Ball screw is one of the key transmission components of computer numerical control machine tools. In view of its easy to appear wear

pitting and other faults

leading to tool wear and machining accuracy reduction. Therefore

a fault location method of ball screw based on current time-domain synchronous average residual characteristic statistics is proposed

by collecting the motor current signal of ball screw and using time-domain synchronous average method based on phase to estimate deterministic signals. Then the fault-related signals in the residual signals were highlighted by removing the deterministic signals from the original signals. Finally

the characteristic values of the residual signals were extracted

and the fault detection was carried out with the kernel partial least square method. The fault identification and positioning of the ball screw were carried out.