In order to improve the accuracy and feedback efficiency of dynamic assessment results of English teaching, a teaching effectiveness optimization model for deep learning target adaptation was proposed. The model collects data such as platform logs, writing texts, oral audio, reading tests and interactive feedback, and completes standardization processing, feature extraction and learning ability portrait. Then, the index system of language foundation, discourse understanding, expression generation, thinking processing, interaction and collaboration and self-regulation are constructed, and the mapping relationship between dynamic evaluation index and deep understanding, comprehensive expression, transfer application and other goals is established. The experiment was carried out with 96 students for 16 weeks. The comprehensive score of the experimental class increased from 71.36 to 84.21, the feedback acceptance rate reached 84.36%, the evaluation time was reduced to 8.4 minutes, and the average error was 3.18%. The results show that the proposed model can improve the efficiency of English teaching evaluation and the achievement of deep learning goals.
