooling load based on time classification

The uncertainties of cooling load significantly impact the operations performance of building energy systems. A novel interval prediction method is proposed to quantify the uncertainties of cooling load. Due to the cooling load having a similar changing trend from 1:00 to 24:00, a cooling load classification method based on the time of the day is proposed to analyze the uncertainty. Firstly, the gated recurrent unit algorithm is used to construct the point prediction model. Secondly, the kernel density estimation method is employed for probability statistics of prediction error. Moreover, the percentile method constructs the upper and lower bounds of the cooling load prediction interval with given confidence levels. Finally, a commonly used interval prediction method is compared to the proposed method, and a real cooling load dataset is used to assess the prediction method. The results demonstrate that the point prediction model has a mean absolute percentage error of 10.19% in the interval prediction dataset. The interval prediction results illustrate that the prediction interval coverage probability is 91.75%, and the prediction interval normalized average width is 20.83% in the confidence level of 90%. Compared with the commonly used method, the proposed method has a narrower interval.

A comparison of machine learning models for suspended sediment load classification

The suspended sediment load (SSL) is one of the major hydrological processes affecting the sustainability of river planning and management. Moreover, sediments have a significant impact on dam operation and reservoir capacity. To this end, reliable and applicable models are required to compute and classify the SSL in rivers. The application of machine learning models has become common to solve complex problems such as SSL modeling. The present research investigated the ability of several models to classify the SSL data. This investigation aims to explore a new version of machine learning classifiers for SSL classification at Johor River, Malaysia. Extreme gradient boosting, random forest, support vector machine, multi-layer perceptron and k-nearest neighbors classifiers have been used to classify the SSL data. The sediment values are divided into multiple discrete ranges, where each range can be considered as one category or class. This study illustrates two different scenarios related to the number of categories, which are five and 10 categories, with two time scales, daily and weekly. The performance of the proposed models was evaluated by several statistical indicators. Overall, the proposed models achieved excellent classification of the SSL data under various scenarios.

Classification of Driver Cognitive Load: Exploring the Benefits of Fusing Eye-Tracking and Physiological Measures

In-vehicle infotainment systems can increase cognitive load and impair driving performance. These effects can be alleviated through interfaces that can assess cognitive load and adapt accordingly. Eye-tracking and physiological measures that are sensitive to cognitive load, such as pupil diameter, gaze dispersion, heart rate (HR), and galvanic skin response (GSR), can enable cognitive load estimation. The advancement in cost-effective and nonintrusive sensors in wearable devices provides an opportunity to enhance driver state detection by fusing eye-tracking and physiological measures. As a preliminary investigation of the added benefits of utilizing physiological data along with eye-tracking data in driver cognitive load detection, this paper explores the performance of several machine learning models in classifying three levels of cognitive load imposed on 33 drivers in a driving simulator study: no external load, lower difficulty 1-back task, and higher difficulty 2-back task. We built five machine learning models, including k-nearest neighbor, support vector machine, feedforward neural network, recurrent neural network, and random forest (RF) on (1) eye-tracking data only, (2) HR and GSR, (3) eye-tracking and HR, (4) eye-tracking and GSR, and (5) eye-tracking, HR, and GSR. Although physiological data provided 1%–15% lower classification accuracies compared with eye-tracking data, adding physiological data to eye-tracking data increased model accuracies, with an RF classifier achieving 97.8% accuracy. GSR led to a larger boost in accuracy (29.3%) over HR (17.9%), with the combination of the two factors boosting accuracy by 34.5%. Overall, utilizing both physiological and eye-tracking measures shows promise for driver state detection applications.

