1 Introduction
In recent years, my country's installed wind power capacity has maintained rapid growth. Wind turbine units can be divided into two main categories: constant speed constant frequency (CSCF) units and variable speed constant frequency (VSCF) units. VSCF units allow the wind turbine to adjust the generator speed according to changes in wind speed. This flexible control strategy can absorb gust energy, reduce mechanical stress, and maximize energy capture, thus improving wind energy utilization. These advantages are unmatched by CSCF units, making VSCF technology a current research hotspot in the domestic and international wind power generation field.
VSCF (Variable Reinforced Conversion) wind turbines mainly consist of doubly fed induction generators (DFIG) and direct-driven permanent magnet synchronous generators (D-PMSG). Among these, brushed DFIGs are the mainstream, typically employing high-speed motors requiring gearboxes for speed increase. Furthermore, the presence of carbon brushes and slip rings in the motors contributes to a complex system structure and inconvenient maintenance. D-PMSGs, on the other hand, eliminate the gearbox, reducing mechanical transmission components and maintenance costs, increasing system reliability and overall efficiency. Additionally, the use of a full-power converter enhances the system's ability to cope with grid faults. Therefore, it can be predicted that D-PMSGs will become the mainstream for future wind power generation.
2. Wind turbine modeling and characteristic simulation
There are many types of wind turbines, but they can be divided into horizontal axis wind turbine generators and vertical axis wind turbine generators according to the direction of the wind turbine axis. In commercial operation, wind power systems mostly use horizontal axis wind turbines.
Currently, wind turbine modeling mainly employs two methods: one based on blade element theory and the other based on fluid energy conversion theory. Blade element theory divides the blade into several micro-elements (blade elements), calculates the torque of each element through force analysis, and then sums all the torques to obtain the total output torque of the wind turbine. Fluid energy conversion theory, on the other hand, uses aerodynamic formulas and design parameters to derive the wind energy utilization coefficient, and then uses the wind energy capture formula to obtain the wind turbine's output power. The above simulation analysis focuses on the static characteristics of wind turbines. However, in actual systems, wind turbines also exhibit dynamic characteristics, primarily including tower shadow effects and shear effects. These two effects cause torque pulsations. Therefore, to accurately reflect the output torque of a wind turbine, its dynamic characteristics must be considered during the modeling process.
Wind speed models are a crucial component of wind turbine simulation systems, determining the dynamic characteristics of the simulation process. Therefore, establishing reliable wind speed models is essential. Currently, many experts both domestically and internationally have researched wind speed models. For example, some models employ the Weibull distribution and combine it with the exponential distribution to form a four-logarithmic hybrid model; others utilize the turbulence effect of wind speed, superimposing turbulence components onto the average wind speed component to establish a wind speed model. The turbulence component is solved using stochastic process theory, employing the wind power spectrum and coherence function to derive the Fourier spectrum of wind speed, followed by inverse Fourier transform; still others establish wind speed models based on power spectral density, using an autoregressive moving average (ARMA) model to establish a wind speed model that satisfies certain power spectral density characteristics. However, these models are overly complex due to the numerous influencing factors they consider, and some parameters are difficult to obtain. Therefore, in the laboratory, a four-component wind speed model composed of basic wind, gust wind, gradual wind, and random wind can be used. This model is simple, has clear physical meaning, and is highly practical.
In wind turbine simulation systems, the prime mover typically uses a DC motor or an asynchronous motor. DC motor control systems primarily employ DC pulse width modulation (PWM) technology, controlling the DC motor to operate stably at the optimal torque point on the wind turbine's characteristic curve based on calculated optimal torque. Asynchronous motor wind turbine simulation systems mainly use vector control technology, controlling the decoupled torque components to achieve wind turbine simulation. However, some systems also employ direct torque control (DTC) strategies to achieve flexible control of asynchronous motors.
3D-PMSG topology
The d-pmsg uses a full-power converter. Depending on the wind turbine capacity, the converter can generally be divided into two categories: low-voltage systems and medium-voltage systems. Low-voltage systems have a grid voltage level below 690V and mostly use a two-level structure; medium-voltage systems have a grid voltage level above 690V and use a multi-level structure. Currently, the voltage level of wind turbines is typically 690V.
