Adaptive control of physiological systems with offset estimation

In many applications of control engineering in physiological systems, the output level corresponding to no input (i.e. the offset) is unknown and time-varying. Adaptive controllers that do not account for output offset may fail when presented with such systems. Furthermore, the standard least-squares technique for offset compensation during adaptive control may fail. To overcome these problems a floating identifier (FI) was developed. It removes DC components from the input/output data before model dynamics estimation. After estimation, the DC components are used to determine the offset. The FI is a shell around the parameter identification algorithm and can be used with most ARMA (autoregressive moving-average) type system identifiers. Computer simulations and animal experiments have been used to compare an adaptive controller coupled to the FI with the same controller coupled to two other standard least-squares identifiers. The animal experiments involved control of mean arterial pressure in the anesthetized dog by sodium nitroprusside. Phenylephrine is used as a disturbance.<>