Research On Key Technologies Of Low-frequency Signal Detection Based On SQUID

The magnetic anomaly signal is a typical low-frequency weak magnetic field signal.Through its analysis,information such as the position,speed,and material of the magnetic target can be obtained.Therefore,magnetic anomaly detection is an important means of anti-submarine warfare.With the continuous development of submarine degaussing and noise reduction technology,its magnetic anomaly signal is getting weaker and weaker,which can be detected by superconducting quantum interferometer(SQUID)with extremely high sensitivity.SQUID usually needs to be used in a magnetically shielded environment,but the high cost of a magnetically shielded room with excellent performance and a huge space occupation greatly limit the development and promotion of mobile magnetic anomaly detection.In a non-magnetic shielding environment,SQUID will be subject to a lot of noise interference.In order to obtain a high-quality signal,it needs to be denoised.This thesis mainly studies the key technologies in the signal postprocessing process of the SQUID-based magnetic anomaly detection system in the nonmagnetic shielding environment,suppresses the noise by designing a suitable denoising algorithm,and further studies the detection algorithm of the magnetic anomaly signal.Aim for higher signal-to-noise ratio and higher probability of detection.The main work of this thesis is as follows:(1)The magnetic anomaly signal and the noise in the SQUID detection process are modeled and analyzed.The magnetic target to be detected is modeled using the magnetic dipole model,the expression of the magnetic anomaly signal is obtained,and the time domain and frequency domain characteristics of the magnetic anomaly signal are summarized.Then,the noise encountered in the SQUID detection process was studied emphatically,and the environmental magnetic field noise and the internal noise of the SQUID device were analyzed and modeled respectively.Finally,the above-mentioned mixed noise is added to the magnetic anomaly signal,and the amplitude is normalized,and then the magnetic anomaly signal detected by the SQUID magnetometer is simulated.(2)In order to solve the problem that the magnetic anomaly signal detected by SQUID is submerged in the noise and difficult to extract in the non-magnetic shielding environment,this thesis proposes a joint denoising algorithm based on improved adaptive noise-assisted complete integration empirical mode decomposition(CEEMDAN)and adaptive wavelet entropy threshold.First,the signal is decomposed by CEEMDAN,and the selection of the intrinsic mode function(IMF)component is subjective,and the crosscorrelation coefficient index and energy index are used to distinguish the noise and signal components,and then the adaptive wavelet entropy threshold is used for the noise component Denoising,and finally reconstruct the denoised component and signal component to obtain the denoised signal.The simulation results show that the algorithm in this thesis improves the signal-to-noise ratio,and the waveform distortion is small.Under low signal-to-noise ratio,the algorithm in this thesis improves the signal-to-noise ratio by 1～3d B compared with other denoising algorithms,which verifies that the algorithm in this thesis has a good denoising effect.(3)In order to solve the problem of poor performance of existing detection algorithms under the condition of low SNR,a noise-enhanced detection algorithm based on neural network is proposed in this thesis.The algorithm trains the model by means of noise augmentation.This thesis focuses on the selection principle and addition method of noise.The results show that the algorithm in this thesis achieves high-accuracy detection under low signal-to-noise ratio,and achieves a detection probability of 90% at-6d B.And by comparing with other algorithms,the validity of the algorithm in this thesis is verified.