Application of Pattern Recognition for the Mitigation of Systematic Errors in an Optical Incremental Encoder

The accuracy and reliability in the measurement of the instantaneous angular speed (IAS) by incremental encoders are fundamental for speed and torque control systems. An error in the acquisition of this variable may cause instabilities or even render the system unfeasible. Many works deal with this problem, which in some cases is caused by the imperfections in the disk manufacturing or also shaft misalignments. Due to the systematic character of these type of error, in this work an error pattern in the encoder signal of a vehicle dynamometer was identified by means of a calibration process and detected along the experiment by minimizing the total variation of the corrected angular speed curve. By making use of this procedure, not only the measurement became more accurate, but also the direction of rotation and angular position of the rotary system could be detected with only one channel of the encoder sensor being acquired. After the application of this method, the measurement error was reduced by 84.22%, even in transient conditions, without the application of further filtering methods, which may unwantedly lead to phase shifting in the IAS signal.