Radial variations of broad-sense heritability in wood properties and classification of load\u2013deflection curves in static bending for six half-sib families of Chamaecyparis obtusa

Wood properties (annual ring width, tracheid length, microfibril angle [MFA], basic density, and air-dry density) and mechanical properties (modulus of elasticity [MOE], modulus of rupture [MOR], bending work, and compressive strength) in 34-year-old Chamaecyparis obtusa trees of six half-sib families were measured from pith to bark to clarify radial variations in inheritance of these traits and the relationships between wood properties and mechanical properties. In addition, within-tree and among-family differences in the load–deflection curves were discussed. Radial variations of all wood properties were fitted to linear or nonlinear mixed-effects models with random effects of families. The MFA was correlated with MOE in all radial positions, whereas air-dry density correlated with all mechanical properties in mature wood. Radial variations in broad-sense heritability differed between wood properties. A relatively higher broad-sense heritability was recognized in almost all wood properties for mature wood. Based on the results, it was concluded that mechanical properties in mature wood can be effectively improved using MFA and air-dry density as criteria. In addition, the types of load–deflection curve in mature wood differed from those in juvenile wood, suggesting that not only elastic properties, but also plastic properties in C. obtusa are affected by genetic controls, especially in mature wood.

Joint Feature Selection of Power Load in Time Domain and Frequency Domain Based on Whale Optimization Algorithm

Accurate classification of power load is the premise of demand-side response capability evaluation, which is of great significance for demand-side management. Therefore, this study proposes a selection method of joint characteristics of power load in time domain and frequency domain based on whale optimization algorithm. The electrical measurement data of six typical electrical equipment are selected to form the original power load identification dataset. Firstly, the time domain features and frequency domain features of the power load data are extracted, and then the joint characteristics of the power load are screened by Whale optimization algorithm (WOA). Finally, the selected feature information is used as the input to verify the performance of WOA on the joint characteristics of power load under back propagation neural network (BP), extreme learning machine (ELM), support vector machine (SVM), k-nearest neighbor (KNN), decision tree (DT), and naive Bayesian (NB) classifiers. The results show that WOA can effectively screen out the 15 most helpful feature attributes for power load identification, which can not only improve the accuracy of power load identification but also effectively reduce the time cost of the algorithm, which is of reference value for further demand-side response strategy. At the same time, it is of great significance for intelligent dispatching of power system and improving the economic operation of the power system.

Experimental Investigation of Energy Efficient Room Air Conditioners for Mini-Stores

An experimental investigation of energy efficiency room air conditioners for the mini-stores was presented and illustrated in the paper. The experiment has been carried out in two well-insulated rooms with two different compressor technologies (fixed-speed and inverter) and essential equipment in a laboratory located at the School of Heat Engineering and Refrigeration (Hanoi University of Science and Technology). The authors conducted the test with three different heat load categories and weather conditions to examine the energy efficiency based on several factors such as room temperature and power consumption of air conditioner from records. Results show the significant effect of various aspects on the efficiency and the behaviour of air-cooled air conditioners including the indoor heat load and the outdoor temperature. The results also suggest a suitable cooling capacity choice that helps improve the energy efficiency of air conditioners besides the technical data such as CSPF (or SEER) and power consumption.

Towards a sustainable monitoring: A self-powered smart transportation infrastructure skin

Sustainable monitoring of traffic using clean energy supply has always been a significant problem for engineers. In this study, we proposed a self-powered smart transportation infrastructure skin (SSTIS) as an innovative and bionic system for the traffic classification of a smart city. This system incorporated the self-powered flexible sensors with net-zero power consumption based on the Triboelectric Nanogenerator (TENG) and an intelligent analysis system based on artificial intelligence (AI). The feasibility of the SSTIS was tested using the full-scale accelerated pavement tests (APT) and the long-short term memory (LSTM) deep learning model with a vehicle axle load classification accuracy up to 89.06%. This robust SSTIS was later tested on highway and collected around 869,600 pieces of signals data. The generative adversarial networks (GAN) WGAN-GP (Wasserstein GAN - Gradient Penalty) was used for data augmentation, due to the imbalanced data of different vehicle types in actual traffic. The overall accuracy for on-road vehicle type classification improved to 81.06% using the convolutional neural network ResNet. Finally, we developed a mobile traffic signal information monitoring system based on cloud platform and Android framework, which enabled engineers to obtain the vehicle axle-load information mobilely. This study is the emerging design and engineering application of the self-powered flexible sensors for smart traffic monitoring, which provides a significant advance for intelligent transportation and smart cities in future.

