| 1. |
- Adin, Veysi, et al.
(författare)
-
Tiny Machine Learning for Damage Classification in Concrete Using Acoustic Emission Signals
- 2023
-
Ingår i: 2023 IEEE International Instrumentation and Measurement Technology Conference (I2MTC). - : IEEE. - 9781665453837
-
Konferensbidrag (refereegranskat)abstract
- Acoustic emission (AE) is a widely used non-destructive test method in structural health monitoring applications to identify the damage type in the material. Usually, the analysis of the AE signal is done by using traditional parameter-based methods. Recently, machine learning methods showed promising results for the analysis of AE signals. However, these machine learning models are complex, slow, and consume significant amounts of energy. To address these limitations and to explore the trade-off between model complexity and the classification accuracy, this paper presents a lightweight artificial neural network model to classify damage types in concrete material using raw acoustic emission signals. The model consists of one hidden layer with four neurons and is trained on a public acoustic emission signal dataset. The created model is deployed to several microcontrollers and the performance of the model is evaluated and compared with a state-of-the-art machine learning model. The model achieves 98.4% accuracy on the test data with only 4019 parameters. In terms of evaluation metrics, the proposed tiny machine learning model outperforms previously proposed models 10 to 1000 times. The proposed model thus enables machine learning in real-time structural health monitoring applications.
|
|
| 2. |
- Adin, Veysi, et al.
(författare)
-
Tiny Machine Learning for Real-Time Postural Stability Analysis
- 2023
-
Ingår i: 2023 IEEE Sensors Applications Symposium (SAS). - : IEEE conference proceedings. - 9798350323078
-
Konferensbidrag (refereegranskat)abstract
- Postural sway is a critical measure for evaluating postural control, and its analysis plays a vital role in preventing falls among the elderly. Typically, physiotherapists assess an individual's postural control using tests such as the Berg Balance Scale, Tinetti Test, and time up-and-go test. Sensor-based analysis is available based on devices such as force plates or inertial measurement units. Recently, machine learning methods have demonstrated promising results in the sensor-based analysis of postural control. However, these models are often complex, slow, and energy-intensive. To address these limitations, this study explores the design space of lightweight machine learning models deployable to microcontrollers to assess postural stability. We developed an artificial neural network (ANN) model and compare its performance to that of random forests, gaussian naive bayes, and extra tree classifiers. The models are trained using a sway dataset with varying input sizes and signal-to-noise ratios. The dataset comprises two feature vectors extracted from raw accelerometer data. The developed models are deployed to an ARM Cortex M4-based microcontroller, and their performance is evaluated and compared. We show that the ANN model has 99.03% accuracy, higher noise immunity, and the model performs better with a window size of one second with 590.96 us inference time.
|
|
| 3. |
- Aranda, Jesus Javier Lechuga
(författare)
-
Towards Self-Powered Devices Via Pressure Fluctuation Energy Harvesters
- 2021
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- The growing interest in the Internet of Things has created a need for wireless sensing systems for industrial and consumer applications. In hydraulic systems, a widely used method of power transmission in industry, wireless condition monitoring can lead to reduced maintenance costs and increase the capacity for sensor deployment. A major problem with the adoption of wireless sensors is the battery dependence of current technologies. Energy harvesting from pressure fluctuations in hydraulic systems can serve as an alternative power supply and enable self-powered devices. Energy harvesting from pressure fluctuations is the process of converting small pressure fluctuations in hydraulic fluid into a regulated energy supply to power low power electronics. Previous studies have shown the feasibility of pressure fluctuation harvesting. However, for the development of self-powered sensor systems, the methods and techniques for converting pressure fluctuations into electrical energy should be further investigated.This thesis explores the methods, limitations, opportunities and trade-offs involved in the development of pressure fluctuation energy harvesters in the context of self-powered wireless devices. The focus is on exploring and characterizing the various mechanisms required to convert pressure fluctuations into electrical energy. In this work, an energy harvesting device consisting of a fluid-to-mechanical interface, an acoustic resonator, a piezoelectric stack, and an interface circuit is proposed and evaluated. Simulations and experimental analysis were used to analyse these different components for excitation relevant to hydraulic motors.The results of this work provide new insights into the development of power supplies for self-powered sensors for hydraulic systems using pressure fluctuation energy harvesters. It is shown that with the introduction of the space coiling resonator for pressure fluctuation amplification and a detailed analysis of the fluid interface and power conditioning circuits, the understanding of the design and optimization of efficient pressure fluctuation energy harvesters is further advanced.
|
|
| 4. |
- Aranda, Jesus Javier, et al.
