Conclusion

The procedure for diagnosing and prognosticating and estimating residual life with hybrid methods combines the advantages and disadvantages of experience-based and knowledge-engineering methods, data-based methods and physical model-based methods. The case studies presented highlight that the components of a hybrid method need to be carefully selected to perform the diagnosis and prognosis for a specific degradation of a given piece of equipment. Selecting the optimum hybrid method for a specific application requires a potential user to meet three major challenges. The first challenge is to determine the best hybrid method based on the data and information available. In theory, the optimal method would correspond to one that requires little information and data on equipment to minimize computational volumes and times. Where data-driven methods do not have sufficient data, they need to be...