Sufficient conditions for the convergence are given for the second-order LTI systems, by which learning gains can be chosen.

• 依据收效性条件，学习增益的选取借系统矩阵的估计值，但在一定建模误差下，仍能保证算法的收效性。