The increasing proportion of inverter-based resources (IBRs) in power grid has introduced novel characteristics to short-circuit current calculation method

and causes convergence problem in conventional algorithms

needing new development of computational approaches. This study analyzes the fault equivalent models under typical low voltage ride-through control strategies for IBRs. These models exhibit control strategy switching transients and nonlinear current-limiting characteristics

which contribute to non-smooth features in the equations of short-circuit current calculation. Then

mathematical expressions for the short-circuit current calculation that accounts for nonlinear and non-smooth behaviors are established. Analysis demonstrates that the convergence of fixed-point algorithm is poor

and Newton-Raphson algorithm may exhibit numerical oscillations when solving these equations. The non-smooth feature of control strategy switching is identified as the primary cause of convergence failures. A new Trust Region Gradient updated with quasi-Newton technique algorithm (TRGN) is proposed

which uses quasi-Newton intercept approximation instead of gradient and combines trust region to solve non-smooth equations. This algorithm can ensure global convergence of short-circuit current calculation in high proportion new energy grids

and its specific application method is given. Finally

based on the IEEE118 example

the proposed algorithm is proved to have better performance in solving short-circuit current calculation with non-smooth characteristics by comparing with Newton-Raphson method and adaptive Levenberg-Marquardt algorithm.