One of the reasons structural health monitoring (SHM) techniques are preferred over the traditional visual inspection is that the assessment of the condition is not dependent on the expertise and experience of the person performing the inspection. In reality it is seen that the performance of the SHM techniques is still dependent on the inputs and the preferences of the engineer and as a result, on the experience and the skill of the engineer. The paper presents a semi-automated methodology for the damage detection and localization in a scaled model of an offshore support structure using fibre Bragg grating strain sensors for two damage scenarios namely deterioration of the support condition (simulated by introducing change in the attachment of a leg of a tripod) and a circumferential fatigue crack (simulated by unbolting the flange).The method consists of the offline setup step where the human inputs are provided. The system is trained based on the inputs of system performance, the factors such as thresholds for damage detection, the signal processing, etc. are determined. Once the system is trained, it is then deployed for automated, online damage detection. For the damage identification, a two-step damage method is employed which makes use of the ensemble empirical mode decomposition (EEMD) technique. In the first step, if the damage metric is above a certain threshold, damage is said to be detected. Once the damage has been detected, the EEMD based metric is again used to calculate the damage localization index (DLI) which is used for the localization of the damage. The DLI is used to overcome biases due to higher relative amplitude of vibration of some rosettes. Application of the methodology for the different damage scenarios on several different excitation conditions indicates that the developed methodology is robust and may be used for damage identification in tripod structures.