Abstract: Due to the problem of multi-motion mode coupling of industrial robots,the traditional health monitoring scheme needs to install sensors at each joint separately,which is difficult to meet the needs of actual industrial field.A 6-joints industrial robot is taken as the research object,and the health monitoring method of industrial robot in multi-motion mode switching scenario is studied based on vibration signals.Firstly,robot motion modes are divided by jump point algorithm,and signal intervals corresponding to different motion modes are obtained.Secondly,the monitoring indexes are extracted from the signals of different motion modes.Finally,the health monitoring of different joints of industrial robots is realized based on the control chart method.The proposed method is verified by the experimental data of industrial robot degradation.It shows that the proposed method can realize the health status monitoring of 6 joints of the robot with only two vibration sensors.The motion pattern partitioning algorithm proposed in this paper costs shorter running time,has higher single precision and better repeatability when analyzing a large number of historical degraded data.The proposed method can effectively avoid the misjudgment of monitoring results which may be caused by the difference of motion mode coupling and sampling signal under the condition using only a few sensors,making the monitoring results more accurate and reliable,and more suitable for the health status monitoring of industrial robots in actual industrial sites.