<p>Wind turbines are crucial in the production of electricity using renewable energy, but their efficiency and lifespan are affected by erosion of the wind turbine blades (WTBs), which is difficult to detect and impacts energy production. Therefore, wind farms require proper maintenance to prevent failures and maintain competitive operational costs. However, early detection of erosion is a major technical challenge. This study proposes an innovative method based on pressure coefficient (<InlineEquation ID="IEq100"> <EquationSource Format="TEX">$c_p$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>c</mi> <mi>p</mi> </msub> </math></EquationSource> </InlineEquation>) measurements to automatically identify different levels of erosion in the WTB’s, which does not require complex instrumentation. Unlike traditional methods such as vibration analysis and piezoelectric sensors used to measure acceleration, temperature, strain, or force, the <InlineEquation ID="IEq101"> <EquationSource Format="TEX">$c_p$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>c</mi> <mi>p</mi> </msub> </math></EquationSource> </InlineEquation> measurement provides a highly efficient alternative because the coefficient represents an aerodynamic component of the WTB, which has a direct relationship with the drag force (<InlineEquation ID="IEq102"> <EquationSource Format="TEX">$F_D$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>F</mi> <mi>D</mi> </msub> </math></EquationSource> </InlineEquation>) and lift force (<InlineEquation ID="IEq103"> <EquationSource Format="TEX">$F_L$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>F</mi> <mi>L</mi> </msub> </math></EquationSource> </InlineEquation>), complementing the structural monitoring of the WTB. The results show that the proposed methodology allows for a highly accurate classification of levels of erosion and estimation of associated power losses. This demonstrates the feasibility of using <InlineEquation ID="IEq104"> <EquationSource Format="TEX">$c_p$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>c</mi> <mi>p</mi> </msub> </math></EquationSource> </InlineEquation> for structural and energy monitoring of wind turbines, with potential application in real-world conditions. A machine learning approach was applied to classify different erosion levels based on controlled experiments in a wind tunnel with a 1 kW blade. The method was validated with multiple ML algorithms, demonstrating high classification accuracy. Additionally, power losses due to erosion were estimated at various wind speeds, highlighting the feasibility of this methodology for structural and energy monitoring.</p>

Detection of Wind Turbine Blade Erosion and Power Loss Analysis Using the Pressure Coefficient and Machine Learning

期刊论文
发表日期：

2025年12月10日

查看全文

Javier Carmona,
Marco Arjona,
Josué Enríquez Zárate,
Leonardo Trujillo

摘要

Wind turbines are crucial in the production of electricity using renewable energy, but their efficiency and lifespan are affected by erosion of the wind turbine blades (WTBs), which is difficult to detect and impacts energy production. Therefore, wind farms require proper maintenance to prevent failures and maintain competitive operational costs. However, early detection of erosion is a major technical challenge. This study proposes an innovative method based on pressure coefficient ( $c_p$ $c_{p}$ ) measurements to automatically identify different levels of erosion in the WTB’s, which does not require complex instrumentation. Unlike traditional methods such as vibration analysis and piezoelectric sensors used to measure acceleration, temperature, strain, or force, the $c_p$ $c_{p}$ measurement provides a highly efficient alternative because the coefficient represents an aerodynamic component of the WTB, which has a direct relationship with the drag force ( $F_D$ $F_{D}$ ) and lift force ( $F_L$ $F_{L}$ ), complementing the structural monitoring of the WTB. The results show that the proposed methodology allows for a highly accurate classification of levels of erosion and estimation of associated power losses. This demonstrates the feasibility of using $c_p$ $c_{p}$ for structural and energy monitoring of wind turbines, with potential application in real-world conditions. A machine learning approach was applied to classify different erosion levels based on controlled experiments in a wind tunnel with a 1 kW blade. The method was validated with multiple ML algorithms, demonstrating high classification accuracy. Additionally, power losses due to erosion were estimated at various wind speeds, highlighting the feasibility of this methodology for structural and energy monitoring.