IELTS Band 7 Complete Prep Course
Click Here ===> https://ssurll.com/2te2Qm
The proposed method was applied and validated to diagnose the gear fault as shown in Figure 4. The FFT-based frequency domain features (kurtosis, skewness, and crest factor) were extracted to classify the fault in each gear and the results were stored in a database. Then the proposed ECG-labeling was used to normalize the feature space of the data set and various ECG-based classifiers: SVM, k-NN, PLS-DA, and MLP were implemented for the fault categories. The results showed that the MLP-driven weighted combination classifier improved the classification accuracy making it 100% accurate for identifying the gear fault in the data set.
However, some other methods are more effective in representing the fault conditions. Pandey et al. [60] used the FFT-based Pulse Width (or ressentment) and the crest factor features to diagnose the bearing fault in the drive-train circuit of an aircraft which is used to lift a helicopter. More features were considered for the fault state classification, and the features of ridge line slope feature, the second and the higher order of harmonic and the first two orders of harmonics were extracted. The ridge line slope feature hence became the most effective feature in representing the fault condition in the data set for diagnosing the gear fault. Sectorman et al. [68] used the cone-like TAEF and the hertzian crest factor features to diagnose issues of machine parts. TAEF is calculated by measuring the peak-to-peak amplitude over a smoothed envelope of the vibration signal [67]. As an alternative to the cone-like TSFE, Sectorman et al. [68] used the hertzian crest factor feature. Sectorman et al. [67] have concluded that i) both cone like feature and hertzian crest factor are not suitable features for identifying the fault condition, ii) it is not a good idea to measure TAEF before smoothing peaks, as it will reduce the TAEF and iii) the hertzian feature is good for identifying the fault types in the discretized band samples, whereas cone like feature is suitable only for identifying the fault types in the continuous waveform. d2c66b5586