Classification of load and rotational speed at wire-race bearings using Convolutional Neural Networks with vibration spectrograms

Abstract We propose a model to classify axial, radial loading profiles and rotation speed of wire-race bearings using Convolutional Neural Networks with spectrograms. Software-based models are able to measure quantities of machine elements. In many cases a physical sensor takes up a lot of space and is expensive. Wire-race bearings are lightweight bearings that can be installed in a space-saving manner. For predictive maintenance of the load-dependent lifetime estimation there is the need to record the stress conditions on the bearing. While force sensors have the ability to record these necessary stresses, they cannot be installed space-savingly. With the proposed setup using an accelerometer attached to the housing of a wire-race bearing, it is possible to both save space and reduce the number of physical sensors. Furthermore, we can show that the model is able to predict unseen data from 168 different load profiles with an overall accuracy of 86.3 %. In several experiments we generated vibration data which is processed using the fast Fourier transform algorithm and spectrograms. We show that image classification of spectrograms with Convolutional Neural Networks significantly outperforms other models with fast Fourier transforms fed into Artificial Neural Networks or logistic regression.

Chirplet transform‐based machine‐learning approach towards classification of cognitive state change using galvanic skin response and photoplethysmography signals

AbstractRecent advances in the field of cognitive sciences have provided evidence to classify activities with different cognitive engagement levels. High cognitive load and sustained attention may predispose an individual to psychological and physical stress and may result in a lack of performance. Cognitive load monitoring of professionals in high precision fields has gathered significant attention in the current research scenario. The proposed work in continuity has tried to demonstrate the use of two non‐invasive physiological sensing modalities viz. galvanic skin response (GSR) and photoplethysmography (PPG) to elucidate effectiveness in picking up differences of two induced cognitive state categories based on machine learning and statistical methods. This paper demonstrates how GSR and PPG signals data can be effectively utilized to examine the shift in cognitive load. The present work proposes the use of general linear chirplet transform (GLCT) to evaluate the time‐frequency characteristic of GSR and PPG signals and utilize the statistical features in classification. Random Forest, Decision Tree, and k‐Nearest Neighbours demonstrated an accuracy of 92.13%, 88.0%, and 86.13% respectively on a dataset of 20 subjects against an optimized feature set, thus demonstrating the effectiveness of the proposed methodology for differentiating pre‐defined categories of cognitive load. The study shows the potential of GSR and PPG signals attributes and time‐frequency representation using GLCT to monitor cognitive load in real‐life conditions. The proposed work also indicates the possibility to extend the same to attention‐demanding fields such as aviation, medicine, and manufacturing for effective remediation of fatigue periods with increased cognitive demand, which, if sustained, may lead to cognitive decline in the long run.

Assessment of Activity Profiles in Older Adults and Lower Limb Bone Parameters: Observations from the Hertfordshire Cohort Study

As muscle strength and function decline with age the optimal high-impact physical activity (PA) required for bone remodelling is rarely achievable in older adults. This study aimed to explore the activity profiles of community-dwelling older men and women and to assess the relationship between individual PA profiles and lower limb bone parameters. Participants from the Hertfordshire Cohort Study wore triaxial accelerometers for 7 days and counts of low (0.5–1.0 g), medium (1.0–1.5 g), and high (> 1.5 g) vertical-impact activity were calculated. Two years later, participants underwent a pQCT scan of the tibia (4% and 38% sites) to obtain measures of bone mineral density and bone geometry. Linear regression was used to quantify associations between bone and PA loading profiles adjusting for age, sex, loading category, and BMI. Results are presented as β [95% confidence interval]. Bone and PA data were available for 82 participants. The mean (SD) age at follow-up was 81.4(2.7) years, 41.5% (n = 34) were women. The median low-impact PA count was 5281 (Inter-quartile range (IQR) 2516–12,977), compared with a median of only 189 (IQR 54–593) in medium, and 39 (IQR 9–105) in high-impact counts. Positive associations between high-impact PA and cortical area (mm2), polar SSI (mm3), and total area (mm2) at the 38% slice (6.21 [0.88, 11.54]; 61.94 [25.73, 98.14]; 10.09 [3.18, 16.99], respectively). No significant associations were found at distal tibia. These data suggest that maintaining high (> 1.5 g)-impact activity is difficult for older adults to achieve; however, even small amounts of high-impact PA are positively associated with selected cortical bone parameters 2 years later.

