Research on the Construction System of Power Science and Technology Innovation Index and Evaluation System Based on Deep Learning

In the continuous innovation and upgrading of the power system, the science and technology innovation index, as a core indicator for evaluating innovation activities, has gradually gained the attention of researchers and practitioners. However, current evaluation methods generally have limitations such as a single evaluation perspective and lack of dynamic response efficiency, making it difficult to fully capture the complexity of innovation driving forces. This paper proposes a construction and evaluation system for the power science and technology innovation index based on a hybrid deep learning model. By integrating CNNs and LSTMs, the system predicts and classifies the innovation index based on deep extraction of innovation driving forces. At the same time, this paper constructs a multidimensional and multi-level innovation evaluation index system that can dynamically evaluate the innovation status of the power system. Through empirical research, the system was tested on multiple real power system data sets and compared with traditional evaluation methods. The experimental results show that the system is superior to existing models in prediction accuracy, dynamic response capability and computational efficiency, and the prediction error of the innovation index is controlled within 5%.