ABSTRACT
Fault alarm time lag is one of the difficulties in fault diagnosis of wind turbine generators (WTGs), and the existing methods are insufficient to achieve accurate and rapid fault diagnosis of WTGs, and the operation and maintenance costs of WTGs are too high. To invent a new method for fast and accurate fault diagnosis of WTGs, this study constructs a stacking integration model based on the machine learning algorithms light gradient boosting machine (LightGBM), extreme gradient boosting (XGBoost), and stochastic gradient descent regressor (SGDRegressor) using publicly available datasets from Energias De Portugal (EDP). This model is automatically tuned for hyperparameters during training using Bayesian tuning, and the coefficient of determination (R2) and root mean square error (RMSE) were used to evaluate the model to determine its applicability and accuracy. The fitted residuals of the test set were calculated, the Pauta criterion (3σ) and the temporal sliding window were applied, and a final adaptive threshold method for accurate fault diagnosis and alarming was created. The model validation results show that the adaptive threshold method proposed in this study is better than the fixed threshold for diagnosis, and the alarm times for the GENERATOR fault type, GENERATOR_BEARING fault type, and TRANSFORMER fault type are 1.5 h, 5.8 h, and 3 h earlier, respectively.
Subject(s)
Algorithms , Electric Power Supplies , Bayes Theorem , Machine Learning , PortugalABSTRACT
The object of the present study was to investigate the feasibility of applying ultraviolet-visible and shortwave near-infrared diffuse reflectance spectroscopy (UV-vis-SWNIR DRS) coupled with chemometrics in qualitative and simultaneous quantitative analysis of drug polymorphs, using cimetidine as a model drug. Three polymorphic forms (A, B and D) and a mixed crystal (M1) of cimetidine, obtained by preparation under different crystallization conditions, were characterized by microscopy, X-ray powder diffraction (XRPD) and infrared spectroscopy (IR). The discriminant models of four forms (A, B, D and M1) were established by discriminant partial least squares (PLS-DA) using different pretreated spectra. The R and RMSEP of samples in the prediction set by discriminant model with original spectra were 0.9959 and 0.1004. Among the quantitative models of binary mixtures (A and D) established by partial least squares (PLS) and least squares-support vector machine (LS-SVM) with different pretreated spectra, the LS-SVM models based on original and MSC spectra had better prediction effect with a R of 1.0000 and a RMSEP of 0.0134 for form A, and a R of 1.0000 and a RMSEP of 0.0024 for form D. For ternary mixtures, the established PLS quantitative models based on normalized spectra had relatively better prediction effect for forms A, B and D with R of 0.9901, 0.9820 and 0.9794 and RMSEP of 0.0471, 0.0529 and 0.0594, respectively. This research indicated that UV-vis-SWNIR DRS can be used as a simple, rapid, nondestructive qualitative and quantitative method for the analysis of drug polymorphs.