On Methodology And Application Of Weak Characteristic Signal Detection Based On Stochastic Resonance

Weak characteristic signal detection is one of the key techniques for condition monitoring and early fault detection of mechatronic systems. It is of great importance for guaranteeing the reliable running of large mechatronic equipment, and avoiding the occurrence of malignant accidents. With the faint characteristic signal, heavy environment noise, and real-time requirement of detection, the weak characteristic signal detection is very difficult. The researchers in this field always try their bests to achieve rapid detection of weak characteristic signal from heavy background noise.Stochastic resonance (SR) is a kind of nonlinear phenomenon, which can convert partial energy of noise into the energy of signal, and enhance the transmission of weak input signal through the nonlinear system. It can detect signal with lower signal-to-noise ratio (SNR) than linear methods. The model and algorithm of SR are rapid, efficient and can be realized through hardware. It is of great importance for utilizing SR in weak characteristic signal detection.Supported by the NSFC research project, On Methodology and Application of Weak Characteristic Signal Detection Based on Stochastic Resonance, this dissertation takes the detection of typical weak characteristic signals in mechatronic systems as the research object, and investigates the detection methods of weak characteristic signals on the basis of SR principle and the corresponding application. The theoretical foundations of SR and stability condition of the discrete SR model are analyzed at first in the dissertation. The influences of various parameters on the detection of faint periodic signal, weak aperiodic shock signal and weak digital pulse signal are numerical studied respectively. And the detection algorithms are proposed. The experiment system of SR model is built up by circuit's simulation. The algorithms and experiment system are applied to the detection of weak characteristic signals in mechatronic systems. The main points can be summarized as follows.1. The developing trends of weak characteristic signal detection of mechatronic systems are analyzed, and the major limits of existing weak characteristic signal detectionmethods are summarized. The advantage, development course and current research situation of the weak characteristic signal detection method based on SR are discussed in detail. The main problems of this method are pointed out.2. Adiabatic approximate theory and linear response theory of SR in bistable system are analyzed. The parameters of the polynomial potential well are optimized, and the stability condition of the discrete model of bistable system is put forward. The algorithm solving the discrete model is proposed. And the SR mechanism of the bistable system is verified.3. For detecting weak sinusoidal signal in heavy background noise, the detection strategy is analyzed and the meaning of improving the local SNR of the weak signal is pointed out. The weak signal detection model based on SR is compared with the best-linear filter. Through a kind of normalized scale transform, the frequency restriction condition of the signal to be detected is eliminated. The influences of various parameters on the detection performance are numerical studied and the detection algorithm is put forward. At last, the method is used for extracting the weak characteristic frequency component indicating the little crackle trouble in bearing inner circle from heavy background noise.4. For detecting weak aperiodic shock signal, the behavior and measurement description of aperiodic SR are discussed in detail. The interwell SR model is combined with the single-potential well approximation model of bistable system to detect the weak aperiodic shock signal. The influences of various parameters on the detection performance are numerical studied and the detection algorithm is put forward. And the method is applied to detect the weak shock signal in leak detection using sound pulse.5. For the reliable and accurate transmission of weak digital pulse signal, SR method is used to extract the weak digital pulse signal in heavy background noise. The signal enhanced-transmission model based on bistable system is discussed and the error bit ratio, channel information capacity is used as the measurement of digital signal transmission. The influences of system parameters on the performance of the signal transmission are numerical studied. The detection algorithm is proposed and the numerical experiment results show that the weak digital pulse transmission performance can be enhanced using SR method.6. The experiment system of SR is built up by circuit's simulation. The SR behavior of the system is theoretically analyzed and verified. The experiment results for detecting weak sinusoidal signal and weak digital pulse signal show that it can enhance the transmission of these two kinds of weak signals. At last, the experimental system is applied to the real-time pretreatment of raster strip signal in laser displacement measure system and the results show that it is fit for real-time signal processing occasions.In a summary, the detection of typical weak characteristic signals in mechatronic systems based on SR theory is investigated. The stability condition of the discrete model of bistable system is proposed. The frequency range of the detecting signal is extended through a kind of normalized scale transform. The detection algorithms for faint sinusoidal signal, weak aperiodic shock signal and weak digital pulse signal are put forward. The algorithms are applied to the fault feature extraction of rotor-bearing system and weak shock signal detection in leak detection using sound pulse. The experiment system of SR by circuit's simulation is built up. And the system is used in the detection of weak sinusoidal signal, in the enhanced-transmission of weak digital pulse signal and the pretreatment of raster stripe signal in laser displacement measure system. The SR method is shown to be a practical and fruitful approach for weak characteristic signal detection in mechatronic systems.