It is widely accepted that conventional boost algorithms are of low efficiency and accuracy in dealing with big data collected from wind turbine operations. To address this issue, this paper is devoted to the application of an adaptive LightGBM method for wind turbine fault detections. To this end, the realization of feature selection for fault detection is firstly achieved by utilizing the maximum information coefficient to analyze the correlation among features in supervisory control and data acquisition (SCADA) of wind turbines. After that, a performance evaluation criterion is proposed for the improved LightGBM model to support fault detections. In this scheme, by embedding the confusion matrix as a performance indicator, an improved LightGBM fault detection approach is then developed. Based on the adaptive LightGBM fault detection model, a fault detection strategy for wind turbine gearboxes is investigated. To demonstrate the applications of the proposed algorithms and methods, a case study with a three-year SCADA dataset obtained from a wind farm sited in Southern China is conducted. Results indicate that the proposed approaches established a fault detection framework of wind turbine systems with either lower false alarm rate or lower missing detection rate.

Uniaxial trackers are widely employed as the frame for solar photovoltaic (PV) panel installation. However, when used in sloping terrain scenarios such as mountain and hill regions, it is essential to apply a solar-tracking strategy with the sloping factors considered, to eliminate the shading effects between arrays and reduce the electricity production loss due to terrain changes. Based on a uniaxial tracker on the sloping terrain of a PV farm located in Ningxia, this study established a uniaxial solar-tracking strategy for sloping terrain by integrating a spatial projection model with a dynamic shadow assessment method. In the proposed strategy, the optimal tilt angle of the PV array and related desirable adjustment are identified taking into consideration major parameters such as the shadow area ratio S and the average solar irradiance intensity G. A tool underpinned by Matlab Simulink has also been developed to realize the proposed solar-tracking strategy. With the input of a simulated ramp signal β and the dynamically changed time parameters, the tracking angle of PV arrays over the simulated duration is accurately predicted, followed by a series of experimental validations conducted on the winter solstice and a typical sunny day (15 September). Moreover, the study also explored the terrain impacts on solar tracking by comparing the sloping terrain and flat terrain applications. The analytic and experimental results indicate that (a) the maximum value of the G(β) function could serve as the input to identify the optimal tracking angle; (b) the application of the flat terrain tracking (FTT) strategy in sloping terrain would result in a reduction of average solar irradiance intensity harvested by the PV arrays with varying degrees; (c) in the context of an east–west −7° sloping terrain, compared with the FTT strategy, the sloping terrain tracking (STT) strategy enabled anti-shading tracking, and then increased the daily PV electricity yield by 0.094 kWh/kWp, which is around 1.48% of the daily energy production; (d) given a measurement with annual scale, the STT strategy could cause a 1.26% increase in the energy harvesting with a flat uniaxial PV array on a −7° slope terrain, achieving an annual increase of 25.16 kWh/kWp. The experimental comparative analysis validated the precision of the proposed solar-tracking model, which has far-reaching significance for achieving automatic solar-tracking of PV modules, as well as improving the capacity and efficiency of PV systems.