Adaptive control of servohydraulic materials-testing machines: a comparison between black- and grey-box models

Abstract

In a materials-testing machine a specimen is subjected to actuation such that is load force, position or extension follows some prescribed cyclic pattern. Good testing relies on tight and consistent closed-loop control — yet the plant's dynamic behaviour is a function of the test-piece's stiffness, which is probably unknown a priori and which may vary during the test: hence adaptation is required for acceptable performance. As a basis for the design, a ninth-order physical model is successively reduced to a first-order transfer function whose parameters depend on the specimen stiffness, which can be deduced using on-line data. Comparisons are made between a black-box MRAC design using the ‘minimum controller synthesis’ method and a grey-box approach which adaptively modifies PI controller settings based on estimated stiffness. The final practical design is evaluated using a range of test specimens, showing that the grey-box adaptive controller is highly effective for industrial use.