This system employs a back-to-back dual PWM converter topology. The generator-side converter achieves maximum power point tracking (MPPT) by controlling the motor speed, while the grid-side converter stabilizes the DC bus voltage and regulates active and reactive power on the grid side. Decoupled control of the converters allows for flexible control of motor torque and efficiency, improving system operating characteristics. While this structure is complex to control, requires a large number of high-power IGBTs, and has a relatively high system cost, it significantly improves system performance and shows promising application prospects. The WG2000KFP 2MW full-power converter manufactured by Hefei Sungrow Power Supply Co., Ltd. in China utilizes this structure.
For parallel operation of converters, the system's current capacity can be increased without increasing the converter voltage level by decomposing a large-capacity converter into several smaller-capacity converters. There are generally two methods: one is to connect the multi-phase windings of a direct-drive permanent magnet wind turbine generator to several smaller-capacity converters, with each converter connected in parallel on the grid side and then connected to the grid via a step-up transformer; the other is to have the wind turbine drive multiple permanent magnet synchronous generators via a gearbox, which are then connected in parallel on the grid side via smaller-capacity converters and finally connected to the grid via a step-up transformer. This structure reduces the capacity of a single generator and the corresponding power electronic converter. However, both methods have disadvantages such as high transmission current, high line and switching losses, and increased costs due to the step-up transformer.
Multilevel converter topologies in wind power systems mainly include two types: neutral-point-clamped (NPC) converters and cascaded H-bridge converters. A back-to-back neutral-point-clamped topology is presented, where both the front and back ends employ neutral-point clamping. Alternatively, a topology with uncontrolled rectification at the front end and neutral-point clamping at the back end can be used. Compared to two-level back-to-back converters, back-to-back neutral-point-clamped converters offer higher voltage withstand capabilities and fewer harmonics in the output voltage at the same DC bus voltage. Therefore, multilevel converters are more advantageous in wind power systems above 2MW. Cascaded H-bridge converters are generally directly coupled to multiphase permanent magnet synchronous generators. This topology requires independent DC current for clamping, and the number of DC power supplies used by the cascade units increases with the number of output waveform levels.
4D-PMSG converter control strategy
4.1 Maximum output power adjustment
The d-pmsg operates in different variable speed constant frequency states under different wind speeds. When the wind speed is lower than the rated wind speed, the wind energy captured by the wind turbine is less than the rated power. In order to improve the operating efficiency of the wind turbine, the motor speed must be adjusted to achieve MPPT. When the wind speed is equal to or greater than the rated wind speed, the pitch angle of the wind turbine needs to be adjusted to limit the wind energy captured, maintain the rated output power, and control the unit to stop operation when the wind speed is greater than the cut-out wind speed.
Currently, there are three main methods for capturing the maximum power point:
(1) Tip speed ratio control. The rotational speed is adjusted in a timely manner when the wind speed changes to maintain the optimal tip speed ratio and the best power output. However, wind speed detection is difficult, the rotational speed adjustment accuracy is low, and the increased cost and maintenance difficulty are due to the need for a precise anemometer.
(2) Power signal feedback. The power output of the wind turbine is correlated with the wind speed, and the rotational speed is adjusted according to the power output and wind speed curve. However, since each wind turbine is different, it is difficult to unify the power and wind speed curve, so its practical value is not high.
(3) Hillclimb Searching (HCS). This method involves continuously applying small disturbances to the wind turbine's rotational speed, determining the turbine's speed increment based on changes in generator output power, and repeatedly searching until the turbine reaches its maximum power point. The search step size can be either fixed or variable. However, this method is unsuitable for wind turbines with large rotational inertia and presents challenges in coordinating step size with system stability and efficiency, as well as avoiding tracking incorrect rotational direction.
