4results about How to "Improve the phenomenon" patented technology

The invention relates to the technical field of current transformers, in particular to an anti-direct-current magnetic bias double-iron-core current transformer and a preparation and simulation verification method of the anti-direct-current magnetic bias double-iron-core current transformer. According to the preparation method of the anti-DC magnetic bias double-iron-core current transformer, the size parameters of the double iron cores are determined firstly, the first iron core and the second iron core are obtained according to the size parameters, the air gap size is determined according to the electrical performance index of the current transformer, and the air gap is formed in the preset position of the first iron core; a first coil is wound on a first iron core provided with an air gap, a second coil is wound on a second iron core, the first coil and the second coil are arranged in parallel, and the first iron core and the second iron core are stacked in the preset direction. The magnetic saturation phenomenon of the current transformer can be improved, the linearity difference of the current transformer is avoided, and the precision of the current transformer is greatly improved.

A burner and a gas stove. A gas distribution member (100) of the burner is provided with an air passage (103), a gas passage (102), and a gas homogenizing portion (101, 119). An outer ring burner cap (300) and an inner ring burner cap (200) are both provided on the gas distribution member (100). The gas mixing cavity of the inner ring burner cap (200) is in communication with the gas outlet of the gas passage (102). The gas mixing cavity of the outer ring burner cap (300) is in communication with the gas mixing cavity of the inner ring burner cap (200) through the air passage (103). The gas homogenizing portion (101, 119) is in communication with the gas passage (102) and the air passage (103). Gas is conveyed to the gas mixing cavity of the inner ring burner cap (200) through the gas passage (102), and is further conveyed to the gas mixing cavity of the outer ring burner cap (300) through the air passage (103). Since the gas distribution member (100) is provided with the gas homogenizing portion (101, 119) which is in communication with the gas passage (102) and the air passage (103), the gas inflow of the inner ring burner cap (200) and the gas inflow of the outer ring burner cap (300) can be balanced, thus the phenomenon of insufficient combustion can be mitigated, and CO emissions can be reduced.

The invention discloses a self-supervised speech enhancement method based on de-statistical instance normalization. The self-supervised speech enhancement method comprises a training stage and a testing stage. In the training stage, training corpora are obtained, a speech enhancement model is constructed and trained, parameters and a loss function required by the speech enhancement model are obtained, and enhancement of noisy speech is realized in the test stage. The speech enhancement model comprises a feature extraction module and a speech enhancement module, the feature extraction module selects a self-supervised learning model WavLM, and the speech enhancement module selects BLSTM as a main framework; de-statistical instance normalization is provided between the feature extraction module and the speech enhancement module, the problem of distribution offset generated by statistical attributes such as a mean value and a variance in sequence data along with time change is solved, and the accuracy of model prediction and the denoising effect are improved. According to the self-supervised speech enhancement method, the speech enhancement performance can be effectively improved, higher speech perception quality and intelligibility are obtained, and high-quality speech enhancement is realized.