Optimized BP Neural Network Model for Intelligent Evaluation of Teaching Quality in Business English Education Using Wireless Network Integration

When evaluating the quality of instruction, the current back propagation neural network model has issues, including delayed convergence and a propensity to slip into local optimality. The current business English teaching quality evaluation technology lacks intelligence and scientificity. This paper improves the BP neural network algorithm and integrates wireless network technology to construct an intelligent teaching quality evaluation and optimization method. Regarding data processing, the study uses a wireless network to collect students’ learning data. It standardizes and preprocesses the data to ensure their correctness and consistency. Given the problem of slow convergence of traditional algorithms, the enhanced Adam optimization algorithm is used to improve the learning rate of the BP neural network, and the rectified linear unit activation function is used to replace the traditional Sigmoid function to avoid the gradient disappearance problem. Dropout regularization technology and an adaptive learning rate are employed to improve the model’s capacity for generalization. Following model establishment, a multidimensional teaching quality evaluation system is built using the improved BP network model. Based on real-time data collection, an adaptive optimization feedback mechanism is adopted to optimize and modify the teaching according to the evaluation results to improve the teaching quality. The test results show that in terms of convergence speed, the improved BP neural network takes 40[Formula: see text]min, while the conventional BP neural network takes 55[Formula: see text]min. Through real-time data input, this technology can also help teachers improve teaching methods, increase student participation and homework completion rates, and ultimately improve teaching levels.