Erqinhu Ke

With the advancement of intelligent industrial equipment and the growing demand for system digitalization, parametric modeling of pump operational states has become increasingly important. This is especially true for large pumps, where real-time monitoring remains a major challenge. This paper proposes a pump characteristic curve prediction method based on transfer learning. By leveraging characteristic curve data from small, easily testable source domains for pre-training, the learned features are transferred as initial conditions for training performance models of other pump types. The test results show that neural network models pre-trained with transfer learning achieve faster prediction speeds and lower error rates. Transfer learning also demonstrates strong adaptability to characteristic curve data from various pump categories. Under varying volumes of target data, prediction accuracy improves significantly. Notably, when data are limited, the transfer learning approach achieves a prediction error of 5.2%, compared to 48.1% for direct deep learning modeling. Moreover, the proposed method effectively reduces prediction range beyond the scope of the original data.

This paper discusses the energy-saving control method of a pipeline network system based on reinforcement learning and a genetic algorithm and compares it with traditional control methods such as constant-pressure control and non-frequency conversion control. The purpose is to improve the operational efficiency of an offshore oil and gas field water injection system. This paper simulates and verifies the experimental platform of a water injection system pipe network in offshore oil and gas fields and evaluates the optimization effect of different control strategies under different flow rates. The experimental results reveal that under a varying flow rate, the water injection system harnessing the GA and RL exhibits a remarkable energy-saving advantage over traditional control methods. Specifically, the GA strategy achieves an average energy-saving rate of 22.51%, with a maximum energy-saving rate of 38.14% under low flow rate, while the RL strategy attains an average energy-saving rate of 18.39%. These methodologies not only furnish novel solutions for the real-time optimal scheduling of water injection systems in offshore oil and gas fields but also proffer practical guidance, thereby paving the way for technological advancement and sustainable development in the industry.

The use of pumps as turbines has been gaining more and more attention in recent years. The present work mainly investigates the influence of blade wrap angle on the internal flow and pressure fluctuation characteristics of centrifugal pumps as turbines. Five different wrap angles (35°,45°, 55°, 65°, and 75°) for a forward-curved impeller were numerically analyzed under multiple operating conditions. The accuracy of numerical simulation was validated by experimental results. The results show that maximum efficiency is achieved with a blade wrap angle of 35°, and the highest efficiency flow point gradually decreases as the blade wrap angle increases. It is found by conducting entropy production theory analysis that the high-entropy production rate regions in PATs are concentrated in the volute tongue and impeller blade inlet regions, and that the entropy production rate at the impeller inlet region increases and then decreases as the blade wrap angle decreases. In addition, pressure pulsation was affected not only by dynamic and static interference but also by an irregular vortex around the impeller; its magnitude under Q_t is higher than 0.8Q_t for blade wrap angles of 55° and 75°. The primary frequency of pressure pulsation within the impeller is the axial frequency fn and its multiples, and the frequency with the largest amplitude is 3fn. The periodicity of vortices is closely related to the periodicity of pressure pulsation. And it is suggested that a PAT with a 35° blade wrap angle is advantageous for improving the stability of a turbine.