(författare)
-
Self-powered wireless sensor using a pressure fluctuation energy harvester
- 2021
-
Ingår i: Sensors. - : MDPI AG. - 1424-8220. ; 21:4
-
Tidskriftsartikel (refereegranskat)abstract
- Condition monitoring devices in hydraulic systems that use batteries or require wired infrastructure have drawbacks that affect their installation, maintenance costs, and deployment flexibility. Energy harvesting technologies can serve as an alternative power supply for system loads, eliminating batteries and wiring requirements. Despite the interest in pressure fluctuation energy harvesters, few studies consider end-to-end implementations, especially for cases with lowamplitude pressure fluctuations. This generates a research gap regarding the practical amount of energy available to the load under these conditions, as well as interface circuit requirements and techniques for efficient energy conversion. In this paper, we present a self-powered sensor that integrates an energy harvester and a wireless sensing system. The energy harvester converts pressure fluctuations in hydraulic systems into electrical energy using an acoustic resonator, a piezoelectric stack, and an interface circuit. The prototype wireless sensor consists of an industrial pressure sensor and a low-power Bluetooth System-on-chip that samples and wirelessly transmits pressure data. We present a subsystem analysis and a full system implementation that considers hydraulic systems with pressure fluctuation amplitudes of less than 1 bar and frequencies of less than 300 Hz. The study examines the frequency response of the energy harvester, the performance of the interface circuit, and the advantages of using an active power improvement unit adapted for piezoelectric stacks. We show that the interface circuit used improves the performance of the energy harvester compared to previous similar studies, showing more power generation compared to the standard interface. Experimental measurements show that the self-powered sensor system can start up by harvesting energy from pressure fluctuations with amplitudes starting at 0.2 bar at 200 Hz. It can also sample and transmit sensor data at a rate of 100 Hz at 0.7 bar at 200 Hz. The system is implemented with off-the-shelf circuits.
|
|
| 5. |
- Bader, Sebastian, 1984-, et al.
(författare)
-
A Comparison of One- and Two-Diode Model Parameters at Indoor Illumination Levels
- 2020
-
Ingår i: IEEE Access. - 2169-3536. ; 8, s. 172057-172064
-
Tidskriftsartikel (refereegranskat)abstract
- Indoor photovoltaic (PV) application gains in attraction for low-power electronic systems, which requires accurate methods for performance predictions in indoor environments. Despite this, the knowledge on the performance of commonly used photovoltaic device models and their parameter estimation techniques in these scenarios is very limited. Accurate models are an essential tool for conducting feasibility analyses and component dimensioning for indoor photovoltaic systems. In this paper, we therefore conduct a comparison of the one- and two-diode models with parameters estimated based on two well-known methods. We evaluate the models' performance on datasets of photovoltaic panels intended for indoor use, and illumination conditions to be expected in indoor environments lit by artificial light sources. The results demonstrate that the one-diode model outperforms the two-diode model with respect to the estimation of the overall I-V characteristics. The two-diode model results instead in lower maximum power point errors. Both models show a sensitivity to initial conditions, such as the selection of the diode ideality factor, as well as the curve form of the photovoltaic panel to be modeled, which has not been acknowledged in previous research.
|
|
| 6. |
- Bader, Sebastian, 1984-, et al.
