Uncertainty reduction through Active Disturbance Rejection

The theme of modern control is how to get around the unknowns, i.e. model uncertainties and disturbances, so that they do not degrade what is valued: stability and performance. That is, the unknowns are accepted as part of the system. Another option perhaps, proposed here, is to first make a frontal attack on the unknowns, to reduce their effects and then, only then, invoke the existing well-established methodology to deal with the remnants. In particular, it is shown that the amount of uncertainties can be reduced by way of active disturbance rejection, implemented in an inner loop to produce a well-behaved plant, which is then regulated by another controller in the outer loop. What's new here is a two degree of freedom design to deal with the unknowns: they are first actively estimated and rejected; then the remaining uncertainty, mostly in high frequency, is dealt with by, say, an Hinfin controller. The result is a hybrid Hinfin-active disturbance rejection control (H-ADRC) strategy. A motion control scenario is used to illustrate how the new approach could benefit problem-solving in the real world.