A power supply system for wind turbine yaw motor control

Abstract

This invention discloses a power supply system for wind turbine yaw motor control, specifically relating to the field of wind turbine yaw motor power supply control. It addresses the stability issues of bus voltage drop and energy surge-triggered servo shutdown during simultaneous startup of multiple motors. The system employs wind prediction to rearrange the startup sequence, pre-charges the isolation converter to create a flexible bus, and adaptively adjusts the current limiting slope using a startup stability coefficient trained with both voltage drop rate and peak width duration as parameters, dynamically waking up the motors in segments. Subsequently, energy is reinjected synchronously to replenish the energy storage capacitor and smooth subsequent voltage fluctuations. The terminal feeds regenerated energy to low-voltage loads in the nacelle according to priority and resets the threshold. The entire process completes a closed loop of energy storage, surge reduction, and residual energy utilization within a single wind direction switching cycle, maintaining a safe bus voltage drop window, significantly reducing the risk of drive shutdown, shortening yaw response time, decreasing tower fatigue peak, while simultaneously improving regenerated energy utilization and reducing heat dissipation pressure.

Description

Technical Field

[0001] This invention relates to the field of power supply control for wind turbine yaw motors, and more specifically, to a power supply system for controlling wind turbine yaw motors. Background Technology

[0002] In offshore wind farms, the nacelle relies on multiple yaw motors working together to achieve rapid wind accretion. The existing publicly available solution, "Wind Power Yaw Servo Drive System and Control Method Based on Vector Control" (application number CN115929542A), directly connects the U / V / W three-phase terminals of all yaw motors in parallel to the output of the same servo driver, thus making the DC bus the sole power supply channel. The servo performs vector control based on the resolver signals of each motor, thereby maintaining consistent speed during normal operation. This structure operates smoothly under stable wind conditions, meeting the requirement for uniform yaw.

[0003] When a sudden crosswind occurs, the main controller issues a start command all at once, causing the torque of multiple motors to rise synchronously, and the peak current instantaneously converges into a single bus. Due to a lag in response, the DC link capacitor cannot replenish energy in time, causing a sudden drop in bus voltage. The servo driver triggers undervoltage or overcurrent self-protection and stops output. The nacelle cannot continue to turn towards the windward side, and the rotor, deviating from the prevailing wind direction, generates periodic torque pulses, exacerbating tower fatigue. The root cause lies in the fact that the existing topology only focuses on smoothing the output current, lacking segmented energy storage and transient current limiting design on the bus side; the energy peak is not buffered and directly impacts the bus, triggering protection logic and forming a chain fault of "multiple start-up-voltage drop-shutdown".

[0004] To address the aforementioned problems, a technical solution is provided. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a power supply system for wind turbine yaw motor control. This system reorders the start-up sequence based on wind direction prediction, pre-charges the isolation converter to build a flexible busbar, and adaptively adjusts the current-limiting slope using a start-up stability coefficient trained with both voltage drop rate and peak width duration as parameters, dynamically waking up the motor in segments. Subsequently, energy is reinjected to simultaneously replenish the energy storage capacitor and smooth subsequent voltage fluctuations. The terminal feeds regenerated energy to low-voltage loads in the nacelle according to priority and resets the threshold. The entire process completes a closed loop of energy storage, impact reduction, and residual energy utilization within a single wind direction switching cycle, maintaining a safe voltage drop on the busbar, significantly reducing the risk of drive shutdown, shortening the yaw response time, and decreasing tower fatigue peak. Simultaneously, it improves regenerated energy utilization and reduces heat dissipation pressure, thus solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] Sequence generation module: Real-time monitoring of wind direction changes, acquisition of motor no-load current and inertia, and generation of segmented start-up sequences in descending order of inertia;

[0008] Energy pre-charge module: Utilizes an isolated bidirectional converter to pre-charge the energy storage capacitors of each section to the target energy level according to the startup sequence, and injects the initial energy threshold into the DC bus to ensure energy preparation before the motor soft start;

[0009] Dynamic current limiting module: Triggers the first stage of the motor and uses ramp current limiting logic to constrain the starting current. During startup, it synchronously acquires the voltage drop rate characteristics and peak width duration characteristics, and uses the results of preset model analysis to dynamically adjust the current limiting slope and the startup sequence of the next stage.

