Сучасний стан підходів до моніторингу технічного стану лопатей вітрових турбін з використанням інформаційних технологій

Abstract

Nowadays wind energy is one of the most important and promising sources of environmentally clean renewable energy. Wind turbine blades are among the most expensive components. Depending on the size, their manufacturing costs range between 10 % and 20 % of total manufacturing costs. Moreover, the size of blades has increased in recent years, leading to greater efficiency and energy production, but presenting higher failure probability. It is extremely important to avoid critical blade failures, because when damaged blades liberate, they have the potential to damage not only the turbines they were attached to, but also other turbines in their vicinity. In order to increase the reliability and safety of wind turbine operation, as well as to reduce costs due to maintenance and downtime in a non-working state, it is necessary to apply modern methods of monitoring the condition of large-sized and highly loaded parts of wind power plants using information technologies. The main types of defects and their classification are considered. The influence of the rotation speed of the turbine and the presence of a damage in the blade on the oscillation natural frequencies was analyzed. The main types and methods of non-destructive testing (NDT) are presented. The acoustic method is considered in detail, as it is rapidly developing and is promising for the field of green energy. The classification of acoustic methods of NDT is provided based on the studied literature. An analytical review of publications considering NDT methods for diagnosing wind turbine blades, including the ones which use unmanned aerial vehicles (UAVs), was conducted. The advantages and disadvantages of each method are shown. The analysis of NDT approach of wind power plants using machine learning based on Gaussian processes to predict natural frequencies of one blade based on the statistical data of the distribution of natural frequencies of neighboring blades and ambient temperature was carried out. The description of the full cycle of the system's functioning, from data collection to decision-making about the possible presence of a defect in the structure, is provided. This paper has summarized and analyzed the most important advances done in the field of NDT in the last few years. The considered approaches can serve as a basis for building new highly reliable methods for detecting dangerous defects in the blade material at the early stages of their development.