I. Working Principle of Wind Turbine Generator
The principle of wind power generation is to use wind power to drive the windmill blades to rotate, and then use a speed increaser to increase the rotation speed, thereby causing the generator to generate electricity. According to windmill technology, a breeze speed of about three meters per second (a very light breeze) is sufficient to start generating electricity.
Wind power is becoming a global trend because it has no fuel issues and does not produce radiation or air pollution.
Wind power generation is popular in countries like Finland and Denmark; my country's wind power industry has also seen rapid development in recent years. Small wind power systems are highly efficient, but they are not simply composed of a single generator head; rather, they are small systems with a certain level of technological sophistication: a wind turbine + charger + digital inverter. A wind turbine consists of a turbine head, rotor, tail fin, and blades. Each part is important, and their functions are as follows: the blades receive wind power and convert it into electrical energy through the turbine head; the tail fin ensures the blades are always facing the wind direction to maximize wind energy; the rotor allows the turbine head to rotate flexibly, enabling the tail fin to adjust its direction; the rotor of the turbine head is a permanent magnet, and the stator windings cut magnetic lines of force to generate electricity.
Because wind power is unstable, wind turbines output alternating current ranging from 13 to 25V. This current must be rectified by a charger before being used to charge the battery, converting the electrical energy generated by the wind turbine into chemical energy. Then, an inverter with protective circuitry converts the chemical energy in the battery into 220V AC mains power to ensure stable operation.
Mechanical connection and power transmission: The blades of a horizontal axis wind turbine are connected to a universal flexible coupling via a gearbox and its high-speed shaft, transmitting torque to the generator's drive shaft. This coupling should have excellent damping and vibration absorption characteristics, absorbing appropriate radial, axial, and angular misalignments, and preventing overload of the mechanical device. Another type is the direct-drive wind turbine, where the blades are directly connected to the motor without a gearbox.
II. Techniques for Diagnosing Wind Turbine Defects
After more than a decade of rapid development, defect diagnosis techniques have evolved into numerous methods based on various principles. Compared to earlier methods, these approaches have made significant progress in terms of detection, diagnosis, and robustness, and have formed relatively complete system structures for linear time-invariant systems. All defect diagnosis methods can be categorized into three types: knowledge-based methods, analytical model-based methods, and signal processing-based methods. In recent years, new methods have emerged, such as discrete task-based diagnosis methods and online learning-based diagnosis methods.
The so-called defect diagnosis based on analytical models involves comparing the measurable information of the target system with the prior information of the system expressed by the model, generating residuals, and then analyzing and processing these residuals to complete the defect diagnosis. A residual is a linear or nonlinear function composed of the input and output information of the system being diagnosed, independent of its normal operating conditions. In the absence of defects, the residual is equal to zero or approximately zero (in a certain sense), while when defects occur in the system, the residual should significantly deviate from zero. To facilitate defect identification, the residual should belong to one of the following two categories.
1) Structured residuals: For each defect, the residual has different components corresponding to it. When the defect is diagnosed, these specific components change from zero to non-zero.
2) Fixed-direction residuals: These refer to residual vectors that have different directions corresponding to each defect. Generally, defect detection and separation techniques based on analytical models involve two phases:
(l) Residual generation: This is the process of using an appropriate algorithm to process the input and output of the system to obtain the residual signal;
(2) Residual evaluation: is the process of using appropriate decision functions and decision rules to determine the probability of defects occurring.
Based on the different methods of residual generation, the FDI methods of the model can be further divided into environmental estimation methods, equivalent space methods, and parameter estimation methods. Although these three methods are developed independently, they are not isolated from each other, but rather have certain interrelationships. The equivalent space method and the observer method are structurally equivalent; while the connection between the parameter estimation method and the observer method is that the residuals obtained by the observer method include the residuals obtained by the parameter estimation method, so the two methods are essentially complementary.
