Many different monitoring systems for assessing the condition of power transformers are now commercially available. While they offer various advantages, they also present certain limitations and significant costs. Currently, no universal solution exists due to the complexity of the task. This contribution introduces an innovative approach based on the use of existing cost-effective vibration sensors typically employed for rotating machinery monitoring. These wireless sensors communicate via radio waves and can be installed on any transformer in service. The measured signals are processed using an industrial-grade artificial intelligence analytics platform, aiming to detect deviations in the monitored parameters which indicate potential mechanical defects that could lead to failure. Successful tests conducted on an in-service power transformer demonstrated a high correlation between the vibration spectrum and current, as expected. Additionally, the system's sensitivity was validated through deformation tests performed on a small-scale test transformer. The proposed approach is promising, and the potential of integrating additional sensors into the system to measure additional parameters has been identified.
