Search

Ultrasonic diagnostics is the earliest way to predict industrial faults. Usually, a contact microphone is employed for detection, but the recording will be contaminated with noise. In this paper, a dataset that contains 10 main faults of pipelines and motors is analyzed from which 30 different features in the time and frequency domains are extracted. Afterward, for dimensionality reduction, principal component analysis (PCA), linear discriminant analysis (LDA), and t-distributed stochastic neighbor embedding (t-SNE) are performed. In the subsequent phase, recursive feature elimination (RFE) is employed as a strategic method to analyze and select the most relevant features for the classifiers. Next, predictive models consisting of k-Nearest Neighbor (KNN), Logistic Regression (LR), Decision Tree (DT), Gaussian Naive Bayes (GNB), and Support Vector Machine (SVM) are employed. Then, in order to solve the classification problem, a stacking classifier based on a meta-classifier which combines multiple classification models is introduced. Furthermore, the k-fold cross-validation technique is employed to assess the effectiveness of the model in handling new data for the evaluation of experimental results in ultrasonic fault detection. With the proposed method, the accuracy is around 5% higher over five cross folds with the least amount of variation. The timing evaluation of the meta model on the 64 MHz Cortex M4 microcontroller unit (MCU) revealed an execution time of 11 ms, indicating it could be a promising solution for real-time monitoring. [ABSTRACT FROM AUTHOR]

Abstract the process of modern integrated circuit (IC) design has been challenged by many factors. One of the most important challenges is the variation of device and circuit parameters during the manufacturing process. In this paper, the effects of manufacturing process variations on the gate delay have been modeled and an accurate yet low-cost simulation method for estimation of the circuit performance has been proposed. This additive method takes advantage of a 3-parameter probability density function (PDF), known as Burr distribution, to estimate the delay of each gate on the critical path. In this work, it is demonstrated that our proposed method is more accurate than previously proposed methods by taking into account the skewness of delay PDF. Although our proposed method is based on a 3-parameter PDF, we demonstrate that the simulation cost of our proposed method is no more than the conventional 2-parameter Gaussian PDF. We have compared the accuracy of our proposed method against the HSPICE simulation results. Moreover, we have compared the accuracy of our method with the most recent works with a 2-parameter PDF. The results for ISCAS85 benchmark circuits in our work have shown for 99 percentile points with average errors of 3.62, 3.49 and 2.78% in 90 nm, 45 nm and 22 nm technologies respectively. [ABSTRACT FROM AUTHOR]

Uncontrolled expansion of combustion wave in spark ignited internal combustion engine causes knock effect which seriously degrades efficiency and lifetime of the engine. Thus, accurate knock detection and control are essential for obtaining a desired performance from the engine. Usually, knock sensor is used to detect this phenomenon but it has limited accuracy especially at engine high-speed rotations because of natural vibration and external noises. In this study an effective method based on Non-Local Mean (NLM) algorithm has been proposed to improve the knock detection accuracy. This method is evaluated based on four different indicators and four engine cylinders. The results show 52.9 % improvement in knock detection. Also feasibility of real time execution of this method based on embedded hardware has been studied. [ABSTRACT FROM AUTHOR]