Data plays a crucial role in Structural Health Monitoring (SHM), determining the success and efficiency of the entire system. Sensor data helps in the early detection of damage or deterioration in structures, allowing for timely corrective measures before issues become severe. Data analysis allows for the optimization of maintenance and repair schedules, saving costs and resources. However, during the data collection process from SHM systems, data loss and errors frequently occur. These issues are primarily due to sensor installation errors and transmission path problems. To address this, this study proposes the application of machine learning to reconstruct lost or erroneous data. The results show that the reconstructed data is highly accurate and can represent actual data for SHM purposes. This study demonstrates the potential of machine learning in enhancing SHM, ultimately contributing to the safety, longevity, and cost-effectiveness of transportation infrastructure.