A Critical Review on Wind Turbine Power Curve Modelling Techniques and Their Applications in Wind Based Energy Systems

Table 5

Comparison of modelling methods.

Models	Data required for modelling	Merits	Demerits	Applications
Polynomial models; (linear, quadratic, binomial, cubic, and Weibull based)	, , , and of turbine	(i) Simplicity; (ii) Limited data required; (iii) Parameter calculation is easy	(i) Do not follow curvature of power curve; (ii) Accuracy is poor; (iii) Sometimes more than one expression are used to describe the shape of curve	Suitable for power prediction and energy estimation during initial resource assessment and designing of small systems
Manufacturer’s curve fitting	Manufacturer’s curve	(i) Need less data	(i) Requires manufacturer’s power curve data; (ii) Fairly accurate; (iii) Many expressions may be required for accurate representation of curve	Suitable for power prediction and energy estimation during initial resource assessment and designing of small systems
Cubic splines	Manufacturer’s curve	(i) Exact fit	(i) Variance of data is not taken into account	Power prediction
4PL model	Manufacturer’s curve, actual data of wind farm	(i) Consider inflection point on curve; hence shape of curve is represented more accurately than the earlier models; (ii) One expression is required	(i) Asymmetry of curve not modelled	Online monitoring; further research on power prediction during design and power forecasting applications is required
5PL model	Manufacturer’s curve; actual data of wind farm	(i) Consider inflection point on curve and asymmetry of curve is modelled more accurate than the earlier and 4PL models; (ii) One expression is required	(i) Parameter estimation is difficult	Further research on power prediction during design and power forecasting and online monitoring applications is required
ANN	Actual data of wind farm	(i) Found to be accurate than other methods	(i) Black box approach	Wind power assessment for sizing and power forecasting, and online monitoring applications, suitable for group of turbines
Clustering	Actual data of wind farm	(i) More accurate than the regression method	(i) Accuracy depends on the number of cluster centres	Wind power assessment for sizing and power forecasting, and online monitoring applications, suitable for group of turbines
-NN	Actual data of wind farm	(i) Performance variable in different studies	(i) Accuracy depends on value of; (ii) Less training time as instance based scheme	Wind power assessment for sizing and power forecasting, prediction, and; online monitoring applications, suitable for group of turbines
Model trees(REP, M5P, and bagging tree)	Actual data of wind farm	(i) Fairly accurate	(i) Much research not available	Applicability in power prediction and; online monitoring to be explored
ANFIS	Actual data of wind farm	(i) Integrates best features of fuzzy systems and neural networks; (ii) Accurate method; (iii) Fewer parameters required in training therefore faster training compared to NN; (iv) Tunable membership functions	(i) Computational complexity	Wind power assessment for sizing and power forecasting for energy trading Online monitoring applications, suitable for group of turbines
Copula model	Actual data of wind farm	(i) Considers joint probability distribution of wind speed and power; (ii) Includes measures of uncertainty in performance estimates	(i) Needs advanced method for parameter estimation of marginals	Applications for condition monitoring to be investigated