Research On Weak Signal Detection Based On SR And LMD

With the improvement of enterprise management level, equipment health management arises at the historic moment, and gradually got the attention. To ensure the health condition of equipment, monitoring and fault diagnosis technology is indispensable. Prompt detection and elimination of early weak fault of mechanical equipment can prevent the further development of fault and avoid the occurrence of major accidents effectively. Therefore, condition monitoring and fault diagnosis technology for the industrial production is significant and far-reaching. This paper takes rolling bearing as the experiment object, combined with noise reduction based on stochastic resonance method and LMD decomposition method, focuses on the research of weak signal detection technique for early failure of mechanical equipment.Stochastic resonance is a characteristic noise reduction method. Insteading of suppressing noise, the method uses the high frequency noise to enhance energy of low frequency weak signal. However, in the engineering application, the selection of structure parameters of the stochastic resonance bistable system and calculation of step size is very difficult. To solve this problem, a stochastic resonance theory based on GAPSO method is proposed. GAPSO is a kind of particle swarm optimization algorithm which mixtures genetic thought. It integrates the advantages of particle swarm optimization algorithm and genetic algorithm, which achieves a fast convergence, a local search ability, and synchronization optimization of structural parameters and computing step size. With simulation analysis and engineering application, it is proved that the convergence speed of this method is fast and the method is simple. There will be a promising application prospect of weak signal extraction in incipient faults.When the signal is influenced by strong noise, the method only through a single stochastic resonance cannot achieve desired effects in noise reduction. Therefore a cascaded bistable stochastic resonance method will be applied. However, for the practical engineering processing signal, there still lays a problem that the parameters of cascaded bistable stochastic resonance system cannot be selected adaptively. In order to solve this problem, a GAPSO method will be used in the cascaded bistable stochastic resonance. For the problem that LMD decomposition effect affected by the noise, this paper presents a cascaded bistable adaptive stochastic resonance noise reduction method to improve the LMD decomposition. This method makes full use of the unique noise reduction advantage of cascaded bistable adaptive stochastic resonance, significantly improves the quality of LMD decomposition. Both simulation experiment and engineering application prove that this method can effectively extract weak signal in the early fault.After successful theory research, an experiment is applied. Taking key components of 6 axes NC machining center as the object of study, the network based independent monitoring unit and embedded online monitoring unit are constructed.