马达作为液压系统的主要执行装置，其巨大的噪声辐射已难以满足低音化的要求；由于当前马达噪声主要来源不明确以及噪声定位精度较低等问题，致使马达降噪效果始终不理想。因此，为明确马达噪声的主要来源，进一步提升马达噪声的定位精度，采用Pumplinx软件建立马达的流体仿真模型，得到马达配流盘处流体振动力变化信息；通过ADAMS和AMESim软件联合仿真，得到马达运动时柱塞撞击缸体所引起的振动力变化信息；结合有限元瞬态动力学分析方法，采用ANSYS中瞬态分析方法得到马达壳体和后端盖表面的振动位移响应；以ANSYS中得到的马达壳体和后端盖表面的振动信息作为声学边界条件，结合边界元分析方法，在LMS Virtual.Lab中仿真得到马达的声场，明确马达噪声的主要来源及噪声的主要发生区域；其次，设计马达声强噪声测试试验台，获得马达声强变化云图，验证多物理联合仿真结果的准确性；然后综合考虑观测矩阵、稀疏表达以及重构算法的关系，基于正则化正交匹配追踪重构算法（ROMP）明确马达主要噪声的定位区域；最终再次通过声强测试试验台验证优化后重构发算法提升定位精度的可行性。结果表明，马达模型多物理场仿真的结果是正确的，噪声主要来源于配流盘处的压力冲击和柱塞的碰撞，主要噪声区域分布于配流盘处，新的主要定位精度达到25 mm，实现了对马达主要噪声来源的确定和定位精度的提升。

As the main actuator in hydraulic systems, motors generate significant noise radiation that increasingly fails to meet low-noise requirements. Due to unresolved issues such as unclear primary noise sources and low loca-lization accuracy, noise reduction in motors remains challenging. Therefore, to identify the main noise sources and improve localization accuracy, this study employed multiple approaches: A fluid simulation model of the motor was established using Pumplinx software to obtain the variation of fluid excitation forces at the motor’s port plate. A co-simulation using ADAMS and AMESim was conducted to acquire the variation of excitation forces caused by pistons impacting the cylinder block during motor operation. Combined with transient finite element analysis, the transient analysis method in ANSYS was used to obtain the vibration displacement response on the surfaces of the motor housing and rear end cover. Using this vibration data from ANSYS as acoustic boundary conditions, a boundary element analysis was performed in LMS Virtual Lab to simulate the motor’s acoustic field, thereby identifying the main noise sources and primary noise generation areas. Subsequently, a motor sound intensity noise test bench was designed to obtain sound intensity variation cloud maps, verifying the accuracy of the multi-physics simulation results. Then, considering the relationships among the observation matrix, sparse representation, and reconstruction algorithm, the regularized orthogonal matching pursuit reconstruction algorithm was adopted to determine the loca-lization areas of the main motor noise. Finally, the feasibility of the optimized reconstruction algorithm in improving loca-lization accuracy was verified with the sound intensity test bench. The results show that the multi-physics field simulation of the motor model is correct, the main noise sources are the pressure impact at the valve plate and piston collisions, the main noise area is distributed around the valve plate, and the new localization accuracy reaches 25 mm, achieving the determination of the main motor noise sources and an improvement in localization accuracy.