Search

Quality inspection in the industrial production field is experiencing a strong technological development that benefits from the combination of vision-based techniques with artificial intelligence algorithms. This paper initially addresses the problem of defect identification for circularly symmetric mechanical components, characterized by the presence of periodic elements. In the specific case of knurled washers, we compare the performances of a standard algorithm for the analysis of grey-scale image with a Deep Learning (DL) approach. The standard algorithm is based on the extraction of pseudo-signals derived from the conversion of the grey scale image of concentric annuli. In the DL approach, the component inspection is shifted from the entire sample to specific areas repeated along the object profile where the defect may occur. The standard algorithm provides better results in terms of accuracy and computational time with respect to the DL approach. Nevertheless, DL reaches accuracy higher than 99% when performance is evaluated targeting the identification of damaged teeth. The possibility of extending the methods and the results to other circularly symmetrical components is analyzed and discussed. [ABSTRACT FROM AUTHOR]

This paper describes a method for the uncertainty-based combination of signal processing techniques for the identification of rotor imbalance. The main idea of the proposed method is to compute the imbalance with different algorithms and to average the different algorithms’ results. The method is based on the data fusion at feature level and uses the measurement uncertainty of the imbalance as a figure of merit for the weight computation. A static, a dynamic, and a hybrid implementation are presented. In the static one, the weights are computed in a dedicated training phase, in which four algorithms (Fourier transform and quasi-harmonic fitting of signal denoised with Hilbert-Huang Transform, Hilbert Vibration decomposition, and Wavelet Packet decomposition) have been used to estimate the known imbalance of car wheels. In the dynamic one, the weights are computed at runtime by estimating the difference between each predictor and the actual signal. The hybrid approach is the combination of the two algorithms. Results of simulations and experiments evidenced the validity of the data fusion, with uncertainty reductions between 10 and 40%, with larger benefits in presence of non-stationary disturbances. [ABSTRACT FROM AUTHOR]

Abstract: This paper describes a method for the automatic identification of acoustic events using a weighted average of sound pressure and sound intensity measured at the vicinity of airports. The classification is based on the combination of different parameters using a technique conceptually similar to the sensor fusion: the indications of different classifiers are merged using the classification uncertainty as a figure of merit. The method uses the results of a training phase for the observation of statistical distributions of sound pressure and sound intensity related parameters. The different parameters’ weights are computed analyzing the overlap of probability distributions of takeoffs and landings, so that more relevance is given to the quantities presenting a low risk of misclassification. The proposed method does not require any arbitrary assumption about the parameter effectiveness, given that the indications of multiple (potentially infinite) classifiers can be merged together with weights that minimize the chance of misclassification. The method has been validated with measurements performed at the Milan Malpensa airport (Italy). Results outlined that the proposed classification criterion correctly identifies approximately 99% of events. [Copyright &y& Elsevier]