A structurally based approach to model order reduction using bond graphs

This paper introduces an approach to model order reduction that retains structural information in the reduced order model (ROM). Neither traditional methods such as aggregation nor modern techniques that employ H_∞ theory maintain a link between the structure of the original model and the ROM. In other words, the state variables and the coefficients of the ROM have no connection to the state variables of the original model. In contrast, the algorithm developed in this paper, which uses tree-structured transfer functions derived from bond graphs, maintains a subset of the original state variables in the reduced order model and maintains structural significance in the state variable coefficients. The advantage of the tree structure is that the ROMs are available directly from a single transfer function. The algorithm is outlined and the properties of the reduced models are indicated. The algorithm and three modern methods are applied to an eighth-order electromechanical system, and the absolute errors between the original and ROMs are compared