On-line parameter estimation for improved sensorless control of synchronous motors

An on-line synchronous motor parameter estimation technique is introduced. The motor is driven sensor-lessly with a Kalman filter. A second Kalman filter, that works in parallel with the main one, estimates the motor model parameters. Thanks to the parameter estimation, the accuracy of the Kalman filter estimations of position and speed increase and thus the global performance of the drive. Theoretical results are confirmed by simulations and by experimental application on an industrial drive. The final goal is to drive a brushless DC (BLDC) motor for a high-dynamic industrial application without any need of position sensors, off-line calibration, a first-use calibration time nor predefinite start-up sequence.

I. Nomenclature

Phase resistance

Phase inductance

Torque constant

Friction constant

Pole pairs

Inertia

Sampling period

Phase current

Phase voltage

Rotor speed

Rotor position

Torque

LHS

Left hand side

RHS

Right hand side

State variable:

State vector:

or

Parameter:

Parameter vector:

or

where denotes the instant, in the discrete time, at which the variable, parameter or vector are considered. The estimated variable, parameter or vector are indicated by an hat, like: and . Matrices are indicated with an uppercase bold letter like .