Power load identification based on Long-and-Short-Term Memory network and Affinity Propagation clustering algorithm

Load monitoring and decomposition have broad application prospects in the field of smart grid. Based on non-invasive load monitoring technology, a power load identification method based on Long-and-Short-Term Memory (LSTM) and Affinity Propagation (AP) clustering algorithm is proposed. The original power load data is processed by AP clustering algorithm, and a large number of measured sample data are divided into multiple clusters. The continuous variable state equipment load is discretized to obtain the power set of the sample equipment. The LSTM network model of load classification is established, input the processed data into the network model, and the optimal working state sequence and corresponding equipment power under the influence of error are obtained. The results obtained by LSTM algorithm alone and the predicted data obtained by combining LSTM algorithm with AP algorithm are compared with the actual load data, and the superiority of the algorithm is judged according to the degree of difference. The results show that the deviation between the two algorithms and the actual data is about 2%–7%, which is much lower than the deviation when LSTM algorithm is used alone.

Impaired detection of omicron by SARS-CoV-2 rapid antigen tests

Since autumn 2020, rapid antigen tests (RATs) have been implemented in several countries as an important pillar of the national testing strategy to rapidly screen for infections on site during the SARS-CoV-2 pandemic. The current surge in infection rates around the globe is driven by the variant of concern (VoC) omicron (B.1.1.529). Here, we evaluated the performance of nine SARS-CoV-2 RATs in a single-centre laboratory study. We examined a total of 115 SARS-CoV-2 PCR-negative and 166 SARS-CoV-2 PCR-positive respiratory swab samples (101 omicron, 65 delta (B.1.617.2)) collected from October 2021 until January 2022 as well as cell culture-expanded clinical isolates of both VoCs. In an assessment of the analytical sensitivity in clinical specimen, the 50% limit of detection (LoD50) ranged from 1.77 × 106 to 7.03 × 107 RNA copies subjected to the RAT for omicron compared to 1.32 × 105 to 2.05 × 106 for delta. To score positive in these point-of-care tests, up to 10-fold (LoD50) or 101-fold (LoD95) higher virus loads were required for omicron- compared to delta-containing samples. The rates of true positive test results for omicron samples in the highest virus load category (Ct values < 25) ranged between 31.4 and 77.8%, while they dropped to 0–8.3% for samples with intermediate Ct values (25–30). Of note, testing of expanded virus stocks suggested a comparable RAT sensitivity of both VoCs, questioning the predictive value of this type of in vitro-studies for clinical performance. Given their importance for national test strategies in the current omicron wave, awareness must be increased for the reduced detection rate of omicron infections by RATs and a short list of suitable RATs that fulfill the minimal requirements of performance should be rapidly disclosed.