[0010] Energy recovery module: When the bus voltage recovers to the upper edge of the recovery threshold, the next stage of the motor is triggered, and the regenerative energy generated in the previous acceleration stage is injected into the energy storage capacitor through the bidirectional converter to reduce the second voltage fluctuation;

[0011] Cyclic Replenishment Module: After yaw adjustment, the isolated bidirectional converter takes over the brake regenerative energy and supplies it to the cabin auxiliary loads according to priority. Then, the energy storage capacitor is reset to the target energy threshold to retain a stable energy reserve for the next round of startup.

[0012] In a preferred embodiment, the sequence generation module includes the following:

[0013] Monitor the wind direction and nacelle orientation, calculate the deviation between the two and compare it with a preset threshold to determine whether yaw adjustment is needed. If the necessity of adjustment is confirmed, obtain the no-load current and inertia parameters of each yaw motor, arrange all yaw motors in descending order of inertia value and divide them into multiple segments to balance the total inertia of each segment, generate a start-up sequence to activate each segment in sequence, starting from the segment containing the motor with the highest inertia value.

[0014] In a preferred embodiment, the energy pre-charging module includes the following components:

[0015] The starting energy requirement of each motor segment is calculated based on the motor inertia and target speed. The starting energy requirement is adjusted by the energy redundancy coefficient to set the pre-charge target energy of each energy storage capacitor segment. The pre-charge target voltage of each energy storage capacitor segment is calculated based on the energy storage capacitor capacity and the pre-charge target energy. Each energy storage capacitor segment is pre-charged according to the starting sequence and the energy storage capacitor voltage is monitored in real time until the pre-charge target voltage is reached. The initial energy threshold required by the DC bus is calculated based on the starting energy requirement of the first motor segment and the amplification factor. Energy is injected into the DC bus through an isolated bidirectional converter and the DC bus voltage is adjusted to support motor starting.

[0016] In a preferred embodiment, the energy pre-charging module further includes the following:

[0017] The energy redundancy factor is a value greater than 1, used to calculate the pre-charge target energy of each segment of the energy storage capacitor by multiplying the starting energy requirement of each segment of the motor by the energy redundancy factor.

[0018] In a preferred embodiment, the dynamic current limiting module includes the following:

[0019] The smoothness of motor starting is evaluated by real-time monitoring of the instantaneous voltage attenuation ratio of the bus voltage and the peak width of the starting current. The instantaneous voltage attenuation ratio is obtained by collecting the bus voltage curve and calculating the difference between the lowest point and the starting point, divided by the recovery time. The peak width of the starting current is obtained by extracting the peak segment and measuring the half-peak width, which is then normalized to the rated current of the motor.

[0020] In a preferred embodiment, the dynamic current limiting module further includes the following:

[0021] The instantaneous voltage attenuation ratio and the peak width of the current wave are input into a pre-trained random forest regression model to calculate the start-up stability coefficient. Based on the comparison between the start-up stability coefficient and the preset safety threshold, the current limiting slope and the start-up sequence of the next stage of the motor are dynamically adjusted to ensure that the bus voltage fluctuation is controlled within the predetermined range.

[0022] In a preferred embodiment, the energy reinjection module includes the following:

[0023] The DC bus voltage is monitored, and the next motor is triggered to start when the DC bus voltage recovers to the upper edge of the preset recovery threshold. At the same time, the regenerative energy generated by the acceleration of the previous motor is injected into the energy storage capacitor through the isolated bidirectional converter to support the start of the next motor and reduce the fluctuation of DC bus voltage.

[0024] In a preferred embodiment, the energy reinjection module further includes the following:

[0025] The injected power of regenerative energy is determined by the energy transmission efficiency and the injection time, and the starting energy provided by the energy storage capacitor takes priority over the DC bus in order to reduce the extraction of energy from the DC bus.

[0026] In a preferred embodiment, the cyclic replenishment module includes the following:

[0027] After the yaw adjustment is completed, the regenerative energy generated by the yaw motor during the braking phase is captured by the bidirectional converter and transmitted to the DC bus through the bidirectional converter. At the same time, the DC bus voltage is monitored to prevent the voltage from exceeding the preset upper limit. Then, according to the preset auxiliary load priority sequence and the real-time monitored auxiliary load power demand, the regenerative energy is distributed to the auxiliary loads in the cabin in priority order to ensure that high priority loads receive power support first.

