In order to reduce the noise level of wheel loaders caused by gear pumps and realize environmentally sustainable development, the noise generation mechanism of a gear pump was studied, and the influence of flow pulsation and gear impact on noise was analyzed. In order to reduce the interference of other noise sources on the noise level of the gear pump, a noise test rig was established. The mixed noise signals obtained from the rig test were separated using the ICA model. The ICA model includes the following algorithms: The fast Fourier transform (FFT), independent component analysis (ICA) and inverse fast Fourier transform (IFFT). Some theories about the influence of the teeth number and teeth profile on noise were analyzed by theory and simulation. A noise reduction strategy was proposed by increasing the teeth numbers and modifying the teeth profile of the gear pump. The tests results showed that the noise values of the external and the driver’s ear of the wheel loader were reduced to 1 and 2.2 dB (A), respectively. This proves the effectiveness of the optimization scheme of gear pump noise reduction.