Accurate model identification of the inertial mass dynamic of hydraulic cylinder with model uncertainty

Abstract

In this study, a novel double-layered network scheme is proposed for accurate and efficient identification of hydraulic cylinder. The model of real hydraulic cylinder is usually partly known. If gray-box methods or black-box methods are used alone for this kind of system, parameter accuracy, simulation accuracy, the conciseness of identified result and the ease of implementation usually cannot be satisfied at the same time. The proposed double-layered network combines the advantages of gray-box and black-box methods. Accurate parameters of known model are obtained by random sample consensus algorithm in the first layer, even with modeling error. Thus, the value of unknown model can be obtained. Then the unknown model is identified by a simple black-box method in the second layer. Therefore, complete model can be obtained precisely and concisely. The overall method is easy to implement and the obtained physical parameters and model can help the controller design in industries. The proposed method is verified through simulation and real system tests, compared with several existing identification methods. Remarkable results are achieved by double-layered network in the identifications of inertial mass dynamic of hydraulic cylinder.