To address the aforementioned issues, many scholars have proposed improved HCS methods and combined them with intelligent control technology to form hybrid control strategies. For example, small-signal perturbation methods are used to maximize the system's average power; based on grey theory, a dynamic GM(1,1) model is used for prediction, and the least squares method is used for fitting to calculate the maximum power point; an operating point correction module and a direction monitoring module are used to train the adaptive control strategy; a model reference adaptive method is studied and applied to the control of the unit under varying wind speeds to enable the wind turbine to obtain maximum wind energy; rapid maximum wind energy capture control is achieved through offline training of neural networks; and the h∞ hybrid sensitivity robust control principle is adopted to improve the anti-interference capability of the system's speed control.
The above addresses the issue of maximizing wind energy capture by wind turbines when wind speeds are below the rated wind speed. However, when wind speeds are high, to ensure the safe and reliable operation of the wind power system, it is necessary to limit the turbine's rotational speed. This is typically achieved by controlling the turbine's pitch angle to reduce the captured energy. The design of the pitch angle controller relies on the nonlinear dynamic model of the wind turbine; usually, the model is linearized first, followed by the controller design. A pitch angle controller was designed using the pole placement method. However, its performance degrades when the turbine operating point deviates from its linearization point, causing system instability. An improved version of the conventional controller, a variable-gain controller, was designed, continuously adjusting the controller gain as the turbine operating point changes. This method requires precise wind speed, presenting challenges in practical applications. A variable-speed, variable-pitch controller based on fuzzy sliding mode control theory was designed to simulate the nonlinear dynamics of the turbine using high-order polynomials and transcendental equations, exhibiting good robustness. A variable-pitch controller was designed based on active disturbance rejection control theory. A state observer monitors the system state and wind speed disturbances, configuring the nonlinear structure according to the state deviation to achieve good dynamic performance. An h∞ controller was designed and verified through simulation. A nonlinear pitch angle robust controller was designed based on singular perturbation theory and the inverse system method.
4.2 Machine-Side Converter Control Strategy
The topology of the generator-side converter can employ either uncontrolled diode devices or fully controlled devices. For uncontrolled rectifier topologies, the rectifier bridge injects a large amount of low-frequency harmonics into the motor, resulting in significant current distortion. Typically, a boost chopper is used to achieve power factor correction and a wider motor speed range. Current boost circuit control strategies include average current control using fixed-frequency PWM technology and peak current control using variable-frequency PWM technology. While using multiple parallel choppers meets the harmonic content requirements of megawatt-level wind power systems for turbine input current, this two-stage structure reduces system efficiency. A novel single-stage Z-source circuit is employed to achieve flexible buck-boost control, ensuring a wide range of variable-speed operation and high efficiency.
For the generator-side converter using a fully controlled device topology, the main research focuses on system modeling and the control strategy of the permanent magnet synchronous motor (PMSG). Mathematical models of the wind turbine, PMSG, and inverter system in a direct-drive permanent magnet wind power generation system are presented, and the characteristics of a 2.5MW unit are analyzed. To improve system performance, researchers studied the control strategies for the PMSG, which can be broadly categorized into ISD=0 control, unity power factor control, optimal torque control, constant stator voltage control, and efficiency-optimal control. The ISD=0 control strategy employs rotor flux orientation control, where the electromagnetic torque is only related to the q-axis component of the stator current. It is simple to control and has no demagnetizing effect, but the motor power factor is low, and the capacity of the motor and converter cannot be fully utilized. Unity power factor control overcomes the shortcomings of the above strategies, but the motor output torque is relatively small. Optimal torque control, also known as minimum stator current control, maximizes the motor output torque under unit stator current. This control method reduces motor copper losses and improves the system power factor and capacity. Stator voltage-oriented control employs stator voltage-oriented vector control, using an outer loop of voltage and power, and an inner loop of stator current, to control the motor's output power and maximize wind energy capture. It utilizes optimal efficiency control, calculating the setpoint of ISD by relating the minimum values of motor iron losses, copper losses, mechanical losses, and converter losses to ISD. The controller then controls ISD and motor output power to achieve minimum motor losses and maximum maximum power point (MPPT). Meanwhile, some researchers have applied sensorless speed control strategies to wind turbines, proposing to use DC bus voltage for speed observation and directly control MPPT based on grid-side power.
