Continuous wave mud pulse data transmission method based on continuous gradation frequency keying modulation and Convolution neural network demodulation.

Measurement While Drilling (MWD) technology plays a significant role in enhancing the geological steering and subsurface evaluation capabilities of extended-reach wells, challenging horizontal wells, and multilateral wells. With the increasing complexity of underground exploration, there is a heightened demand for the continuous wave mud pulse data transmission capacity. To address the inter-symbol interference(ISI) caused by the inherent inertia of the motor during high-speed data transmission using traditional modulation methods such as Frequency Shift Keying (FSK) and Phase Shift Keying (PSK), a novel approach has been proposed. This method employs Continuous Gradation Frequency Keying (CGFK) modulation combined with Convolution Neural Network (CNN) demodulation for continuous mud pulse data transmission. By controlling the waveform frequency to uniformly increase from zero to a predetermined value and then uniformly decrease back to zero within a symbol period, and utilizing the rate of frequency change as the feature for modulation and demodulation, this method effectively mitigates the issue of ISI caused by the motor's inability to rapidly switch to the next speed due to its inertia during symbol transitions. Simulation tests indicate that, compared to the traditional Matched Filter method, Support Vector Machine (SVM), Long Short-Term Memory (LSTM) networks, and CNN exhibit superior performance in recognizing CGFK, with CNN demonstrating the best results. Physical tests show that CGFK, particularly when assisted by CNN demodulation, possesses the capability to avoid or reduce ISI caused by motor inertia, and achieves favorable information transmission rates and bit error rate (BER) compared to traditional FSK and PSK.