Non-intrusive Sensors Research Articles

A Self-Compensating Non-Intrusive Ring-Type AC Voltage Sensor Based on Capacitive Coupling

In order to reduce the influence of coupling capacitance variations on cable voltage measurement, this paper proposes a self-compensating non-intrusive ring-type AC voltage sensor based on capacitive coupling. A theoretical model of the sensor was established, and the influence of parasitic capacitance changes on sensor output was analyzed. Furthermore, a theoretical analysis shows that the parasitic capacitance between the external cable and the sensing probe, as well as between the ground and the sensing probe, will significantly affect the sensitivity of the sensor and increases the measurement error. A ring-type inductive probe and a signal processing circuit were designed, incorporating a reference signal to compensate for the influence of coupling capacitance variations. Additionally, to minimize the impact of parasitic capacitance on sensor output, the length of the outer ring electrode was extended, and a PTFE housing was designed for protection. A prototype of the sensor was developed and tested. This prototype has a good linear response to AC voltage in the measurement range of 0–1000 V with a linearity of 0.86%. The effects of changes in cable diameter and cable position on the measurement were tested separately. The worst-case error of the sensor output is less than 6.44%, representing a reduction of 21.4% compared to the uncompensated case. Under external cable interference, the sensor exhibited an output error of less than 1.85%. The results show that the designed sensor can accurately measure cable voltage despite changes in cable diameter or installation position, and also demonstrates effective shielding against external interference.

Unsupervised complex semi-binary matrix factorization for activation sequence recovery of quasi-stationary sources

Industry 5.0 advocates for a sustainable industry, particularly in terms of energy. One of the most fundamental elements of comprehension is knowing when the systems are operated, as this is key to locating energy intensive subsystems and operations. Such knowledge is often lacking in practice. Activation statuses can be recovered from sensor data though. Some non-intrusive sensors (accelerometers, current sensors, etc.) acquire mixed signals containing information about multiple actuators at once. This paper considers different industrial settings in which the identification of binary sub-system activation sequences is sought. In this context, source signals are assumed to be quasi-stationary and they may be correlated. Existing methods either restrict these assumptions, e.g., by imposing orthogonality or noise characteristics, or lift them using additional assumptions, typically using nonlinear transforms.This paper addresses these limitations, and introduces the unsupervised complex semi-binary matrix factorization (ℂSBMF) as its main contribution. In particular, we show that the exact recovery of source activation sequences from non-intrusive sensor data is intrinsically tied to the presence of problematic phase shifts, the causes of which are detailed. A greedy algorithm is proposed, iteratively resynchronizing sources to converge towards the maximum decomposition of each operation despite these phase shifts. The ℂSBMF is verified and compared to existing techniques on synthetic use cases, then validated on experimental data with signals of different nature. To that occasion, the CAFFEINE dataset for unsupervised time series multi-label classification is introduced.

A data-driven agent-based model of occupants’ thermal comfort behaviors for the planning of district-scale flexible work arrangements

In a global context of increasing flexibility in the way workplaces and the districts in which they are located are used, there is a need for occupant-driven approaches to plan urban energy systems. Several authors have suggested the use of agent-based models (ABM) of building occupants in urban building energy simulations due to their ability to reproduce emergent behaviors from individual agents’ actions. However, few works in the literature take full advantage of the ABM paradigm, accounting for both occupant presence and energy-relevant behaviors at the district scale. In this work, we propose a methodology to develop a data-driven, agent-based model of building occupants’ activities and thermal comfort in an urban district. Our methodology combines the use of campus-scale Wi-Fi data to derive feasible occupant activity and location plans, along with thermal preference profiles derived from a longitudinal field study where off-the-shelf, non-intrusive sensors were used to capture the right-here-right-now thermal preference of 35 participants in the same case study district. Our model is then used to explore how different district operation strategies could affect building energy performance in the context of increased workspace flexibility. Our results show that by providing a diversity of building operation conditions, with different buildings having different set point temperatures, occupants’ thermal comfort hours could be improved by an average of about 10% with little effect on district energy performance. Meanwhile, a 6%–15% average decrease in space cooling energy use intensity was observed when implementing occupant-driven ventilation and setpoint controls, regardless of location choice scenario.

