Real-time monitoring for vibration quality of fresh concrete using convolutional neural networks and IoT technology

Vibration quality is critical to ensure the concrete strength, which directly affects the long-term safe operation of concrete structures. The vibration duration and vibration depth are key parameters to guarantee vibration quality. However, traditional manual inspection on concrete surface to judge the vibration duration and estimation of vibration depth is subjective and unreliable. Moreover, existing studies monitor the vibration duration based on the knowledge from prior experiments, ignoring the influence of concrete heterogeneity. Thus, a real-time monitoring method for vibration quality of fresh concrete based on ResNet with 50 layers (ResNet-50) and Internet of Things (IoT) technology is proposed. The IoT-based monitoring framework is proposed to measure vibration depth and capture concrete surface image (CSI). A three-category classification model of CSI is established based on fine-tuned ResNet-50 model using a self-constructed dataset with 15,006 images to determine proper vibration duration. A large-scale hydraulic engineering application verifies the performance of the proposed method.