Search

Wind energy plays an important role in the field of renewable energy due to its lower cost and the maturity of its manufacturing technology. During the last few years, there has been an increase in offshore wind turbine (OWT) projects especially in China. Considering the fact that ocean environment where these offshore wind turbines operate usually has a negative impact on their structural health, the demand for investigating the dynamic characteristics of offshore wind turbines based on sea test is increasing. However, current studies are mainly focused on method development, theoretical analysis and numerical simulation, and few attention has been given to the study of the dynamic characteristics of an actual OWT in different operational states based on sea test. To improve the current studies, in September 2016, November 2016 and April 2017 a series of monitoring campaigns of an actual in service OWT and its monopile were conducted at an offshore wind farm in the sea area near Jiangsu Province, China. For the purpose of acquiring high-quality vibration signal, one type of artificial excitation was adopted to excite the OWT vibration. In these monitoring campaigns, with the help of a group of wireless health monitoring sensor nodes developed by Ocean University of China (OUC), we successfully acquired the vibration signal of the wind turbine in parked state and normal operational state and acquired the vibration signal of the monopile in construction, which means no tower was installed. Afterwards, two modal analysis methods, the eigensystem realization algorithm (ERA) and stochastic subspace identification (SSI), were used to obtain the modal parameters of the tested structure using the vibration signal under artificial excitation and random wave excitation, respectively. The results show that the ERA method can successfully identify the first two order modal parameters of the monopile and the first three order modal parameters of the wind turbine in parked or operational state. However, the SSI method failed to identify the third-order modal parameters of the wind turbine, regardless of whether it is in parked state or operational state (only the first two order modal parameters can be identified); meanwhile, SSI failed to identify the modal parameters of the monopile. Concerning the high-level environmental noise in an offshore site, we applied weak-mode identification (WMI) for the first time to obtain the modal parameters of OWT using vibration signal under both artificial excitation and random wave excitation. The results indicated that WMI has achieved the same good performance as the ERA when the vibration signal under artificial excitation is used. When the vibration signal under random wave is used, WMI successfully identified the first order modal parameters of the monopile and the first three order modal parameters of the wind turbine in parked or operational state, which indicated a better ability than that of the SSI method. We can conclude that the modal parameters of the wind turbine can be obtained by the WMI method and ERA method, and the ERA method is the best way to obtain the modal parameters of monopile under artificial excitation. These monitoring campaign of an actual OWT not only provide modal information support for structural maintenance and operation safety assessment, but also provide a real data reference for wind turbine design. • Vibration signals of an offshore wind turbine under artificial and ambient excitations are obtained, respectively. • Three methods are emplyed to obtain the modal parameters of the tested offshore wind turbine. • ERA is effective in identifying modal parameters from vibration signal under artificial excitation. • SSI and weak-mode identification can obtain modal parameters from vibration signal under ambient excitaiton. [ABSTRACT FROM AUTHOR]

The development of China's manufacturing industry has received global attention. However, research on the distribution pattern, changes, and driving forces of the manufacturing industry has been limited by the accessibility of data. This study proposes a method for classifying based on natural language processing. A case study was conducted employing this method, hotspot detection and driving force analysis, wherein the driving forces industrial development during the "13th Five-Year plan" period in Jiangsu province were determined. The main conclusions of the empirical case study are as follows. 1) Through the acquisition of Amap's point-of-interest (POI, a special point location that commonly used in modern automotive navigation systems.) data, an industry type classification algorithm based on the natural language processing of POI names is proposed, with Jiangsu Province serving as an example. The empirical test shows that the accuracy was 95%, and the kappa coefficient was 0.872. 2) The seven types of manufacturing industries including the pulp and paper (PP) industry, metallurgical chemical (MC) industry, pharmaceutical manufacturing (PM) industry, machinery and electronics (ME) industry, wood furniture (WF) industry, textile clothing (TC) industry, and agricultural and food product processing (AF) industry are drawn through a 1 km× 1km projection grid. The evolution map of the spatial pattern and the density field hotspots are also drawn. 3) After analyzing the driving forces of the changes in the number of manufacturing industries mentioned above, we found that manufacturing base, distance from town, population, GDP per capita, distance from the railway station were the significant driving factors of changes in the manufacturing industries mentioned above. The results of this research can help guide the development of manufacturing industries, maximize the advantages of regional factors and conditions, and provide insight into how the spatial layout of the manufacturing industry could be optimized. [ABSTRACT FROM AUTHOR]