(författare)
-
Distributed Measurement of Light Conditions for Indoor Photovoltaic Applications
- 2020
-
Ingår i: Proceedings of IEEE Sensors. - 9781728168012
-
Konferensbidrag (refereegranskat)abstract
- Ambient light measurements and an understanding of light conditions are essential for the accurate estimation of available energy in indoor photovoltaic applications. Light conditions may vary with respect to illumination intensity, duration, and spectral composition. Although the importance of the light spectrum has been documented in laboratory studies, previous distributed measurement methods are limited to intensity as a measure for output power. In this paper, we propose and implement a system for distributed measurement of light conditions that includes spectral information with low overhead. Based on a prototype implementation, we demonstrate that the illumination intensity and spectrum varies considerably over time and space, which confirms the demand for the proposed solution. We, moreover, characterize the energy consumption of the prototype, demonstrating that long-term, unattended characterization of light conditions can be achieved.
|
|
| 7. |
- Bader, Sebastian, 1984-, et al.
(författare)
-
Instrumentation and Measurement Systems : The Challenge of Designing Energy Harvesting Sensor Systems
- 2024
-
Ingår i: IEEE Instrumentation & Measurement Magazine. - : Institute of Electrical and Electronics Engineers (IEEE). - 1094-6969 .- 1941-0123. ; 27:4, s. 22-28
-
Tidskriftsartikel (refereegranskat)abstract
- With the advent of low-cost and low-power computation, communication and sensor devices, novel instrumentation and measurement applications have been enabled, such as real-time industrial condition monitoring and fine-grained environmental monitoring. In these application scenarios, a lack of available infrastructures for communication and power supply is a common problem. In industrial applications, for example, the machine to be monitored and the monitoring system itself have significantly different technology lifespans, which requires that the monitoring system be retrofitted to machines that are already in use. In environmental monitoring, measurement systems are deployed as standalone devices in potentially remote areas. Consequently, the more autonomous the sensor system can be in terms of required infrastructure, the better it can match application and business needs.
|
|
| 8. |
- Haller, Stefan, 1982-
(författare)
-
Towards Low-Voltage, High-Current : A pioneering drive concept for battery electric vehicles
- 2021
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- The first electric low-voltage vehicles were constructed in the mid-19th century, but by the early 20th century they were progressively replacedby successors with internal combustion engines. As the consequences ofusing fossil fuels are better understood, our society is now transitioning back. The strong driving force towards electric transportation can be traced to several events and trends. The foremost of these is perhaps the rising awareness of climate change and the necessary reduction of the environmental footprint, as well associated political will for change. Alongside this, the pioneering automotive company Tesla, Inc. showed what electric cars are capable of and how to easily charge them along the road. The diesel gate unearthed in 2015, also played a major role. This transition is not without challenges, however. An electric car is expected to be reasonable priced, sustainable, environmentally friendly and electrically safe, even in case of an accident. Overnight charging at home should be possible, as well as the ability to quickly charge while in transit. While the industry has long experience with high-voltage electrical machines, the required battery technology is quite new and low-voltage in nature. Currently, the battery is the most costly part of an electric drivetrain and it has the highest environmental impact. Efficient battery use is therefore key for sustainability and a responsible consumption of the resources available. Nonetheless, most electric vehicles today use lethal high-voltage traction drives which require a considerable isolation effort and complex battery pack. Previous research results showed that a 48 V drivetrain compared to a high-voltage one, increases the drive-cycle efficiency. Hence, similar driving range can be reached with a smaller battery. This thesis provides an introduction to low-voltage, high-current, battery-powered traction drives. With the aim of increasing efficiency, safety and redundancy while reducing cost, a solution that breaks with century-old electric machine design principles is proposed and investigated. An overview and motivation to further investigate 48 V drivetrains with intrinsically safe and redundant machines is provided. The main focus of this work is the practical implementation of multi-phase low-voltage but high-current machines with integrated power electronics as well as components for a 48 V drivetrain. With this work, it is confirmed that today’s MOSFETs are not the limiting factor towards low-voltage, high-current drives. In the first part of this work, two small-scale prototype machines were constructed and tested. The air-cooled, small-scale 1.2 kW proto-type reached a copper fill-factor of 0.84. The machine’s low terminal-to-terminal resistance of 0.23 mΩ, including the MOSFET-based power electronics, allowed continuous driving currents up to 600 A. The resistive MOSFET losses stayed below 21 W. The second part focuses on the key components for a 48 V high-power drivetrain. A W-shaped coil for a multiphase 48 V machine with direct in-conductor cooling was designed and tested. With glycolwater, it reached a current density of 49.5 A/mm2 with 0.312 l/min flowrate. Furthermore, a reconfigurable battery pack for 48 V driving andhigh-voltage charging was investigated.