STATISTICAL PARAMETERS ANALYSIS OF SEDIMENT GRAIN SIZE FROM RAYA RIVER BENGKAYANG REGENCY, WEST BORNEO

The study of the statistical analysis parameters of grain size in the waters of Sungai Raya was carried out in order to understand the characteristic of the sediment in the river and coastal areas. The grain size by sieving method, and analyze the statistical parameters including the average grain size, sorting, skewness, and kurtosis. Sediment samples were acquired from six stations with each station consisting of three sampling points representing river west bank, river midstream, river east bank, and the coastal area consisting of five stations. The results of the sediment analyses showed that the average grain size in the river ranges from 1.23 to 2.77 mm, and on the beach area is between 1,52 - 2,59 mm, classified as fine sand, medium sand, very fine sand, and coarse sand with predominant fine sand in all stations. The results of the statistical parameter analysis of the grain size of the bottom sediment in the waters of Sungai Raya exhibit the average diameter value ranging from 1.23 to 2.77 mm, classified as fine sand to medium sand. This value indicates that the type of sediment in this location is dominated by silt with grain sizes ranging from &lt; 0.05 - 0.002 mm. The sorting value ranges from 0.14 to 1.59 mm, categorized into moderately sorted, moderately well-sorted, very well sorted, and poorly sorted classes. The average value of skewness is 2.17 with the classification of the bed load being very fine and kurtosis value ranges from 0.72 to 1.67 indicates the classification in these waters is platycuric, mesokurtic, leptokurtic, and very leptokurtic. These finding concluded that in the study area there is a variation in the angularity of the sediment grains due to the different hydrodynamic processes

Battered minipile response to low-frequency cyclic lateral loading in very dense sand

This experimental study investigates the response of vertical and battered minipiles to two-way symmetrical low-frequency (0.1 Hz) cyclic lateral loading. Laboratory (1-g) tests were performed on scaled-down minipiles in very dense cohesionless soil, for batter angles of 0°, 25° and 45°. The cyclic loading is classified into two categories: multi-amplitude and long-term single amplitude, where force-controlled load was applied at a constant frequency. The minipiles were instrumented with optic fibres, and strain profiles were obtained at each loading stage, in both compression and tension stroke. The results are presented in terms of hysteresis loops, variation of normalised stiffness, minipile strain and bending moments under cyclic loading. In the multi-amplitude loading category, backbone curves show a stiffer force–displacement response in tension stroke than in compression stroke. For the single-amplitude category, the area of the hysteresis loop is largest for 45° battered minipiles with the lowest accumulated deformation. The normalised stiffness at the end of 50 cycles is highest for 25° minipiles with a value slightly greater than one. The strain profiles along the minipiles show stabilisation of measured strain before the number of cycle reaches 50, for all three battered conditions. A multi-surface hardening constitutive model is used to explain the effect of shearing and cyclic loading, with increasing loading amplitude on 25° battered minipiles. These test results are indicative of better performance capability of 25° battered minipiles, in terms of secant stiffness, compared to the vertical and 45° battered cases.

Dynamic load testing of a modular truss bridge using military vehicles

Determination of appropriate load factors ensures bridge assessments completed using limit states design are applicable for all potential loading scenarios. Extremely limited research has been completed to determine accurate Dynamic Load Allowance (DLA) values for military vehicle loading of bridges, specifically independent values for tracked and wheeled vehicle types. An experimental testing program was used to assess the dynamic load effects caused by these vehicle categories by subjecting a bridge to loading from multiple different vehicle types at different speeds, while controlling deck surface condition. It was found that tracked vehicles generated approximately half the dynamic bridge response of wheeled vehicles. Using this knowledge, tracked vehicle DLA value recommendations relevant for this bridge are made for use in the North Atlantic Treaty Organization (NATO) Military Load Classification (MLC) system, and in the Canadian Highway Bridge Design Code (CHBDC) and American Association of State Highway and Transportation Officials (AASHTO) bridge code. Further research in this area could lead to strategic increases in tracked vehicle mobility.

Resident user load classification method based on improved Gaussian mixture model clustering

In view of the limitation of “hard assignment” of clusters in traditional clustering methods and the difficulty of meeting the requirements of clustering efficiency and clustering accuracy simultaneously in regard to massive data sets, a load classification method based on a Gaussian mixture model combining clustering and principal component analysis is proposed. The load data are fed into a Gaussian mixture model clustering algorithm after principal component analysis and dimensionality reduction to achieve classification of large-scale load datasets. The method in this paper is used to classify loads in the Canadian AMPds2 public dataset and is compared with K-Means, Gaussian mixed model clustering and other methods. The results show that the proposed method can not only achieve load classification more effectively and finely, but also save computational cost and improve computational efficiency.

