The Research Of Blind Signal Processing Method Based On Automobile Brake Noise

Due to the influence of brake structure and the change of the braking condition such factors, the car can produce different levels of noise in braking process. It will not only affect the healthy living of the residents, more serious situation can reduce the performance of the braking system and cause safety accidents, therefore effective governing automobile brake noise will become very important. When mechanical equipment is abnormal at work, it can produce different vibrations and noise of the corresponding, so engineer can explore the sources of noise and control the brake noise through analyzing the brake noise. It is important to the study of brake noise.The main content of this academic dissertation is shown below.The main content in chapter 2 is that performance structure and working process of two kinds of brake are expounded, and the mechanism of noise and vibration are explained in detail. First of all, this paper introduces the basic mathematical model, signal preprocessing technology and evaluation standard of separation effect of Blind Signal Processing in the theoretical research part. Secondly theoretical knowledge of several typical instantaneous blind separation algorithms is studied in detail; finally FastICA algorithm and EASI algorithm is realized by using a computer programming and compared the separation effect of the two methods through the simulation analysis. Observed signals are usually complex signal in real environment, it consists of engine noise tire noise and air flow noise. The mixing process is called the convolution mixture in mathematics. This paper takes blind deconvolution algorithm based on frequency domain to realize the signal separation. Its main idea is to convert convolution mixture of time domain to instantaneous mixing of each frequency point by using the short-time Fourier transform. Then adopting blind source separation algorithm of the complex domain achieves signal separation. Finally using the reverse short-time Fourier transform recovers time domain signals. This article adopts normalization of separation matrix norm and improved KL distance method to solve the permutation of ICA for each frequency point. In the end the author respectively verify the operation effect of complex domain blind source separation algorithm and frequency domain blind deconvolution algorithm by simulation.Finally the author completes the development of brake abnormal sound analysis system by using lab VIEW and MATLAB software and tests the effectiveness of the system through the experiment. Data exchange of two kinds of software is realized by using DLL technology.