[0028] In a preferred embodiment, the cyclic replenishment module further includes the following:

[0029] After allocation, the current energy level of each energy storage capacitor is assessed, and energy is replenished from the DC bus to the preset target energy threshold through a bidirectional converter to ensure that the energy storage capacitor provides sufficient energy reserves for the next yaw adjustment.

[0030] The technical effects and advantages of the power supply system for wind turbine yaw motor control according to the present invention are as follows:

[0031] This invention first rearranges the startup sequence based on wind forecast, pre-charges the isolation converter to build a flexible busbar, and then adaptively adjusts the current limiting slope using a startup stability coefficient trained with both voltage drop rate and peak width duration to dynamically wake up the motor in segments. Subsequently, energy reinjection synchronously replenishes the energy storage capacitor and smooths subsequent voltage fluctuations. The terminal feeds regenerated energy to the low-pressure load in the cabin according to priority and resets the threshold. The entire process completes a closed loop of energy storage, impact reduction, and surplus energy utilization within one wind direction switching cycle, maintaining a safe window for busbar voltage drop, significantly reducing the risk of drive shutdown, shortening yaw response time, and decreasing tower fatigue peak. At the same time, it improves the utilization rate of regenerated energy and reduces heat dissipation pressure, thereby maintaining continuous wind resistance capability in the scenario of multiple startups triggered by extreme crosswinds, reducing the number of unplanned shutdowns, extending the life of capacitors and power devices, and building a self-circulating energy supply chain for auxiliary loads. Overall power generation revenue and operation and maintenance efficiency are improved simultaneously, while reducing the pressure of offshore monitoring and enhancing the predictability of equipment operating conditions. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of a power supply system for controlling a wind turbine yaw motor according to the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Example 1: Figure 1 The present invention provides a power supply system for wind turbine yaw motor control, comprising:

[0035] Sequence generation module: Real-time monitoring of wind direction changes, acquisition of motor no-load current and inertia, and generation of segmented start-up sequences in descending order of inertia;

[0036] Energy pre-charge module: Utilizes an isolated bidirectional converter to pre-charge the energy storage capacitors of each section to the target energy level according to the startup sequence, and injects the initial energy threshold into the DC bus to ensure energy preparation before the motor soft start;

[0037] Dynamic current limiting module: Triggers the first stage of the motor and uses ramp current limiting logic to constrain the starting current. During startup, it synchronously acquires the voltage drop rate characteristics and peak width duration characteristics, and uses the results of preset model analysis to dynamically adjust the current limiting slope and the startup sequence of the next stage.

[0038] Energy recovery module: When the bus voltage recovers to the upper edge of the recovery threshold, the next stage of the motor is triggered, and the regenerative energy generated in the previous acceleration stage is injected into the energy storage capacitor through the bidirectional converter to reduce the second voltage fluctuation;

[0039] Cyclic Replenishment Module: After yaw adjustment, the isolated bidirectional converter takes over the brake regenerative energy and supplies it to the cabin auxiliary loads according to priority. Then, the energy storage capacitor is reset to the target energy threshold to retain a stable energy reserve for the next round of startup.

[0040] In offshore wind farms, wind power systems require multiple yaw motors working in tandem to quickly align with the wind direction, ensuring the rotor efficiently captures wind energy. In existing technology, the U / V / W three-phase terminals of all yaw motors are directly connected in parallel to the output of the same servo driver, with the DC bus serving as the sole power supply channel. This topology, under stable wind conditions, maintains consistent motor speeds through vector control, meeting the requirement for uniform yaw. However, when a sudden crosswind occurs, the main control system issues a start command all at once, causing the torque of multiple motors to surge synchronously, with peak currents instantaneously superimposed on the DC bus. Due to the lag in the DC link capacitor response, energy cannot be replenished in time, leading to a sudden drop in DC bus voltage. This triggers the undervoltage or overcurrent protection mechanism of the servo driver, causing it to stop outputting. As a result, the nacelle cannot continue to turn towards the windward side, and the rotor, deviating from the prevailing wind direction, generates periodic torque pulses, exacerbating tower fatigue and potentially affecting equipment lifespan. The root of the problem lies in the lack of segmented energy storage and transient current limiting mechanisms on the bus side in the existing design. The energy peak is not effectively diverted and buffered, directly impacting the bus and triggering a chain reaction of failures.

