The invention relates to a tide parameter inversion-based tide predication method. A proper stability functional is constructed by introducing a regularization parameter to an objective functional through using a Tikhonov regularization technology against the ill-posed phenomenon generated in a variation assimilation method in order to improve the general variation assimilation method; and a regularization idea combined variation assimilation method is applied to a tide mode to optimize tide forecasting mode parameters in order to improve the tide mode forecasting precision, wherein two unknown parameters in the tide mode, comprising bottom friction coefficient and water depth, undergo combined inversion and synchronous optimization. Inversion parameters are obtained through solving the minimum value problem of the objective functional, so the optimal solution of the problem can be obtained, and the tide mode forecasting precision is effectively improved.