Turbines Predictive Maintenance
Wind turbines are produced in a broad spectrum of vertical and horizontal axes. Neuron soundware focuses on larger turbines that can be used through the electrical grid to contribute to a domestic power supply while selling unused energy back to the utility provider. Large turbine arrays, known as wind farms, are becoming an increasingly significant source of intermittent renewable energy and are being used as part of a plan by many nations to decrease their dependence on fossil fuels.
In less than 3 weeks, the NSW Machine Learning team developed fresh algorithms for predictive maintenance of parts of wind turbines. They developed an end-to-end solution using a range of algorithms and their understanding of application growth that not only forecasts failures but also enables the user to take action and schedule maintenance logistics.
Despite the absence of previous understanding of the wind energy sector and wind turbines, the NSW team rapidly examined and understood the issue and information at hand and provided workable, scalable software in less than 2 weeks by leveraging our expertise in machine learning as well as our understanding of developing user-centric web applications. The AI we created was able to identify the failure of parts of wind turbines up to 15 days in advance by using a range of algorithms, including recurrent and convolutional neural networks, random forest methods, gradient boosting and decision trees. Neuron soundware is now looking to scale-up this solution.