Overhead cranes are subjected to significant cyclic loads and stresses during their operation, which can lead to fatigue cracking and other forms of damage. Their correct and safe operation is critical while lifting and handling loads. Therefore, industrial cranes are subjected to periodic inspections and maintenance. However, inspections only give information on the condition of a crane at a certain moment and focus on propagating damage which makes it difficult to determine the remaining lifetime of the structure. In this paper a Structural Health Monitoring (SHM) system is presented that is based on load monitoring. The SHM system involves the use of strain sensors combined with the operational crane data to continuously monitor the crane loading and response of the structure. This continuous data stream is analysed by a model optimised using machine learning to determine the fatigue life consumption and detect anomalies. This model is based on a digital twin of the overhead crane to determine the critical areas and critical welds of the crane. The output of the SHM system assists maintenance personnel to identify and address potential problems at critical areas before they result in costly failures or accidents. In addition, SHM can help extending the service life of overhead cranes, reducing the need for costly replacements and helping to ensure the continued safety and efficiency of industrial operations. The results are presented of a pilot project implementing the SHM system on two small capacity cranes (12.5 ton).