[318] How different monitoring approaches impact the P-F model – a case study
D Hickey, M Smith, B Simmonds and I Dinwoodie
The Natural Power Consultants, UK
In the wind industry, condition-based monitoring (CBM) is defined as utilising a range of data types in corroboration to assess the health and status of an electromechanical component. The practicality of this technique allows for reliability engineers to engage with operations, servicing and repair teams to facilitate the best and most predictive approach to scheduled maintenance. Classically, these data types are centred around high-frequency vibration measurements, from sophisticated and expensive data acquisition systems. However, a widely used model for assessing the relationship between potential failure (P) and functional failure (F) is the P-F curve; this provides a deterioration model for the preferred method of data analysis that can capture any damage patterns in the most efficient and effective timeframe.
This paper attempts to explain the P-F deterioration model by exploring failure detection from a variety of real wind turbine case studies. Data types ranging from visual inspection, supervisory control and data acquisition (SCADA) and vibration analysis are provided. The paper then offers a philosophical view on how these diagnoses could have been predicted by utilising future techniques that the industry is currently researching, with a particular focus on advancements in machine learning techniques and their application in rotational and electromechanical fault prognostics. A final discussion details the usage of IOT sensors with a cloud-based approach to data storage.
This paper attempts to explain the P-F deterioration model by exploring failure detection from a variety of real wind turbine case studies. Data types ranging from visual inspection, supervisory control and data acquisition (SCADA) and vibration analysis are provided. The paper then offers a philosophical view on how these diagnoses could have been predicted by utilising future techniques that the industry is currently researching, with a particular focus on advancements in machine learning techniques and their application in rotational and electromechanical fault prognostics. A final discussion details the usage of IOT sensors with a cloud-based approach to data storage.