Digital twin with augmented state extended Kalman filters for forecasting electric power consumption of industrial production systems

The work aims to develop an effective tool based on Digital Twins (DTs) for forecasting electric power consumption of industrial production systems. DTs integrate dynamic models combined with Augmented State Extended Kalman Filters (ASEKFs) in a learning process. The connection with the real counterpart is realized exclusively through non-intrusive sensors. This architecture enables the model development of industrial systems (components, machinery and processes) on which complete knowledge is not available, by identifying the model's unknown parameters through short online training phases and small amounts of real-time raw data. ASEKFs track the unknowns keeping models updated as physical systems evolve. When a forecast is needed, the current estimates of the uncertain parameters are integrated into the dynamic models. These can then be used without ASEKFs to predict the actual energy use of the system under the desired operating conditions, including scenarios that differ from typical functioning. The approach is validated offline with reference to the electricity consumption of an automatic coffee machine, which represents a real test environment and a blueprint to design DTs for other industrial systems. The appliance is observed by measuring the supply voltage and the absorbed current. The accuracy of the results is analyzed and discussed. This method is developed in the context of energy consumption prediction and optimization in the manufacturing industry through refined energy management and planning.

Detection of Anomalies in Daily Activities Using Data from Smart Meters.

The massive deployment of smart meters in most Western countries in recent decades has allowed the creation and development of a significant variety of applications, mainly related to efficient energy management. The information provided about energy consumption has also been dedicated to the areas of social work and health. In this context, smart meters are considered single-point non-intrusive sensors that might be used to monitor the behaviour and activity patterns of people living in a household. This work describes the design of a short-term behavioural alarm generator based on the processing of energy consumption data coming from a commercial smart meter. The device captured data from a household for a period of six months, thus providing the consumption disaggregated per appliance at an interval of one hour. These data were used to train different intelligent systems, capable of estimating the predicted consumption for the next one-hour interval. Four different approaches have been considered and compared when designing the prediction system: a recurrent neural network, a convolutional neural network, a random forest, and a decision tree. By statistically analysing these predictions and the actual final energy consumption measurements, anomalies can be detected in the undertaking of three different daily activities: sleeping, breakfast, and lunch. The recurrent neural network achieves an F1-score of 0.8 in the detection of these anomalies for the household under analysis, outperforming other approaches. The proposal might be applied to the generation of a short-term alarm, which can be involved in future deployments and developments in the field of ambient assisted living.

Application of Digitalisation in Regulated Environments for Predictive Failure Modelling

This paper explores the challenges of applying digitalization in regulated pharmaceutical manufacturing environments. A large range of complex equipment including pumps, valves and vessels may be associated with pharmaceutical batch production processes. Maintenance of such equipment are often based on reactive or preventative strategies which are not always effective and not completely successful in preventing costly downtime or scrap. This research examines how predictive maintenance Key Performance Indicators (KPIs) can be developed through data capture using non-intrusive sensors and their integration with production data derived from Programmable Logic Controllers (PLCs), Enterprise Resource Planning (ERP) systems, and Product Lifecycle Management (PLM) systems. The significance of regulation and the associated challenges in applying digitalization within such a highly regulated environment are also considered. This research aims to shed light on the potential benefits and challenges of implementing digital solutions for predictive maintenance in regulated manufacturing environments to contribute to the enhancement of operational efficiency and product quality while reducing costs due to outages.

Learning Motion Primitives for the Quantification and Diagnosis of Mobility Deficits.

The severity of mobility deficits is one of the most critical parameters in the diagnosis and rehabilitation of Parkinson's disease (PD). The current approach for severity evaluation is clinical scaling that relies on a clinician's subjective observations and experience, and the observation in laboratories or clinics may not suffice to reflect the severity of motion deficits as compared to daily living activities. The paper presents an approach to modeling and quantifying the severity of mobility deficits from motion data by using nonintrusive wearable physio-biological sensors. The approach provides a user-specific metric that measures mobility deficits in terms of the quantities of motion primitives that are learned from motion tracking data. The proposed method achieved 99.84% prediction accuracy on laboratory data and 93.95% prediction accuracy on clinical data. This approach presents the potential to supplant traditional observation-based clinical scaling, providing an avenue for real-time feedback to fortify positive progression throughout the course of rehabilitation.

