Diagnostics of Robot Drives Based on DC Motors by Identifiability Criterion of Nonlinear Discrete Model in State Space
Nikitin, Yu. R., Trefilov, S. A.
Article language: English
Abstract. The paper studies robot drives by the identifiability criterion based on a discrete digital control model. Criteria of observability, controllability and identifiability of drives as a function of the rank of an extended state matrix with a measurement matrix are considered, in which the relative errors of the information-measuring system are analytically taken into account. An algorithm is proposed for calculating the identifiability criterion for a nonlinear control system in a discrete linearization version. It is proposed to use identification in terms of the correspondence of the mathematical model to the results of the operation of the object. At each step, the determinant of the extended matrix is calculated, which is compared with a constant that numerically divides the space of the state matrices. Thus, the operation of the drives itself makes it possible to determine its identifiability. As a criterion for the optimality of the identification algorithm, a decision-making optimality criterion is chosen in combination with an identifiability criterion for an optimal control algorithm by the criterion of minimum quadratic form. The vector-matrix model of drives in the state space is presented taking into account the relative accuracy of measuring the state of the information-measuring subsystem of drives. It is proposed for practical problems to determine the identifiability criterion by modeling the state matrix for cases when the state matrix parameters exit the space of realizable parameters of serviceable drives. The research results obtained can be used to build diagnostic systems for robot drives.
Keywords: diagnostics, identification, state space, velocity control, DC motors, robots, modeling
Pages: 24–31Total pages: 8
Funding, support: The reported study was funded by RFBR according to the research project No. 18-08-00772 A.
Year of publication: 2020