Concrete structures dominate modern construction projects; however, issues with concrete quality frequently arise. As a critical factor in the control and assessment of concrete quality, the accurate determination of concrete strength is essential for ensuring the overall quality of construction projects. Current assessment methods exhibit a degree of randomness and subjectivity in determining concrete strength. This paper integrates Bayesian neural networks with commonly used concrete strength evaluation methods, such as the ultrasonic method and the rebound method, to achieve more precise identification results. Finally, an assessment of the concrete strength of existing structures was conducted, and the results were validated using the core drilling method. The findings indicate that the integration of Bayesian neural networks with traditional nondestructive testing methods not only enhances the accuracy of concrete strength assessments but also provides a more reliable tool for engineers working with existing structures.