|
|
| 9. |
- Kelati, Amleset
(författare)
-
Data-driven Implementations for Enhanced Healthcare Internet-of-Things Systems
- 2021
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Healthcare monitoring systems based on the Internet of Things (IoT) areemerging as a potential solution for reducing healthcare costs by impacting and improving the quality of health care delivery. The rising numberof elderly and chronic patient population in the world and the associatedhealthcare costs urges the application of IoT technology to improve andsupport the health care services. This thesis develops and integrates twoIoT-based healthcare systems aiming to support elderly independent livingat home. The first one involves using IoT-based remote monitoring for paindetection, while the second one detects behavioral changes caused by illnessvia profiling the appliances’ energy usage.In the first approach, an Electromyography (EMG )sensor node with aWireless Fidelity (Wi-Fi) radio module is designed for monitoring the painof patients living at home. An appropriate feature-extraction and classification algorithm is applied to the EMG signal. The classification algorithmachieves 98.5% accuracy for the experimental data collected from the developed EMG sensor node, while it achieves 99.4% classification accuracy forthe clinically approved pain intensity dataset. Moreover, the experimentalresults clearly show the relevance of the proposed approaches and provetheir suitability for real-life applications. The developed sensor node for thepain level classification method is beneficial for continuous pain assessmentto the smart home-care community.As a complement to the first approach, in the second approach, an IoTbased smart meter and a set of appliance-level load profiling methods aredeveloped to detect the electricity usage of users’ daily living at home, whichindirectly provides information about the subject’s health status. The thesishas formulated a novel methodology by integrating Non-intrusive ApplianceLoad Monitoring (NIALM) analysis with Machine Learning- (ML) basedclassification at the fog layer. The developed method allows the detectionof a single appliance with high accuracy by associating the user’s Activitiesof Daily Living (ADL). The appliances detection is performed by employinga k-Nearest Neighbors (k-NN) classification algorithm. It achieves 97.4% accuracy, demonstrating its high detection performance. Due to the low cost and reusability advantages of Field Programmable Gate Arrays (FPGA),the execution of k-NN for appliances classification model is performed onan FPGA. Its classification performance was comparable with other computing platforms, making it a cost-effective alternative for IoT-based healthcare assessment of daily living at home. The developed methods have haspractical application in assisting real-time e-health monitoring of any individual who can remain in the comfort of their normal living environment.
|
|
| 10. |
- Ma, Xinyu, et al.
(författare)
-
Estimating Harvestable Energy in Time-Varying Indoor Light Conditions
- 2020
-
Ingår i: ENSsys 2020 - Proceedings of the 8th International Workshop on Energy Harvesting and Energy-Neutral Sensing Systems. - New York, NY, USA : Association for Computing Machinery (ACM). - 9781450381291 ; , s. 71-76
-
Konferensbidrag (refereegranskat)abstract
- Ambient light energy harvesting is a cost-effective and mature approach for supplying low-power sensor systems with power in many indoor applications. Although the spectral information of a light source is known to influence the efficiency and output power of a photovoltaic cell, the spectrum of the ambient illumination is due to measurement complexity often neglected when characterizing light conditions for power estimation purposes. In this paper we evaluate the influence of considering spectral information on the energy estimation accuracy. We create a dataset of varying light conditions in a typical indoor environment based on eight locations. For each location, we compare the energy estimation accuracy with and without spectral considerations. The results of this investigation demonstrate that a spectrum-based method leads to significant performance improvements in cases where the light condition is not defined by a single light source.
|
|
