Figure 1
From: Discriminant non-stationary signal features’ clustering using hard and fuzzy cluster labeling
Discriminant feature clustering for a synthetic example. (a) Normal signal with a sampling frequency of 8 kHz is generated using Equation 1 where (α,σ,μ,a) for each component from 1 to 7 is as following: (1,0.005,0.15,2Π 1000), (1,0.04,0.20,2Π 1500), (1,0.04,0.20,2Π 2000), (1,0.02,0.30,2Π 3000), (1,0.05,0.35,2Π 3500), (1,0.04,0.40,2Π 2500), and (1,0.05,0.50,2Π 500). (b) Abnormal signal is created using the following parameters for each component: (1,0.001,0.15,2Π 1000), (1,0.04,0.20,2Π 1500), (1,0.04,0.20,2Π 2000), (1,0.02,0.30,2Π 3000), (1,0.001,0.35,2Π 3500), (1,0.04,0.40,2Π 2500), (1,0.001,0.50,2Π 500). (c) TF representation of the normal signal. (d) TF distribution of the abnormal signal. (e) Feature space. (f) Clusters representing abnormality and normality features.