Smart architecture for software-defined networking

Software-defined networks (SDN) seek to solve the problems in current network schemes by simplifying their management through their reprogrammability and accessibility to the overall network infrastructure. One aspect to improve in SDN-based schemes is the precise classification of your traffic load, this can improve various aspects such as quality of service, dynamic access control, prioritized random access, among others. This research aims to propose a conceptual architecture of SDN and evaluate different machine learning methods for traffic classification. To this end, SDN architectures are analyzed and different modules are proposed to strengthen their management with the help of low computational cost classifiers. The architecture proposes the following main modules: Capture network traces module, Learning Engine module, and ML-model and Flow classifier. To determine the model to be used in the Learning Engine and Flow classifier modules, different classifiers were evaluated using a database of network traffic, as a result, it was determined that the gradient boosting algorithm is the most suitable to be integrated with the proposed SDN architecture.

Demand Response Method Considering Multiple Types of Flexible Loads in Industrial Parks

With the rapid development of the energy internet, the proportion of flexible loads in smart grid is getting much higher than before. It is highly important to model flexible loads based on demand response. Therefore, a new demand response method considering multiple flexible loads is proposed in this paper to character the integrated demand response (IDR) resources. Firstly, a physical process analytical deduction (PPAD) model is proposed to improve the classification of flexible loads in industrial parks. Scenario generation, data point augmentation, and smooth curves under various operating conditions are considered to enhance the applicability of the model. Secondly, in view of the strong volatility and poor modeling effect of Wasserstein-generative adversarial networks (WGAN), an improved WGAN-gradient penalty (IWGAN-GP) model is developed to get a faster convergence speed than traditional WGAN and generate a higher quality samples. Finally, the PPAD and IWGAN-GP models are jointly implemented to reveal the degree of correlation between flexible loads. Meanwhile, an intelligent offline database is built to deal with the impact of nonlinear factors in different response scenarios. Numerical examples have been performed with the results proving that the proposed method is significantly better than the existing technologies in reducing load modeling deviation and improving the responsiveness of park loads.

Machine Learning Based Self-Balancing and Motion Control of the Underactuated Mobile Inverted Pendulum With Variable Load

In this paper, a novel Machine Learning (ML) based Adaptive Fuzzy Logic-Proportional Integral (AFL-PI) controller was developed for the self-balancing and precision motion control of a two wheeled Underactuated-Mobile Inverted Pendulum (U-MIP) under variable payloads. One of the external disturbances in balance and movement control of the U-MIP is the amount of payload it carries on. To investigate the effectiveness of the proposed controller, a load bar was mounted on top of the U-MIP. The weights of 55gr each can be attached to this bar for variable payloads. The weights on the bar were labeled as three different classes: Low Load (LL), Normal Load (NL) and Heavy Load (HL). Artificial Neural Network (ANN), Linear Discriminant Analysis (LDA), Support Vector Machine (SVM) and k-Nearest Neighbors (k-NN) models were tested to obtain the highest load class estimation. The highest load classification accuracy was achieved with ANN. Therefore, the ANN model was applied on the U-MIP. The balance performance of the U-MIP was compared by applying the classical FL-PI and ANN based AFL-PI controller on the robot. In order to compare the body tilt angle performance of the U-MIP, the optimal FL-PI parameter in LL was applied for NL and HL conditions without changing. Then, the proposed ANN-based AFL-PI controller was implemented on U-MIP. With the proposed controller, the body tilt angle variation of the U-MIP was improved by %29.42 for NL and %55.62 for HL compared to the classical FL-PI controller. The validity of the proposed controller was proved by real experiments.

Augmented Mel-Spectrogram VGG-16 Model for Axial and Radial Load Classification at Wire-Race Bearings

Augmented Mel-Spectrogram VGG-16 Model for Axial and Radial Load Classification at Wire-Race Bearings