Investigating Activity Recognition for Hemiparetic Stroke Patients Using Wearable Sensors: A Deep Learning Approach with Data Augmentation.

Measuring the daily use of an affected limb after hospital discharge is crucial for hemiparetic stroke rehabilitation. Classifying movements using non-intrusive wearable sensors provides context for arm use and is essential for the development of a home rehabilitation system. However, the movement classification of stroke patients poses unique challenges, including variability and sparsity. To address these challenges, we collected movement data from 15 hemiparetic stroke patients (Stroke group) and 29 non-disabled individuals (ND group). The participants performed two different tasks, the range of motion (14 movements) task and the activities of daily living (56 movements) task, wearing five inertial measurement units in a home setting. We trained a 1D convolutional neural network and evaluated its performance for different training groups: ND-only, Stroke-only, and ND and Stroke jointly. We further compared the model performance with data augmentation from axis rotation and investigated how the performance varied based on the asymmetry of movements. The joint training of ND + Stroke yielded an increased F1-score by a margin of 31.6% and 10.6% compared to ND-only training and Stroke-only training, respectively. Data augmentation further enhanced F1-scores across all conditions by an average of 11.3%. Finally, asymmetric movements decreased the F1-score by 25.9% compared to symmetric movements in the Stroke group, indicating the importance of asymmetry in movement classification.

Nonintrusive thermal-wave sensor for operando quantification of degradation in commercial batteries

Monitoring real-world battery degradation is crucial for the widespread application of batteries in different scenarios. However, acquiring quantitative degradation information in operating commercial cells is challenging due to the complex, embedded, and/or qualitative nature of most existing sensing techniques. This process is essentially limited by the type of signals used for detection. Here, we report the use of effective battery thermal conductivity (keff) as a quantitative indicator of battery degradation by leveraging the strong dependence of keff on battery-structure changes. A measurement scheme based on attachable thermal-wave sensors is developed for non-embedded detection and quantitative assessment. A proof-of-concept study of battery degradation during fast charging demonstrates that the amount of lithium plating and electrolyte consumption associated with the side reactions on the graphite anode and deposited lithium can be quantitatively distinguished using our method. Therefore, this work opens the door to the quantitative evaluation of battery degradation using simple non-embedded thermal-wave sensors.

Occupancy State Prediction by Recurrent Neural Network (LSTM): Multi-Room Context

The energy consumption of a building is significantly influenced by the habits of its occupants. These habits not only pertain to occupancy states, such as presence or absence, but also extend to more detailed aspects of occupant behavior. To accurately capture this information, it is essential to use tools that can monitor occupant habits without altering them. Invasive methods such as body sensors or cameras could potentially disrupt the natural habits of the occupants. In our study, we primarily focus on occupancy states as a representation of occupant habits. We have created a model based on artificial neural networks (ANNs) to ascertain the occupancy state of a building using environmental data such as CO2 concentration and noise level. These data are collected through non-intrusive sensors. Our approach involves rule-based a priori labeling and the use of a long short-term memory (LSTM) network for predictive purposes. The model is designed to predict four distinct states in a residential building. Although we lack data on actual occupancy states, the model has shown promising results with an overall prediction accuracy ranging between 78% and 92%.

An Unsupervised Method to Recognise Human Activity at Home Using Non-Intrusive Sensors

As people get older, living at home can expose them to potentially dangerous situations when performing everyday actions or simple tasks due to physical, sensory or cognitive limitations. This could compromise the residents’ health, a risk that in many cases could be reduced by early detection of the incidents. The present work focuses on the development of a system capable of detecting in real time the main activities of daily life that one or several people can perform at the same time inside their home. The proposed approach corresponds to an unsupervised learning method, which has a number of advantages, such as facilitating future replication or improving control and knowledge of the internal workings of the system. The final objective of this system is to facilitate the implementation of this method in a larger number of homes. The system is able to analyse the events provided by a network of non-intrusive sensors and the locations of the residents inside the home through a Bluetooth beacon network. The method is built upon an accurate combination of two hidden Markov models: one providing the rooms in which the residents are located and the other providing the activity the residents are carrying out. The method has been tested with the data provided by the public database SDHAR-HOME, providing accuracy results ranging from 86.78% to 91.68%. The approach presents an improvement over existing unsupervised learning methods as it is replicable for multiple users at the same time.

