During oil and gas drilling, the accuracy of intelligent early kick recognition models relies on their detection of abnormal variation trends in monitoring parameters. These variation trends are contained in the data sequences of monitoring parameters corresponding to early kick samples. Therefore, accurate identification of the abnormal variation trends of monitoring parameters in these samples is of great importance for early kick monitoring. This understanding can serve as a prior knowledge to guide the learning process of intelligent models. Since intelligent models use loss values calculated from their loss functions to determine which samples should be focused on during the model training process, the prior knowledge can be used to modify their loss functions. First, a knowledge-guided attention allocation network is proposed, which allocates attention weights to monitoring parameters and enables the intelligent model to pay greater attention to these critical ones. Then, a knowledge-guided bidirectional long short-term memory (KG-Bi-LSTM) model was established by embedding prior knowledge into its binary cross-entropy loss function. Firstly, in order to reduce the model's attention on the most easily identifiable samples corresponding to normal drilling, the square of the model's prediction error for each normal sample is used as an attenuation factor for loss values corresponding to normal samples. In this way, the model can learn more from kick samples. Secondly, to increase the model's focus on learning the variation trends of monitoring parameters corresponding to the early kicks, a monotonically decreasing exponential function is employed to weight the loss term for kick samples. This approach assigns greater weights to samples from the early kicks, thereby preventing underfitting by emphasizing key trend recognition. Subsequently, to prevent the model from overfitting due to the weighting of loss values, the sum of squares of the network weights is added to the loss function as a constraint term. Thus, a balance between model overfitting and underfitting is achieved through knowledge guidance and weight constraint. Thereafter, a knowledge-guided loss function is obtained by combining the above weighted loss terms and the constraint term. Finally, this loss function is applied to the Bi-LSTM-based kick recognition model to produce a KG-Bi-LSTM version. Experiments were conducted using 4,599 samples from field data, the results show that compared with the prevailing Bi-LSTM model, the proposed KG-Bi-LSTM has a 5.28% increase in recognition accuracy and an earlier alarm time of nearly 2 min.
- Article type
- Year
Open Access
Original Paper
Issue
Radio frequency (RF) signal editing and sampling rate conversion are critical techniques employed in communication and radar systems for the extraction of targeted signals and collaborative processing of multiple standard waveforms. The existing international business tools (such as BIRD’s RF-Editor, and the vector signal analyzers from Rohde & Schwarz and Agilent) offer functions for signal editing and sampling rate conversion; however, they fail to address the urgent demand for cost-efficient and highly adaptive solutions due to economic and technological limitations, particularly in frequency hopping communication systems and multitarget radar tracking systems.
To address these issues, an integrated RF signal editing and sampling rate conversion system was developed herein on the Qt platform. A filter design method based on an improved Kaiser window function was used in the RF signal editing module. This method resolved the trade-off between performance and complexity inherent in traditional Kaiser window designs. A cost function was defined to quantitatively evaluate the performance of the Kaiser window function，which usede the normalized transition bandwidth and minimum stopband attenuation as the performance metrics. This function incorporated a weighting factor, enabling designers to dynamically prioritize transition bandwidth and stopband attenuation based on specific application requirements. An efficient golden ratio search algorithm was employed to optimize the shape parameter of the Kaiser window with minimal computational overhead. It operated within a well-defined initial search range and iteratively refined the solution interval based on objective function evaluations until it satisfied the predefined convergence criteria. The optimized Kaiser window was used to truncate the ideal infinite impulse response and obtain a finite impulse response low-pass filter with a customizable transition bandwidth and stopband attenuation. In addition, direct modulation technology was used for precise spectrum shifting. A sampling rate conversion method that integrated local interpolation and Sinc-windowed function truncation was used in the sampling rate conversion module. The Hanning window smoothed the edges of the Sinc function, whereas the local interpolation strategy considerably reduced the computational complexity; these improvements facilitated rapid conversion across arbitrary sampling rate multiples. By integrating these methodologies and adopting a hierarchical modular design paradigm, a Qt-based RF signal editing and sampling rate conversion system was developed.
Experiments conducted using measured data revealed that 1) the designed filter exhibited excellent frequency-selective characteristics and effectively attenuated signals outside the specified frequency band; 2) the frequency shift function achieved a relative error within 0.8%, satisfying the engineering tolerance requirements for broadband frequency modulation communication systems; and 3) sampling rate conversion introduced minimal signal distortion, with a time-domain mean square error (MSE) of <0.01 and frequency-domain logarithmic spectral distance of <1 dB. Notably, the computational efficiency was enhanced by over 70% compared with that of traditional Sinc interpolation with an upsampling factor of 2.5.
The proposed system could efficiently and flexibly perform RF signal editing and sampling rate conversion with fast response and stable operation. Compared with the existing commercial software solutions, it eliminated the need for costly licensing fees and specialized hardware dependencies, supported cross-platform deployment on both Windows and Linux, and featured a modular architecture that enabled independent function calling and expansion. These characteristics made it more flexible than closed-source commercial software.
京公网安备11010802044758号