- Research area
Offshore wind energy integration – challenges and impacts
University of Hull
- Research project
Low Power High Performance Neuromorphic Circuits for Remote Sensing and Monitoring in the Offshore Wind Sector
- Lead supervisor
- PhD Student
- Supervisory Team
Dr Ali Adawi (Reader in Physics, University of Hull)
Dr Neil Kemp (Assistant Professor in Experimental Condensed Matter Physics, University of Nottingham)
Sensing and monitoring in the offshore wind industry has wide ranging applications from monitoring the condition of the wind turbines and their blades to studying the effects of their installation and operation on marine life. In the case of the wind turbines, there is significant financial incentive for real-time multi-component monitoring of performance factors since it can lead to decreased maintenance and operation costs. Offshore wind turbines are exposed to harsh environmental conditions, including changing humidity, air pressure and temperature; and unpredictable loads. The turbines are in hard-to-access places, and replacing and repairing failed components usually requires service crews, cranes and lifting equipment. The latter, furthers downtime loss and energy production, which then impacts on costs to the consumer. Monitoring of the environmental consequences is however also very important as the installation of wind turbines and their operation can cause significant to damage to marine life and ecosystems. During their installation, the pile driving noise generated is well above the tolerance limits of most animals. It can drown out acoustic communication between mammals and can cause temporary hearing impairment, interfering with their orientation and ability to find prey. At greater distances, up to 20 km or more, the noise of the pile driving triggers stress and behavioural responses that often cause the animals to flee their home grounds.
In recent years, there has been significant progress in developing new types of non-volatile logic-in-memory circuits that have both high computational power and low energy usage, both particular important for remote real-time data analysis and decision making. Much working in this area has arisen because of the discovery of a new fundamental circuit element, called a memristor or memory resistor. This new circuit element has attributes that are highly suited for artificial intelligence applications. Memristors have inherent electronic properties that are extraordinarily similar to the electronic switching and storage properties of biological synapses, the key learning and memory centres of brain. Like synapses, memristors are non-volatile devices, which means they do not use energy to store information, unlike conventional computer memory (SRAM, DRAM) which must be continually powered. Memristors are also high scalable down to the nanoscale, which means it possible to develop high density networks, much greater than that possible with traditional CMOS semiconductor circuits. Thus, there are significant advantages in using memristors devices for use in remote sensing applications, where high performance A.I. and in situ data analysis is required whilst using very little energy.
he PhD student will join an established team of researchers in Physics that have been investigating and developing memristors devices since their recent discovery in 2008. The aim of the research will be to investigate and develop new high performance and low energy neuromorphic circuits for analysing data. This will include both simple circuits, such as those used for signal processing and feature extraction to more advanced systems based on neural networks for learning, recognizing patterns, prediction and high-level decision making and optimization.
An experimental based project will focus on fabricating and characterizing the memristor devices and their integration into circuits, whereas a computational project, would involve modelling new types of circuits with highly evolved learning and predictive features.