An Innovations-Based Data-Driven Kalman Predictor for Predictive Control

TL;DR

This paper introduces a data-driven Kalman predictor that uses measured input-output data and innovations to improve predictive control without requiring offline disturbance measurements.

Contribution

It proposes a novel parametrization of the Kalman filter based solely on input-output data using the innovations form, eliminating the need for offline disturbance measurements.

Findings

Demonstrated effectiveness on a benchmark simulation.

Efficient estimation of the innovations process from data.

Improved predictive control performance without offline disturbance data.

Abstract

A recently developed data-driven Kalman filter requires offline measurement of the process disturbance; a requirement that is often unmet for many practical applications. We propose a solution that parametrizes the Kalman filter exclusively using measured input and output data. The key idea is to use the innovations form which naturally accounts for the process disturbance and measurement noise into a single orthogonal stochastic process. Unlike process disturbances, the innovations process can be estimated directly from input-output data via a numerically efficient projection step. The performance of the method is demonstrated using a benchmark simulation.