Reliable And Robust Predictive Control Of Power Converters In Offshore Wind Energy System

Accelerating the development of renewable energy is essential in anchoring the goal of peak carbon dioxide emissions and carbon neutrality and promoting the construction of a clean,safe.and efficient energy system.Offshore wind power has become one of the most promising forms of new energy generation due to its abundant reserves,high power density,long generation time,and lack of topographic constraints.The energy transfer of offshore wind power involves several processes.such as wind energy capture,energy conversion,and transmission to the grid.Supported by the development of power semiconductor devices and circuit topology theory,power electronic variable current equipment has become the basis for achieving these functions,while the high-quality control of variable current equipment is the core technology for improving the reliability and overall energy efficiency of offshore wind power systems on the grid.However,due to the harsh offshore environment and high operation and maintenance costs,existing variable current equipment and control schemes have poor reliability and parameter mismatch,resulting in short grid connection times and low wind energy utilization of wind turbines.In addition,the stable interconnection of large-scale offshore wind turbines to the grid faces many challenges,reducing their grid reliability and energy conversion efficiency.These problems have become an important bottleneck for offshore wind power’s safe and efficient grid connection.In this paper,we focus on the theory and control technology of power conversion in offshore wind power converter systems,and investigate(ⅰ)motor-side converters,(ⅱ)backto-back unit converters,and(ⅲ)modular multilevel converters in the order of power transfer in the system.The content and contributions of this paper are summarised and summarised below.(1)To improve the reliability of the motor-side converter,this paper proposes a faulttolerant predictive control strategy for motor-side converter phase-breaking faults based on an expansive state observer for motor and converter phase-breaking faults.The harsh offshore conditions lead to a high motor and converter phase failure incidence.The traditional faulttolerant scheme superimposes an appropriate zero-sequence voltage on the original magnetic field directional controller by feed-forward and uses the remaining phases to maintain regular operation.However,the zero sequence voltage solution strongly depends on the motor zero sequence model and is difficult to obtain accurately.For this reason,this paper uses an expansive state observer to achieve an accurate solution for the zero sequence quantities and designs the observer parameters in conjunction with stability analysis.At the same time,this paper proposes a superposition method of the above zero-sequence voltage in the model prediction controller to quickly drive the speed and torque of the motor after a fault to converge around the reference value with the physical limit output of the converter.Test results show that the proposed method can effectively achieve fault-tolerant machine operation under phase failure,and the technique is desensitized to the zero-sequence parameters.Compared to traditional linear control methods,the proposed method has a fast dynamic response and high reliability.(2)To cope with the problem of easy parameter mismatch in unit converters and improve their control robustness,this paper proposes a model-free predictive control strategy based on a fast update mechanism of the state variable look-up table for back-to-back converters.Classical model predictive control relies on the model to estimate the system state,and parameter mismatch will result in the controller being unable to predict the system trajectory correctly.The promising idea is to use historical data to construct look-up tables to predict the system’s future state.However,due to the single-cycle,single-switch nature of finite-set predictive control,this idea suffers from the problem of stagnant look-up table updates.In this paper,a new model-free predictive control method is proposed.The method uses the relationship between the small signal model and the converter output vector to design a calculation method that accurately updates all elements of the look-up table using only the first two control cycles sampled.Based on a fully updated look-up table,the proposed method can correctly predict the system state without needing model parameters,achieving the same performance as classical model predictive control.Comparative test results demonstrate the robustness of the proposed method under a wide range of parameter mismatch conditions.The proposed method also improves the update frequency of the system state look-up table compared to existing model-free methods based on the accurate calculation of state quantity changes.(3)To enhance the unit converter’s low voltage ride-through capability and improve the unit converter grid connection reliability,this paper proposes a dynamic cascade model predictive control strategy for back-to-back converters.Rotor energy storage in conjunction with energy storage units is an auspicious method for dealing with LVRT.However,the traditional control scheme has two limitations:the conventional model predictive control based on fixed weight factors is challenging to cope with changes in the priority of the control target at different stages;the rotor energy storage leads to a mismatch of the considerable inertia turbine speed concerning the random wind speed,causing large fluctuations in speed and DC bus voltage.To address these problems,this paper proposes a model predictive control scheme with a dynamic cascade architecture that prioritizes the control targets and adjusts the number of candidate switching vectors at each stage.Under this architecture,a speed prediction cost function with a power prediction-based supercapacitor control method is designed in this paper.The proposed method avoids the cumbersome process of designing the weighting factors while ensuring control performance and suppressing speed and DC bus voltage fluctuations.The technique can effectively adapt to changes in the priority of control objectives at each stage for low-voltage ride-through conditions of the unit.(4)To reduce the circulating losses of the converters and improve energy conversion efficiency,this paper proposes a direct predictive circulating current control strategy for modular multilevel converters.The modular multilevel converter has an internal circulating current that flows only between three phases and is dominated by a two-fold fundamental frequency component.Traditional linear control and conventional model predictive control often achieve circulating current suppression by cascading circulating current control loops or changing the total number of single-phase sub-module inputs,which suffers from nested cascade structures that reduce dynamic performance and the difficulty of commissioning a large number of linear controllers.In this paper,a three-phase zero-sequence model of the converter is introduced into the circulating current analysis,based on which the idea of directly predicting and controlling the zero-sequence circulating current is proposed,and the synthesis method of the zero-sequence vector and the effective control vector is designed.Test results show that the proposed method achieves fast and accurate tracking of each target reference value of the modular multilevel converter while effectively suppressing the circulating current.The method only needs to determine the direction of the circulating current,avoiding the influence of system parameters such as the bridge arm inductance,and is a robust circulating current suppression strategy.The influence of control parameters on the effect of vector synthesis is analyzed in the paper,which has reasonably practical engineering guidance significance.