2results about How to "Solve saturation" patented technology

The present application relates to a kind of time-domain electric source superconducting electromagnetic detection primary magnetic field compensation method and system, eliminate the influence of primary magnetic field on superconducting sensor observation, improve secondary magnetic field measurement accuracy.First, the primary magnetic field distribution of electric source emission current on ground is calculated, to design secondary field measuring line with primary magnetic field contour line;The compensation current of measuring line is calculated, and compensation transmitter and electric source transmitter are synchronized using GPS, and compensation current is emitted at measuring point to offset primary magnetic field, with the same characteristics of compensation current on the same contour line;By adjusting the load RL parameter of compensation system, the turn-off time of compensation current and electric source emission current is consistent, and the primary magnetic field generated by electric source emission current is completely offset.The present application can solve the problem that superconducting sensor loses lock and magnetic saturation due to the large change rate and range of primary magnetic field in the near-source area of electric source, realize effective observation of superconducting sensor in the near-source area, and expand the measurement area and detection area.

The invention discloses an electrocardiosignal classification method based on a dual-threshold neuromorphic reservoir, which comprises the following steps of: 1, receiving an electrocardiogram time sequence signal, and carrying out adjacent time sequence nonlinear coding operation on the electrocardiogram time sequence signal to generate an enhanced feature vector containing an original signal and a second-order interaction feature; step 2, inputting the enhanced feature vector into a dual-threshold neuromorphic reservoir, and generating a dual-channel pulse response sequence by using a dual-threshold pulse neuron to distinguish signal strength states; and step 3, constructing a joint feature vector according to the two-channel pulse response sequence, and inputting the joint feature vector into a classifier to obtain a classification result of the electrocardiosignals. According to the method, through cooperation of feature enhancement and a double-threshold mechanism, the problems that a traditional liquid state machine is poor in linear separability and neurons are prone to saturation are effectively solved, the number of the neurons is greatly reduced, and meanwhile the accuracy and the energy efficiency ratio of arrhythmia classification are improved.