Design and development of a low-cost non-intrusive passive acoustic sensor

In this paper, we report on the development and performance study of a low-cost non-intrusive passive acoustic sensor (NIPAS), suitable for gas flow rate measurement and monitoring in pipelines and other similar applications. Detailed information on the design considerations, development, construction, and assembly of the sensor and its unit are also provided in this paper. The complete NIPAS unit is made up of two sensor heads fixed at the test section of a designed flow loop to receive the acoustic emission signals generated from the varying gas flow rate and processed by a control unit for data collection and analysis. The electronic unit was enclosed in a Perspex casing for easy interfacing and protection. Observations on the performance of the developed NIPAS are presented as a voltage-time signal and analyzed in both time and frequency domains. Calibration of the NIPAS was done using suitable statistical features, which are the mean, standard deviation, and coefficient of variation of the signals. Results obtained indicate that the developed NIPAS unit shows proof of concept and it is suitable for measuring and monitoring gas flow rates in pipes.

306 Shinyanimalcv: Interactive Web Application for Object Detection and Three-Dimensional Visualization of Animals Using Computer Vision

Abstract Precision livestock farming applies integrated sensors and information technology to provide individualized animal care in a timely manner. A computer vision system, a non-intrusive and cost-efficient sensor technology, may be useful for monitoring animals to support farm management decisions. While the advancement in camera sensors provides a great opportunity for producers to improve animal health and welfare sustainably, the limited availability of user-friendly image data processing software tools substantially hinders the implementation of computer vision in livestock production systems. The objective of this study was to develop ShinyAnimalCV, which is a Shiny-based interactive animal computer vision web application. This software tool offers a user-friendly graphical user interface for object detection and three-dimensional visualization. The object detection module employs the Mask-RCNN to precisely segment the masks (regions) of the focal animals from two-dimensional images. A Python-based computer vision library OpenCV was used to draw minimum bounding boxes covering the segmented masks for object detection. The three-dimensional visualization module takes the depth map file captured from a top-view three-dimensional camera as an input, which contains the numerical distances between the camera and the objects (animals and background). The depth map file was first converted to a heatmap image, followed by identifying and segmenting the animal from the background using the Mask-RCNN model. The object detection module returns detection results, including the detected animal’s location, class/type, and morphological traits (e.g., body length and width). The visualization module generates a segmented mask to extract the height of the animal from the depth map file, which can be used to interactively visualize the three-dimensional surface of the animal and estimate its morphological traits, including length, width, height, and volume. By integrating these two modules into R Shiny, we deployed ShinyAnimalCV on a cloud server with pre-trained Mask-RCNN models using pig and cattle data to allow users to upload custom data and perform object detection and three-dimensional surface visualization. The features extracted by ShinyAnimalCV are expected to be useful for performing animal identification, feed intake monitoring, body weight predictions, and body condition score estimations. We conclude that the newly developed ShinyAnimalCV could facilitate the application of computer vision in the animal science community.

Large scale indoor occupant tracking using distributed acoustic sensing and machine learning

