Blind source separation method for single-channel vibration signals
A technology of blind source separation and vibration signal, applied in measuring devices, instruments, measuring ultrasonic/sonic/infrasonic waves, etc., can solve the problems of difficulty in selecting mother wavelets, poor blind source separation effect, ignoring source signal structure, etc., to avoid signal Sparsity constraints, good separation, physical well-defined effects
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0047] Consider three mechanical vibration source signals:
[0048] the s 1 (t)=5sin(2πf 1 t+5)
[0049] the s 2 (t)=2sin(2πf 2 t+10)
[0050] the s 3 (t)=8cos(2πf 3 t)
[0051] In the formula, s 1 (t) is the first vibration source signal, s 2 (t) is the second vibration source signal, s 3 (t) is the third vibration source signal, f 1 = 80Hz, f 2 = 25Hz, f 3 =150Hz, sampling frequency 1000Hz, sampling points 2048, optional one-dimensional vector A 1 =[2.2895, 6.4194, 4.8448], construct single-channel observation signal
[0052] x 1 =A 1 ×[s 1 ,s 2 ,s 3 ] T (6)
[0053] to x 1 Perform EMD decomposition to get three IMF components, x limf =[c 1 , c 2 , c 3 ] T . Randomly select a mixing matrix A, perform signal reconstruction according to formula (7), and generate two mixed observation signals x 2 , x 3 .
[0054] A = 1.53 1.52 ...
Embodiment 2
[0062] The invention is used to blindly separate the single-channel vibration signal of a cab of a faulty vehicle, extract fault features, and analyze fault causes. The fault feature of the car is that the cab bumps up and down at a speed of 30km / h-60km / h. An ICP acceleration sensor is installed on the bottom plate of the cab to collect test data, and the sampling frequency is 5KHz. When sampling, the car was driving on the B-grade road at a speed of about 50Km / h. Acquired single-channel acceleration time-domain signals such as Image 6 shown.
[0063] Time-delay autocorrelation noise reduction processing is performed on the single-channel vibration observation signal to eliminate the interference of noise signals. The denoised signal is decomposed by EMD to obtain 11 IMF components, and a 2×11 matrix A is selected to reconstruct two new vibration acceleration observation signals through the IMF components.
[0064] A = ...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 