Self-adaptive compressed sensing method for compressed transmission of partial discharge signal

Abstract

The invention discloses a self-adaptive compressed sensing method for compressed transmission of a partial discharge signal, which comprises the following steps of: A, data preprocessing: converting alow signal-to-noise ratio signal into a high signal-to-noise ratio signal through wavelet filtering; B, training an over-complete dictionary after overlapping and partitioning the high signal-to-noise ratio, wherein atoms in the dictionary are only highly related to partial discharge pulses; C, judging whether the signal block contains a partial discharge pulse or not, if so, setting the label ofthe signal block as 1, otherwise, setting the label as 0; D, carrying out compressed sampling on the signal blocks with different labels; and E, performing sparse reconstruction on the compressed andsampled signal to obtain an original signal. According to the method, the defects in the prior art can be overcome, the compression ratio is increased, the on-site acquisition noise can be effectively filtered while the good recovery effect is ensured, and the signal transmission and storage cost is saved.

Description

technical field [0001] The invention relates to the technical field of abnormal detection of electrical equipment, in particular to an adaptive compressed sensing method for compressed transmission of partial discharge signals. Background technique [0002] Partial discharge signals can indicate abnormal conditions in electrical equipment, and prolonged partial discharges can lead to further degradation of insulation. High-voltage discharge will deteriorate the insulation material and greatly reduce the insulation performance of electrical equipment. Therefore, their monitoring is crucial. At present, the application of partial discharge monitoring has been applied to many electrical equipment, such as switchgear, GIS high-voltage switchgear, transformers, generators, buried cables, overhead lines, etc. The main monitoring methods are UHF method, TEV method, ultrasonic method , pulse current method, etc. Among them, the TEV monitoring method has attracted extensive attent...

AI Technical Summary

Problems solved by technology

The traditional compressed sensing algorithm is mainly a state filter algorithm, but the Kalman filter algorithm does not have strict mathematical verification, and only judges which elements in the support set need to be deleted from an empirical point of view, making the estimation result somewhat random

The algorithm requires the measurement matrix to strictly satisfy the RIP to accurately reconstruct the sparse signal, but many practical problems do not meet the above conditions, so the performance of the algorithm will become very poor, resulting in a large decision error

For the method without filter: Although the minimum l1 norm method is the most widely used sparse reconstruction algorithm, its global minimum point is not the real sparsest solution unless some strict constraints are met. In addition, when The correlation between the column vectors of the measurement matrix is ​​enhanced, and the reconstruction performance of the minimum l1 norm method will be severely degraded

Finally, the minimum l1-norm method is slower to reconstruct

The Bayesian algorithm also has limitations. This algorithm requires the support set of the time-varying sparse signal to be fixed, and only the amplitude of the component corresponding to the support set changes with time. This assumption does not meet the actual situation.

Images