Abstract: Strong tracking filtering is an adaptive estimation method that can effectively deal with state mutation and model inaccuracy in dynamic systems. The traditional strong tracking filtering uses the fading factors to adjust the relevant parts of the state transition matrix of the prediction and estimation error covariance formula, resulting in the weak interpretability of the filter model correction and model parameter estimation, and the estimation performance needs to be further improved. To solve the above problems, based on the existing deep analysis that the adjustment effect of strong tracking fading factors is equivalent to the adjustment of process noise covariance, a strong tracking filtering idea that directly uses fading factors to adapt to the process noise covariance part of the dynamic adjustment prediction estimation error covariance calculation formula is proposed, and two kinds of multiple suboptimal fading factor calculation methods that can effectively guarantee the symmetry of covariance matrix are given, direct matrix solution method and matrix trace solution. Compared with the traditional strong tracking filtering method, the new method directly realizes the real-time dynamic feedback correction of the system model parameters under the existing framework, which can not only avoid the filtering divergence caused by asymmetry but also has a very clear principle and effect interpretability. The experimental results show that the new method has the obvious improvement in tracking estimation performance.