Detection of fault conditions in inverter fed ac machines

In modern industrial drive applications inverter fed ac machines are widely used because of their excellent dynamic performance. However, the inverter fed operation has also led to an increase of failures of the till now proverbial robust electrical machines. The components quite often affected by fault conditions are the bearings (50%) followed by the winding isolation (35%) and - in the case of induction machines - the rotor cage (10%). The breakdown of the winding isolation which has its reason in the voltage stress caused by the switching of the inverter, is one of the most critical faults and needs an immediate reaction to prevent further damage. Vibrations together with bearing currents lead to corrosion and fluting of the bearings and thermal stress and vibrations are usually responsible for faults in the rotor cage of induction machines. Though these faults usually develop relatively slowly, a significant reduction of maintenance costs can be achieved by applying monitoring methods. The topic of condition monitoring is most important from the scientific as well as industrial point of view, which can also be seen in the high number of publications in the past years. Most of the known methods are designed to work under line-fed, steady state operation and are based on Fourier transform, space phasor models, or special sensor signals like vibration or axial flux. However, there are only a few methods specialized to work under transient or even under inverter fed operation. To enable a reduction of preventive maintenance for the steadily increasing number of inverter fed drives in security critical applications it is necessary to monitor all possible sources of failure during steady-state as well as transient operation in a monitoring scheme able to work in the noisy environment of modern voltage source inverters. Goal of the proposed project is to investigate and develop a new approach to condition monitoring specialized to inverter fed operation and able to detect and distinct between fault conditions in the stator, bearings, and rotor during all ranges of operation. Core of the method is the transient electrical behavior of the machine. Instead of working with fundamental wave models of the machine, special sensors, or in the frequency domain, the step response of the phase current is exploited resulting from a transient excitation with voltage pulses applied by the inverter and the pulse width modulation (PWM). Hence a dependence on changing machine parameters, an interaction with speed/load dynamics like torque pulsation, or the disturbance of inverter ringing/dead time effects can be avoided. The proposed project is based on results obtained from a former project financed by the FWF.

In modern industrial drive applications inverter fed ac machines are widely used because of their excellent dynamic performance. However, the inverter fed operation has also led to an increase of failures of the till now proverbial robust electrical machines. The components quite often affected by fault conditions are the bearings (50%) followed by the winding isolation (35%) and - in the case of induction machines - the rotor cage (10%). The breakdown of the winding isolation which has its reason in the voltage stress caused by the switching of the inverter, is one of the most critical faults and needs an immediate reaction to prevent further damage. Vibrations together with bearing currents lead to corrosion and fluting of the bearings and thermal stress and vibrations are usually responsible for faults in the rotor cage of induction machines. Though these faults usually develop relatively slowly, a significant reduction of maintenance costs can be achieved by applying monitoring methods. The topic of condition monitoring is most important from the scientific as well as industrial point of view, which can also be seen in the high number of publications in the past years. Most of the known methods are designed to work under line-fed, steady state operation and are based on Fourier transform, space phasor models, or special sensor signals like vibration or axial flux. However, there are only a few methods specialized to work under transient or even under inverter fed operation. To enable a reduction of preventive maintenance for the steadily increasing number of inverter fed drives in security critical applications it is necessary to monitor all possible sources of failure during steady-state as well as transient operation in a monitoring scheme able to work in the noisy environment of modern voltage source inverters. Goal of the proposed project is to investigate and develop a new approach to condition monitoring specialized to inverter fed operation and able to detect and distinct between fault conditions in the stator, bearings, and rotor during all ranges of operation. Core of the method is the transient electrical behavior of the machine. Instead of working with fundamental wave models of the machine, special sensors, or in the frequency domain, the step response of the phase current is exploited resulting from a transient excitation with voltage pulses applied by the inverter and the pulse width modulation (PWM). Hence a dependence on changing machine parameters, an interaction with speed/load dynamics like torque pulsation, or the disturbance of inverter ringing/dead time effects can be avoided. The proposed project is based on results obtained from a former project financed by the FWF.