Left-inversion soft-sensing method for a class of nonlinear DAE sub-systems

Original left-inversion soft-sensing method is only suitable for the nonlinear ODE (ordinary differential equations) system, and in this paper the method is expanded to be suitable for a class of nonlinear DAE (differential-algebraic equations) sub-systems. The expansion includes two aspects: (1) The type of the directly measurable variables and the to-be-estimated variables is expanded. For example, not only the directly measurable state variables but also the directly measurable algebraic variables could be used to realize the soft-sensing. (2) The type of equations is expanded. Besides the differential equations, the algebraic equations could also be used to construct the relations between the directly measurable variables and the to-be-estimated variables. Aiming at these two aspects, the left-inversion soft-sensing algorithm which is suitable for the nonlinear DAE sub-systems is proposed in the paper. Meanwhile, the soft-sensor of synchronous generator is designed based on the proposed method, and then some directly immeasurable variables in generator could be soft-sensed. The simulation results based on Matlab also demonstrate the validity of the proposed method.