Big Data, Artificial Intelligence & Cloud Computing; a smarter way to manage your fleet

Introduction

Industry is emerging exponentially with modern turbines which have intelligent brain in it. The industry forecast is glowing healthy in installing modern wind turbines all across the world in near future along with considerable support from the various investors to support renewable energy. Wind turbine capacity is increasing along with hub height, rotor size, gearbox size and power electronics complexity etc., due to the technological advancement in the different verticals of wind turbine components. At the same time, control systems are also getting complex in nature and the amount of data generated by the sensors, alarms, event logs and virtually calculated data are huge in scale in modern turbines. The meaningful patterns hidden inside the huge data will be helpful in predicting the component life in advance and leads to help in developing predictive asset management. The data can be structured or unstructured forms, operational up-time and performance improvement can be achieved by manipulating big data efficiently.

Approach

This study focuses on a web intelligence application which is developed and the utilisation of big data in predictive asset management using the application at present and in near future. The development of database with clustering techniques to handle big data and computational time required for the instant dynamic response in web based applications while manipulating big data sets are complex in nature.
The conventional techniques and tools have their own limitations while processing large data sets, such as relatively high handling and processing time and less confidence level in prediction which will be eliminated using big data analytics.


Main body of abstract

Historical data sets collected from more than 550 operational wind turbines spread across 15 sites from 5 different manufactures are stored in the database. To collect all the turbine data sets without manufacturer interference, real time data transfer technique has been implemented and integrated with historical database.
The integrated data sets, analog and digital data measured and recorded by sensors, events, alarms, virtual data, weather data and logs are collected from all the turbines seamlessly, are helping to make daily as well as long operational strategic decision. When these complex data sets are processed and analysed properly it can give insight and visibility of emerging trends. A study has been performed using the above fleet of turbines by which improper pitching, yawing, temperature curtailment beyond the design technical specification are observed and fixed. Cloud computing and storage techniques were used to have an enhanced geographical access and considering the massive storage requirements in future.
When wind farm capacity is big and modern turbines age reaches more than 10 or 15 years when more focus required to monitor performance and component life prediction, the amount of data need to be handled will be huge. Parallel processing techniques or grid computing approach need to be considered well in advance to crunch large volumes of data to get extremely fast response.


Conclusion

Big data analytics is an emerging technique in predictive asset management and helps in reducing the cost of energy by improving practices in operations & maintenance strategy like inventory management, plant performance and monitoring the health of wind turbine without manual intervention. Artificial Intelligence and machine learning techniques are all set to add fire power to the rooster. The implemented algorithm and techniques shows encouraging results and more research leads to,
• Improvement in decision making process
• Confident predictive maintenance
• Reduces the operational cost
• Improve the effective inspection of components
• If a component is replaced in a wind turbine, generating a new relation between old data and new component data is a highly complex task in a short span.



Learning objectives
Developing a web intelligence application especially for predictive asset management of wind turbine will help in all the aspects of performance improvement by incorporating big data analytics and machine learning