Axial-flow pumps have complex hydraulic performance under diverse conditions, especially under low-flow conditions, which may occur rotational stall and brings serious system vibration. By combining experiments and numerical simulation analysis, the fine features of the local internal flow field in multi-flow rate conditions are accurately reflected, revealing the mechanism of changes in the internal flow field and hydraulic performance of axial-flow pumps under different flow conditions, especially at low-flow conditions. As the flow rate decreased, the backflow velocity in near wall area of the inlet pipe and circumferential swirl velocity increased gradually, and the tip leakage vortex (TLV) and pressure surface vortex also increased gradually. Based on the average axial velocity V0 under the design flow rate, the axial backflow velocity was −1.18 V0 and the circumferential velocity was −3.53 V0 under the deep stall condition. The maximum vorticity of TLV, secondary vortex and pressure surface vortex was 5 × 106 S−2, and the area of maximum intensity was the largest. The interaction between the secondary vortices and the mainstream intensifies the formation and development of passage vortex and blocks the inlet pipe. Under the influence of comprehensive factors, the entropy production in the inlet channel and impeller region reached the maximum under the deep stall condition, both of which were above 22 × 10−2 kg m2 s−3 K−1, more than twice that outside the saddle zone. This study analyzed the flow instability under biased conditions from multiple levels and perspectives, providing a basis for improving the hydraulic performance under diverse flow conditions in the future.

Runoff prediction is essential in water resource management, environmental protection, and agricultural development. Due to the large randomness, high non-stationarity, and low prediction accuracy of nonlinear effects of the traditional model, this study proposes a runoff prediction model based on the improved vector weighted average algorithm (INFO) to optimize the convolutional neural network (CNN)-bidirectional long short-term memory (Bi-LSTM)-Attention mechanism. First, the historical data are analyzed and normalized. Secondly, CNN combined with Attention is used to extract the depth local features of the input data and optimize the input weights of Bi-LSTM. Then, Bi-LSTM is used to study the time series feature depth analysis data from both positive and negative directions simultaneously. The INFO parameters are optimized to provide the optimal parameter guarantee for the CNN-Bi-LSTM-Attention model. Based on a hydrology station’s water level and flow data, the influence of three main models and two optimization algorithms on the prediction accuracy of the CNN-Bi-LSTM-Attention model is compared and analyzed. The results show that the fitting coefficient, R2, of the proposed model is 0.948, which is 7.91% and 3.38% higher than that of Bi-LSTM and CNN-Bi-LSTM, respectively. The R2 of the vector-weighted average optimization algorithm (INFO) optimization model is 0.993, which is 0.61% higher than that of the Bayesian optimization algorithm (BOA), indicating that the method adopted in this paper has more significant forecasting ability and can be used as a reliable tool for long-term runoff prediction.

Rectification optimization of the forebay in deflection-type inlet pumping stations mostly relies on a trial-and-error approach, which inevitably leads to resource wastage. To precisely improve the adverse flow patterns in the forebay, this study employs numerical simulations to reveal their formation mechanisms, establishes the intrinsic relationship between flow patterns and hydraulic characteristics, and proposes a targeted rectification measure—a “double-bottom sill + baffle plate” combination. The results indicate that five typical flow structures exist in the computational domain of the pumping station. Among them, the deflection mainstream, which has the most significant impact on the forebay, can be visually categorized into four zones: a concentrated-deflection zone, a gradient deflection zone, a low-speed zone, and a deflection–reflection zone. Helical flows and entrainment vortices predominantly occur in the concentrated-deflection zone, and deflection–reflection zone, localized at the forebay inlet, the degree change section, and the outlet, with intensity proportional to inflow condition. Helical flow intensity correlates with flow uniformity, while vortex strength links to local entropy generation. Energy loss at the forebay outlet shows a positive correlation with inflow condition. Post-rectification redistributes the flow velocity at the bottom sills and enhances the streamlined jet and destruct the vertical flow gradient, results show 36.36% higher flow uniformity, 8.14% less head loss, 1.79% lower turbulent kinetic energy, and 6.6% reduced dissipation rate. This study enhances the understanding of the distribution pattern of undesirable flow regimes in the forebay and provides practical guidance for analogous forebay design.

Since large-flow water diversion began in the middle route of the South-to-North Water Diversion Project, inverted siphons have experienced varying degrees of local flow pattern disorder at their inlets and outlets, resulting in a significant decline in hydraulic performance. Taking the Kuhe inverted siphon as a case study, a combination of numerical simulation and on-site testing was used to explore the causes of flow pattern disorder at the outlet of the inverted siphon. Meanwhile, based on the actual engineering situation, the influence of the flow pattern optimization measure of installing a 5D (five times the diameter of the pier) diversion pier at the outlet of the inverted siphon on its hydraulic characteristics was studied. Research findings indicated that before the implementation of flow pattern optimization measures, the Karman vortex street phenomenon was found to occur when water flowed through the piers; the interaction of the vortex streets behind each pier led to flow pattern disorder and affected the flow capacity. After implementation of the flow pattern optimization measures, the diversion piers had a significant inhibitory effect on the formation and development of the Karman vortex street behind the piers under the dispatching and design flow conditions. The flow velocities in each vertical layer were adjusted, with a significant improvement in the flow pattern. The hydraulic loss of the Kuhe inverted siphon was reduced by 11.5 mm, or approximately 7.8%. Under the dispatching flow condition, the water diversion flow of the Kuhe inverted siphon increased by approximately 4.11%. The water diversion capacity of the structure could be effectively enhanced by adding diversion piers to the tails of the piers. This method can be widely applied in similar open-channel long-distance water diversion projects.