The recent COVID-19 pandemic has brought attention to the criticality of implementing measures to prevent viral transmission and reduce the risk of infection. It is necessary to identify crowd-gathering hazards and implement appropriate engineering control measures. Consequently, obtaining comprehensive footfall information at a large scale becomes significant. Vibration-based sensors are currently the preferred choice for footfall detection, with geophones being the commonly utilized sensors known for their effective performance. However, the deployment of geophones in large-scale scenarios presents challenges due to the substantial amount of manual labor involved. To address these limitations, the Distributed Acoustic Sensing (DAS) system, a non-intrusive sensor, is introduced as an alternative solution. While DAS has been successfully applied in various long-distance scenarios, such as pipeline surveillance and real-time train tracking, its effectiveness in footfall detection requires further evaluation. Therefore, this study proposes a machine learning method to achieve footstep recognition using DAS, considering the inherent time dependency and spatial continuity present in footfall trajectories. To further evaluate the effectiveness of DAS, geophones are selected as the benchmark sensors. The evaluation process encompasses footstep recognition and additional data implementation. The results demonstrate that the proposed recognition method achieves high accuracy for both sensors, thereby establishing DAS as a more applicable solution for footfall detection compared to geophones. In conclusion, DAS offers the advantage of flexible space resolution settings and adaptable deployment options, rendering it suitable for applications in diverse scenarios. Its capabilities make it suitable for large-scale occupancy monitoring scenarios.

Online Tg estimation for blade manufacturing

Cost reduction is one of the main drivers in the production of wind turbine blades in the last few years. In this direction, the use of intelligent process monitoring to optimize the infusion, curing and bonding by measuring online the actual process can reduce energy and increase productivity by 20% while ensuring product quality. In this framework, durable non-intrusive and disposable sensors have been developed for the monitoring of resin arrival, the resin’s viscosity during infusion and the on-going Tg during curing by measuring the resin’s electrical resistivity and temperature.

Chemiluminescence- and machine learning-based monitoring of premixed ammonia-methane-air flames

This work presents the development and validation of an algorithm capable of predicting the equivalence ratio and the ammonia fraction of premixed ammonia-methane-air flames using only measured OH*, NH*, CN*, and CH* chemiluminescence intensities as input. This machine learning algorithm relies on Gaussian process regression (GPR). It was trained and validated with data previously recorded in laminar flames, and it was then tested with new data recorded in more practical, turbulent swirl flames. The algorithm performs well for laminar and turbulent flames for wide ranges of equivalence ratio (0.80 ≤ ϕ ≤ 1.20) and ammonia fraction (0 ≤ XNH3 ≤ 0.60). For turbulent swirl flames, the prediction errors in the equivalence ratio and on the ammonia fraction are smaller than 0.05, except for a very small subset of operating conditions where the error is up to 0.10. Additional tests were performed by adding NO* and CO2* to the list of inputs, but this did not improve the predictions. The GPR algorithm was then benchmarked against linear and polynomial regressions and a more conventional way of inferring flame properties from chemiluminescence measurements, namely the ratio-based method. This method relies only on CN*/NO* and NH*/CH* ratios to predict the equivalence ratio and the ammonia fraction. Its prediction errors were often larger than 0.15, which is significantly worse than that of the GPR algorithm. Consequently, this work constitutes a solid basis for the future development of non-intrusive sensors to monitor practical ammonia-methane-air flames.

VIDEO CAMERA TECHNOLOGY FOR VEHICLE COUNTING IN TRAFFIC CENSUS: ISSUES, STRATEGIES AND OPPORTUNITIES

This study provides an overview of the sensor technologies commonly used for automated vehicle classification and counting, with a focus on non-intrusive sensors. Video cameras are found to be the most feasible solution for data collection in traffic census as it can operate in portable mode and used at any location. Several factors must be considered to ensure accurate counting. These involve optimum placement of the camera to ensure that all vehicles can be observed, and the lighting conditions must be considered to ensure good video quality. These further contributes to accurate classification and counting of vehicles by dedicated deep learning algorithm. As the data collection may involve location with poor access to cloud computing and storage, offline processing is therefore recommended. The study also revealed opportunities for solving issues related to strategic placement of video cameras, and development of dedicated deep learning algorithms.