4.3 Grid-side converter control strategy
Since the grid-side converter topology of wind turbines typically uses three-phase voltage-source PWM rectifiers, the control strategies of both can be mutually referenced. Currently, the main control strategies for PWM rectifiers are "indirect current control" and "direct current control," with the latter being widely used due to its fast current response and good robustness. Specific control strategies include linear controls such as vector control, state feedback control, and predictive current control, as well as nonlinear controls such as hysteresis current control, fuzzy control, and neural network control. Vector control is the most widely used, mainly including grid voltage-based control strategies and virtual flux-based control strategies. The former includes voltage-oriented control and direct power control, while the latter includes virtual flux-oriented control and virtual flux direct power control.
In high-power wind turbines, if traditional first-order inductor filtering is used in the grid-connected inverter, a large inductance is required to ensure filtering effectiveness, resulting in large size and high cost. Therefore, some researchers have studied the application of LCL filters in high-power wind turbines. LCL filters can maintain superior filtering performance with relatively small filter parameters, but their zero-impedance resonant point affects system stability, necessitating resonant damping control. Methods include passive damping and active damping. Passive damping attenuates the resonance effect by connecting a resistor in series with the capacitor; this method is stable and reliable, but introduces additional system losses. Active damping eliminates the resonance effect by modifying the control algorithm, such as using capacitor voltage lead correction combined with grid voltage and current dual feedforward compensation to achieve resonance damping. Capacitor current feedback is introduced, and a control strategy using a phase-lead filter to compensate for delay is proposed. Since direct power control lacks a current loop, some researchers have also studied damping control strategies based on direct power control. Damping algorithms generally require the addition of sensors, making research on sensorless damping algorithms practically significant. Active damping, especially damping methods for unbalanced control, has become a research hotspot.
Parallel operation of converters can increase the capacity of a single unit, which is currently a method to achieve high power at low voltage levels. However, differences in converter parameters will generate zero-sequence circulating current, which increases losses and may even damage the converter. Therefore, current sharing is necessary among parallel modules. Generally, circulating current can be suppressed by blocking the circulating current path, or by using multi-winding transformers for electrical isolation on the AC side. These traditional methods require additional components. Therefore, some researchers have studied the use of novel control algorithms to suppress zero-sequence circulating current. A zero-sequence current control strategy based on zero-axis current feedback control is proposed. Experimental results show that the algorithm can effectively suppress zero-sequence circulating current. Carrier phase-shift modulation technology is used to control the outer loop of the total current output and the inner loop of the circulating current, thereby suppressing circulating current and improving the dynamic performance of the system.
LVRT control strategy of 5D-PMSG
5.1 my country's LVT Standards
When a fault or disturbance occurs in the power grid, the voltage at the wind farm's grid connection point drops. However, within a certain voltage drop range, the turbine can still operate continuously connected to the grid and provide reactive power to support the grid's recovery. Figure 3 shows the low-voltage ride-through requirements for wind farms in my country's draft "Technical Regulations for Wind Farm Connection to the Power System" (2009 revised draft).
By comparing the LVRT standards of countries such as Denmark, Germany, and the United States, it can be seen that my country's requirements for the LVRT capability of wind farms are relatively low, and there is also a lack of regulations on reactive power compensation, peak shaving and frequency regulation capabilities, and protection configuration of wind farms.
5.2pmsg's LVRT control scheme
d-pmsg typically uses full-power back-to-back converters to isolate the wind turbine from the grid. The system also has flexible reactive power control capabilities, giving it a greater advantage in low-voltage ride-through capability compared to doubly-fed induction generators.
When a grid fault occurs in a DC-PMSG converter, due to the limited thermal capacity of the converter and the imbalance between input and output power in the DC link, excess energy charges the bus capacitor, leading to DC link overvoltage. Common methods to suppress the increase in bus voltage include adding a crowbar circuit to the DC link and limiting the electromagnetic torque of the generator.
