Abstract: Abstract: In the combined sluice-pump station project, adverse flow pattern always happens in the forebay, which will result in the unfavorable inflow pattern on the inlet section of the passage intake. Based on this, the physical experiment device and computational domain are prepared separately, which share the same dimensions and parts containing a diversion canal, passage intake, forebay, diversion pier and, check gate, so that the flow in the forebay is predicted and verified by the test data at the model water level which is 0.125m, corresponding to the prototype water level of 2.5m. The velocities of the measure points on the chosen sections in the forebay is measured with a LGY-Ⅱ intelligent velocity instrument and recorded. To gain the steady flow in the test rig, the flow rate was adjusted depending on the valve and electromagnetic flowmeter. In order to better understand the inner detail flow structure in the forebay, several analysis planes which are the surface plane, the bottom plane, the inlet of passage intakes 1#, 4#, and 5#, and the plane before the passage intake are sliced. In the original case, the crosswise velocity of the main forward flow around the diversion pier is large which leads to a deflective flow. With the effect of viscous force, the neighboring water changes its direction, and brings out a larger scale bias flow. Then the water changes its direction again, due to the low pressure zone, and across flow occurs once more whose direction is opposite to the initial cross flow. At last, the water moves along the diversion pier to the inlet of forebay, where a mixed flow takes place and the steady large circulation zone starts from the isolated pier to the cross-section which is close to the inlet of the passage intake in the forebay is produced. The axial velocity distribution is uniform and left-right & up-down symmetry on the inlet of passage intake 1#, the high velocity area is in the center of the section. The axial velocity distribution is not uniform on the inlet of passage intake 5#, and the high velocity area is on the left of the inlet. To solve the problem, a single long diversion pier and the combined measures of "long diversion pier + short diversion pier" were put forward which were marked as improvement scheme 1 and 2. Compared with the original case, the recirculation zone was decreased and advanced with both measures, and the water could be smoothly guided into passage intake 4#. The axial velocity distribution on the inlet of passage intake with both rectification measures was well improved. For passage intake 1#, 4#, and 5#, axial velocity distribution increased 2.6%, 7.8%, and3.6% respectively with improvement scheme 1, as well as 2.7%, 18%, and 1.8% with improvement scheme 2. As a whole, rectification results with improvement scheme 2 were better than improvement scheme 1, but the structure of improvement scheme 1 is simple and convenient for construction. By discussing the calculated and test average axial velocity, the maximum and minimum relative error was 6.29% and 0.63% respectively. After a single long diversion pier was installed, the standard deviation of calculated and test axial velocities decreased. The calculated results were well satisfied with test results and the prediction well verified and credible.