Micro-certification, as an emerging competency certification method, has the advantages of strong flexibility and high relevance. In order to solve the problem of unclear personalized professional development path for kindergarten teachers, this study constructs a micro-certification system based on blockchain technology, and adopts a quasi-experimental research methodology to carry out a two-school-year practical study in two kindergartens in a city. The study uses the federal learning algorithm, differential privacy protection mechanism and other technical means to design a four-layer system architecture with a multimodal resource presentation layer, a data management layer, an online learning input analysis layer, and a visualization display layer. By analyzing the pre and post-test data from 128 valid questionnaires, the results show that: the paired-sample t-test Sig values for the five dimensions of teacher professional development are all 0.000, which is significantly lower than the 0.05 significance level; the decentralized federated learning model based on differential privacy increases the running time by only 0.19-0.22 seconds per round on the MLP MNIST dataset; and the system communication overhead in the case of 10 users participated reached 28.04M; the t-test value of the effect of teachers’ overall professional level improvement was -19.115. The study confirms that the micro-certification system can effectively promote the personalized professional development of kindergarten teachers, and provides a feasible path to build a digital teacher education ecosystem.
