Analysis and Diagnosis of Wind Turbine Blades-Angle Abnormality Based on Anti-noise Hilbert-Huang Transform and Genetic Glowworm Swarm Optimization

• Main Authors: Wei-Ciao Jhang, 張惟喬
• Other Authors: Chun-Yao Lee
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/aayem9

碩士 === 中原大學 === 電機工程研究所 === 102 === This study proposes an anti-noise Hilbert-Huang transform (AHHT) approach with the aim to enhance the angle-detection capability of wind turbine blades. Artificial neural network (ANN) are applied the features to recognize the blades faults. First, this study proposes a novel model of blades unbalanced signal analysis based on the anti-noise Hilbert–Huang transform (AHHT). The AHHT combines a particle swarm optimization (PSO) algorithm with an Hilbert-Huang transform(HHT) based on ensemble empirical mode decomposition(EEMD). The AHHT model is used to slove interference problems of signals in interference environments. Second, this thesis extract the feattures of current signals in the signal analysis method, and input these features to a genetic glowworm swarm optimization (GGSO) to select the effective features subset of features set. The GGSO combines a glowworm swarm optimization (GSO) algorithm with an genetic algorithm (GA). Finally, this thesis utilizes fast Fourier transform (FFT), multi-resolution analysis (MRA), HHT and AHHT, for blades unbalanced analysis. The results indicate that the AHHT has the best detection accuracy. Moreover, the inferior features may cause the interference phenomenon, and subsequently, the detection accuracy of probabilistic neural network (PNN) decline significantly. This study utilizes GGSO and sequential forward selection (SFS), for feature selection problem. The results indicate that the GGSO has the best detection accuracy.