With the wide application of multi-agent systems, consensus algorithms have received extensive attention. Among them, the Dynamic Average Consensus (DAC) algorithm is beneficial in ensuring the stability and correctness of dynamically changing node data in multi-agent systems. However, most existing DAC algorithms only consider the security of time-evolving reference signals, ignore the security of estimator states, and do not consider that each agent has its own weight. To solve these privacy problems, we propose a Privacy-Preserving Dynamic Weighted Average consensus (PP-DWAC) algorithm for multi-agent systems. The algorithm allows each agent to have an individual weight, which ensures the privacy and security of the time-evolving reference signal and estimator state without affecting the tracking accuracy of the algorithm. Finally, through detailed numerical simulation experiments, we can prove that compared with the current representative algorithms of the same type, the PP-DWAC algorithm has obvious advantages in security, convergence, and dynamism. It shows that the algorithm has broad application prospects.