Statistical and time domain signal analysis of the thermal behaviour of wind turbine drive train components under dynamic operation conditions

Abstract

Gearboxes and generators are fundamental components of all electrical machines and the backbone of all electricity generation. Since the wind energy represents one of the key energy sources of the future, the number of wind turbines installed worldwide is rapidly increasing. Unlike in the past wind turbines are more often positioned in arctic as well as in desert like regions, and thereby exposed to harsh environmental conditions. Especially the temperature in those regions is a key factor that defines the design and choice of components and materials of the drive train. To optimize the design and health monitoring under varying temperatures it is important to understand the thermal behaviour dependent on environmental and machine parameters. This paper investigates the behaviour of the stator temperature of the double fed induction generator of a wind turbine. Therefore, different scenarios such as start of the turbine after a long period of no load, stop of the turbine after a long period of full load and others are isolated and analysed. For each scenario the dependences of the temperature on multiple wind turbine parameters such as power, speed and torque are studied. With the help of the regression analysis for multiple variables, it is pointed out which parameters have high impact on the thermal behaviour. Furthermore, an analysis was done to study the dependences in the time domain. The research conducted is based on 10 months of data of a 2 MW wind turbine using an adapted data acquisition system for high sampled data. The results appear promising, and lead to a better understanding of the thermal behaviour of a wind turbine drive train. Furthermore, the results represent the base of future research of drive trains under harsh environmental conditions, and it can be used to improve the fault diagnosis and design of electrical machines.