A fast sensor for non-intrusive measurement of concentration and temperature in turbine exhaust

Accurate and rapid measurement of water vapor concentration and temperature in the exhaust of gas turbine engines is critical for monitoring their health and operational performance. To monitor the two parameters simultaneously, a non-intrusive sensor based on tunable diode laser absorption spectroscopy is developed using 8 laser beams targeting 3 transitions of water vapor. The sensor has in situ validated on a commercial auxiliary power unit. Seven parallel laser beams at 5000.2 cm−1 are 6.3 mm spaced at the plume boundary to characterize the edge effect between hot exhaust plume and cold surrounding flow. One beam, operating at the dual wavenumbers of 7185.6 cm−1 and 7444.4 cm−1 penetrates the plume through centerline to measure temperature via ratio thermometry. Then, this temperature information is used to obtain the concentration of water vapor. A temporal resolution of 4 ms is achieved for the 8-beam measurement, enabling high-speed and simultaneous quantification of the edge effects and the plume parameters. Results indicate water vapor concentration and temperature in the exhaust measured by the developed sensor are highly consistent with the traditional benchmarks, e.g., extractive sampling and thermocouples, with a difference of 0.02% and 3 °C, respectively. Enabled by its continuous millisecond-level measurements, the sensor, for the first time, reveals hidden engine behaviors and combustion dynamics that are unable to be observed by the traditional methods, thus facilitating next-generation real-time gas turbine engine control towards low emissions.

Characterization of in-cylinder spatiotemporal flame and solid particle emissions for ethanol-gasoline blended in gasoline direct injection engines

Stricter regulations for internal combustion engines have necessitated the inclusion of eco-friendly fuels with conventional ones to reduce gas and solid emissions. Gasoline direct injection (GDI) engines, however, have been found to produce more particle number and mass emissions than port fuel injection engines. To address this, bioethanol fuel has been selected as an additive eco-friendly fuel for gasoline, with the aim of reducing particle emissions. In this study, we quantitatively characterize in-cylinder spatiotemporal flame to feature combustion performance and particle emissions for ethanol fuel in the GDI combustion chamber. We examined three engine combustion modes differed in equivalence ratio and injection strategy to produce various combustion results to quantify combustion performance and particle emission characteristics. Using an eight-channel non-intrusive flame luminosity sensor in one cylinder, we measured the flame front and its propagation direction. At the tailpipe, we measured particulate emissions using an engine exhaust particle sizer and a micro soot sensor coupled with a catalytic stripper that removed semi-volatile compounds. Our study found that increasing ethanol fuel content for different combustion modes resulted in distinct combustion and flame development. The particle size distribution also showed different patterns at each combustion mode, depending on the ethanol content. Lean combustion modes yielded high diffusion flame intensity compared to stoichiometric combustion mode, which resulted in a larger amount of particle formation. The physical properties of ethanol fuel were found to predominantly determine the fuel-air mixture quality, combustion process, and flame development at each combustion mode. An increase in ethanol fuel content at lean-homogeneous mode resulted in higher diffusion flame intensity and larger particle emissions. Our comparative study of flame and solid particles for ethanol fuel content improves the understanding of in-cylinder combustion processes and their correlations.

A TEG-Based Non-Intrusive Ultrasonic System for Autonomous Water Flow Rate Measurement

Residential water meters accommodate various methods of power provisioning. Electromagnetic and ultrasonic meters, for example, often rely on a battery-like external power source, whereas mechanical meters harvest energy from water flow through an impeller. Although energy harvesting (EH) minimizes maintenance needs driven by battery depletion/replenishment, placing a physical element into the flow adversely affects water pressure. This intrusive EH/sensing technique is not user-friendly either since the meters with impellers need to be embedded into pipes by skilled personnel. Hence, this paper proposes a non-intrusive sensor system powered by thermoelectric generators (TEGs) for <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">plug-and-play</i> water flow rate measurement. This system, equipped with a custom-made energy management unit (EMU), adopts ultrasonic sensors, a task-based computing scheme, and a LoRa module for autonomous sensing and reporting of the flow rate. After summarizing thermoelectricity and delta time-of-flight ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\Delta$</tex-math></inline-formula> ToF)-based ultrasonic sensing theory, we provide the system model and design details with a particular focus on the EMU. Then, we experimentally evaluate the system under varying conditions, demonstrating their impact on average sensing and transmission periods. The results unveil that our proposal can achieve high measurement precision ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\pm 1.4\%$</tex-math></inline-formula> ), comparable to its intrusive and battery-powered counterparts, and thus has the potential of replacing the residential water meters.