The Implementation Of Detection And Extraction Of Weak Signal In Spectrum Monitoring

With various types of electromagnetic interference increasing, spectrum monitoring and management become more and more difficult. Under the influence of noise, it is possible that some signals have the risk of being submerged to be a weak signal. Based on the real-time spectrum analyzer, this paper will study the method that solves the problem mainly caused by gaussian white noise in the detection and extraction of weak signals.In this paper, based on the compare of the characters and the working conditions of normal method by simulation experiments, we design the improved detection scheme and complete the developing and testing of the software that can realize the detection scheme.The main work of the thesis includes the following three points:1, the design for the scheme of multi-scale variable step size adaptive filter. This paper combine the multi-scale idea of wavelet transform and the scheme of variable step-size LMS algorithm. What’s more, in order to further improve the efficiency of algorithm, we adapt the parameters in the conventional step change formula so that the filter can adapt itself according to the present situation variety. Adaptive filtering is actually a process of continuous approximation to the signal, so it is more suitable for detecting a stable and non-abrupt signal.2, the design for the scheme of multi-band random resonance detection. Because of the shortcoming of the theory of adiabatic approximation which is only applicable for small frequency weak signal in stochastic resonance method, this paper further contracts the whole monitoring frequency band to the frequency band which can generate random resonance effect to realize the detection of frequency band on the basis of normalized scale transformation method. At the same time, for the case that random resonance can only detect the single frequency signal the wavelet transform is first used to realize the frequency band separation of the signal to be measured, and then the separated frequency band signals are respectively passed through the random resonance system to achieve the purpose of multi-frequency signal detection. The stochastic resonance scheme can be used for some weak signals with periodic characteristics, such as the vibration signals detected in the fault diagnosis of the device.3, based on the existing spectrum analyzer architecture, we design software program of achieving the conventional and improved algorithm to complete the weak signal detection function with the use of MFC class library, and complete the software testing.The results of the final software test confirm the feasibility of the algorithm program and prove that the detection software can proper function to achieve the desired goal.