Showing papers by "Consolatina Liguori published in 2022"

The Obtainable Uncertainty for the Frequency Evaluation of Tones with Different Spectral Analysis Techniques

Abstract: Spectral analysis is successfully adopted in several fields. However, the requirements and the constraints of the different cases may be so varied that not only the tuning of the analysis parameters but also the choice of the most suitable technique can be a difficult task. For this reason, it is important that a designer of a measurement system for spectral analysis has knowledge about the behaviour of the different techniques with respect to the operating conditions. The case that will be considered is the realization of a numerical instrument for the real-time measurement of the spectral characteristics of a multi-tone signal (amplitude, frequency, and phase). For this purpose, different signal processing techniques can be used, that can be classified as parametric or non-parametric methods. The first class includes those methods that exploit the a priori knowledge about signal parameters, such as the spectral shape of the signal to be processed. Thus, a self-configuring procedure based on a parametric algorithm should include a preliminary evaluation of the number of components. The choice of the right method among several proposals in the literature is fundamental for any designer and, in particular, for the developers of spectral analysis software, for real-time applications and embedded devices where time and reliability constrains are arduous to fulfil. Different aspects should be considered: the desired level of accuracy, the available elaboration resources (memory depth and processing speed), and the signal parameters. The present paper details a comparison of some of the most effective methods available in the literature for the spectral analysis of signals (IFFT-2p, IFFT-3p, and IFFTc, all based on the use of an FFT algorithm, while improving the spectral resolution of the DFT with interpolation techniques and three parametric algorithms—MUSIC, ESPRIT, and IWPA). The methods considered for the comparison will be briefly described, and references to literature will be given for each one of them. Then, their behaviour will be analysed in terms of systematic contribution and uncertainty on the evaluated frequencies of the spectral tones of signals created from superimposed sinusoids and white Gaussian noise.

A novel image processing technique based on deep learning for water consumption detection

Abstract: In recent years, traditional image processing techniques have seen the introduction of novel tools, able to face issues that are not always handy with classical vision algorithms. For example, classical image processing algorithms (measurement, detection of features, and many others) require a controlled environment, like illumination, target positioning, and vibration that can influence the scene for the correct operation. On the other hand, the machine learning approaches enabled image processing techniques also in non-controlled environments. One of these applications can be represented by developing a leak detector at the household level, based on processing pictures of the mechanical water meter dial. The proposed research investigates using a deep learning approach to detect the minimal movement of the water meter needles related to water leakage. In particular, a CNN was trained to correlate successive differences on the water meter dial images taken with an applied calibrated water flow. From this analysis, it is possible to detect the absence of periods with null consumption and thus detect small water losses.

Image Scaling Effects on Deep Learning Based Applications

Abstract: The sophistication and high accuracy of Deep Neural Networks have gotten significant attention in recent years, with a wide range of applications making use of their capabilities. However, the deployment of such networks still faces limitations due to the high volume of data to be processed and the high computational requirements. In this article we focus on the effects that data volume reduction, due to image compression and scaling down the image resolution, will have on the detection accuracy for the design case of a powered wheelchair guidance system. Throughout our analysis we show that the reduction in image resolution to a factor of $16\times$ in image area alongside with JPEG compression provides a detection accuracy of over 0.93 in mAP, while the additional error in the position estimation of the caregiver is less than 0.5 cm. By reducing the data volume we inherently reduce the communication energy consumption, which is reduced by more than one order of magnitude. These results prove that we can overcome the complexity of high data volume for the deployment of DNNs in resource constrained IoT applications by interlacing the effects of image compression and resolution reduction, maintaining the accuracy and reducing the node energy consumption.

Development of a new speed measurement technique based on deep learning

Abstract: The more pervasive use of Artificial Intelligence (AI) has enabled feature extraction enhancement in several fields, particularly in image processing applications. Thanks to AI, it is possible to use low-cost devices (e.g., webcams and surveillance cameras in complex scenarios) like vehicle speed measurement, obtaining a significant reduction of instrument costs and a great spread of their use. The paper presents a new industrial measurement instrument for vehicle speed based on non-dedicated hardware devices and innovative image processing methods like Regional CNN (Convolutional Neural Network). The proposed hardware is based on a generic surveillance camera or even a simple webcam and an R-CNN to transform it into an intelligent tool capable of estimating the speed of a vehicle and tracking its movement under controlled conditions. One of the essential aspects of the work concerns the metrological characterization of the proposed method. Measurement uncertainty has been evaluated. The metrological characterization of approaches using artificial intelligence can be fundamental for spreading such technologies in practical scenarios and impulse the industrial development of enhanced tools that can comply with legal regulations for speed measurement. The measured velocities of a car under test have been compared with a reference constituted by the vehicle speed retrieved by the ABS ECU.

Acoustic Monitoring of Environmental Noise Based on Sampling Approach

Abstract: Generally the estimation of traffic noise levels is carried out by using acoustic monitoring systems. In this paper, starting from the results of a study on the environmental noise pollution of a city of the South of Italy, a statistic procedure focused on the bootstrap method is presented to evaluate the sound pressure levels of one month acquisition. The experimental verification against real data shows the consistency of the algorithm for forecasting the long-term noise indicator.

Parameter estimation for fault detection and classification in centrifugal pumps

Abstract: Fault detection and diagnosis, have been topics of great interest in both academia and industry. A well-designed strategy can considerably save operating costs by minimizing services downtime and replaceable items costs. A good knowledge of the operating machine’s dynamics, a good measurements technique can significantly add value to the strategy design. Different approaches are nowadays used to effectively detect and diagnose faults. They can be summarized as data-driven, analytically based, and knowledge-based methods. This paper discussed a parametric approach on a steady-state for fault detection on pumps. The parametric estimation method requires good modeling of the system dynamics and proper use of the parameters as features for designing a diagnosis algorithm. The paper demonstrates that the correlation induced by the variation between the head and torque could be easily detected. The paper also suggests using the likelihood ratio test, in order to discretize parameters likely to hold direct information on the fault.