Research on soft measurement algorithm of torque measurement in transmission system

A dynamic torque testing method based on computer system is put forward in this article, which adopts non-contact measurement and associates with virtual instrument technology to do high speed sampling of two-way sine signals to acquire two sets of discrete data. Cross-correlation principle is applied to treat these data to solve phase spectrum of cross correlation function. Finally torsion angle can be gotten to realize measurement of torque by acquisition of phase difference of measurand signals. This measuring method emphasize software algorithm, adopt information processing technique and simplify complex hardware circuit. Measurement result is more accurate by application of software algorithm to eliminate influence of temperature drift, acoustic noise and interfering signal. The key problem of using correlation method to measure phase lies with complete alternation sampling. When signal frequency changes, value of sample frequency should be rectified timely. For this reason, a late-model generalized predictive algorithm is inducted, which is generalized predictive adapting optimization algorithm with time-varying forgetting factor recursive least squares - recursive extended least squares (RLS - RELS) performance criteria weighted. By on-line recursion to estimate parameters, the algorithm speeds up the computing speed of system, avoid matrix calculation, increase efficiency and accuracy and enhance robustness. Frequency tracing based on this algorithm can realize whole period sampling of effective signal.