An overvoltage protection circuit is added to the DC link, and during faults, the crowbar circuit absorbs excess energy. Several DC-side crowbar circuits are introduced and compared, including those with added unloading circuits, buck circuits, and auxiliary converters. These methods all add extra components, causing problems in system design and installation, and also increase the control algorithm, affecting the overall performance of the unit. A method is proposed to limit the power input to the DC link by limiting the generator's electromagnetic torque during grid faults, thereby achieving low-voltage ride-through of the unit. This method transfers the power imbalance in the DC link between the wind turbine and the generator, utilizing the buffer energy storage effect of rotating components. However, with the wind turbine's output power remaining constant, this will cause the generator to accelerate, thereby triggering the wind turbine's pitch controller to limit wind energy capture.
Currently, some scholars have studied the reactive power support capability of converters to the grid during grid faults. When the grid voltage drops, the grid-side converter operates in statcom mode, redistributing the setpoints of active and reactive currents according to the magnitude of the voltage drop. However, because the converter's output current must simultaneously maintain the stability of the DC bus voltage, the reactive power that the converter can provide to the grid is limited, thus not fully demonstrating the advantages of a full-power converter. Therefore, some scholars have studied how grid-side converters can provide rated current to output reactive power under grid faults, employing a novel control strategy. The generator-side converter controls the DC bus voltage and the motor stator voltage, while the grid-side converter controls the active and reactive power flowing to the grid, thereby ensuring maximum reactive power support to the grid.
6 Research on Grid-Connected Operation Technology of Wind Farms
With an increasing number of wind turbines being connected to the grid, fluctuations in their output power and the inherent characteristics of the turbines are impacting the power quality of the grid. This impacts issues such as voltage deviation, voltage fluctuation, flicker, frequency deviation, harmonic pollution from the public grid, and grid imbalance. It also affects grid stability, peak shaving, frequency regulation, and grid dispatching. Therefore, it is necessary to conduct power quality and stability analyses on the grid after wind farms are connected to ensure compliance with IEC and national standards.
Currently, when wind farms are connected to the grid, reactive power compensation devices are generally installed to suppress voltage fluctuations and flicker in order to improve power quality. These devices include parallel capacitor banks, static var compensators (SVCs), and active power filters. While this method can quickly compensate for reactive power and maintain voltage stability at the wind farm connection point, it cannot regulate the active power output of the wind farm. Therefore, some scholars have researched using energy storage systems to control the active power output of wind power generation. For example, combining a battery energy storage system (BES) with wind power generation units can reduce the impact of wind farm output fluctuations on the grid. Using supercapacitor banks as energy storage devices, this paper introduces a wind farm power regulation system and proposes energy management control strategies to effectively suppress active and reactive power fluctuations in wind farms. Flexible AC transmission systems are also being applied to wind farms to improve their power quality.
As the proportion of wind power capacity in the power system increases, the instability of wind farm output has a significant impact on the power system. For example, the reactive power characteristics of wind farms may cause voltage instability or even voltage collapse; when wind farms form an isolated grid with other power generation methods, they can affect the stability of the grid frequency; at the same time, the power injected by wind farms changes the original power flow distribution and power direction of the system, affecting system stability. The aforementioned reactive power compensation devices can solve these problems, and system stability can also be improved through methods such as optimizing reactive power flow distribution and strengthening the grid structure.
Wind farm power prediction is of great significance to the operation of the power system and is conducive to the formulation of reasonable dispatch plans. Current prediction methods are mostly based on time series analysis, artificial neural networks, and Kalman filtering algorithms. For example, the WPMS system developed by IST in Germany uses a neural network method, and its root mean square error (RMSE) is 7%–19% of the installed capacity. These power prediction methods are based on statistical principles and require sample data, making power prediction for newly built wind farms difficult. In recent years, many scholars at home and abroad have adopted wind power prediction methods based on physical principles, such as the Previento system developed by the University of Oldenburg in Germany, whose RMSE is less than 10% of the installed capacity. Currently, there are no similar system application examples in China, but research on the physical methods of wind power prediction and the proposal of prediction methods suitable for power grids provide a reference for engineering applications.
7. Conclusion
This article details the control technologies related to direct-drive permanent magnet wind power generation systems, focusing on the control strategies of full-power converters. It also analyzes the low-voltage ride-through of the unit and the grid-connected operation technology of the wind farm. It can be seen that many scholars at home and abroad have conducted extensive research on this topic, providing theoretical support for its better promotion and application.