[0041] To address this issue, this invention proposes a power supply system for wind turbine yaw motor control. It optimizes the motor startup process through segmented startup, energy dispatch, and adaptive regulation, reducing the impact of peak current on the bus and ensuring stable bus voltage. This answer focuses on the specific processing logic of the sequence generation module, specifically how to generate a segmented startup sequence based on real-time wind direction changes and motor parameters, laying the foundation for subsequent energy management and motor startup control. The sequence generation module is the starting point of the entire solution; the sequence it generates directly affects the energy distribution and impact control effects of subsequent steps.

[0042] The purpose of the sequence generation module is to detect changes in the incoming wind angle in real time, synchronously read the no-load current and inertia parameters of all yaw motors, and generate a segmented start-up sequence according to the inertia from largest to smallest, providing a basis for subsequent energy scheduling. The following is a detailed breakdown of the technical logic:

[0043] S1.1: Real-time detection of changes in the incoming wind angle.

[0044] The cabin needs to adjust its orientation according to changes in wind direction to maximize wind energy capture. To do this, the system first continuously monitors the current direction angle of the wind using a wind direction sensor, recording it as the incoming wind direction angle. This angle is expressed in degrees, and may be a value measured at a specific moment. Next, the system obtains the current orientation angle of the cabin, also recorded in degrees, called the cabin orientation angle. To determine if the two are consistent, the deviation between the incoming wind direction angle and the cabin orientation angle needs to be calculated. The specific calculation method is: subtract the cabin orientation angle from the incoming wind direction angle, and take the absolute value to obtain the deviation angle, still in degrees.

[0045] To facilitate subsequent decision-making, the system pre-sets a deviation threshold angle, in degrees, for example, 5 degrees. This threshold indicates that when the deviation angle exceeds this value, it signifies that the wind direction change is sufficient to affect power generation efficiency, requiring adjustment of the nacelle orientation. After calculation, the deviation angle is compared with the deviation threshold angle. If the deviation angle is greater than the deviation threshold angle, the wind direction change is deemed significant, and further processing is required; if the deviation angle is less than or equal to the deviation threshold angle, the system maintains its current state and continues to monitor wind direction changes.

[0046] This process ensures that adjustments are only triggered when wind direction changes reach a certain level through precise numerical comparisons, avoiding frequent and meaningless actions, thereby improving system efficiency and equipment lifespan.

[0047] S1.2: Read the no-load current and inertia parameters of the yaw motor.

[0048] Once it is confirmed that the nacelle orientation needs to be adjusted, the system needs to obtain the key parameters of the yaw motors involved in the adjustment to support the generation of subsequent start-up sequences. Assume there are multiple yaw motors in the wind power system, denoted as the total number n, and numbered as the first motor, the second motor, and so on up to the nth motor. Each motor has two key parameters: no-load current and moment of inertia.

[0049] For each motor, the no-load current is expressed in amperes, referring to the current value of the motor when it is running without load. This value can be directly read from a pre-stored database or measured in real time using a current sensor. This parameter reflects the basic electrical energy consumption of the motor during operation. The moment of inertia, expressed in kilograms multiplied by square meters, measures the dynamic response characteristics of the motor during startup and acceleration. It can also be extracted from the database or obtained through measurement. Taking the first motor as an example, its no-load current is denoted as the first no-load current, and its moment of inertia as the first moment of inertia; the second motor corresponds to the second no-load current and the second moment of inertia, and so on, up to the nth motor.

[0050] The system pairs the no-load current and moment of inertia of each motor. For example, the parameter pair for the first motor is the combination of the first no-load current and the first moment of inertia. All the parameter pairs are aggregated to form an initial dataset containing n parameter pairs, corresponding to the first through nth motors. This dataset provides the necessary information for subsequent processing, ensuring that the generation of the startup sequence can be optimized based on the actual physical characteristics of the motors.

[0051] S1.3: Generate segmented startup sequence.

[0052] After acquiring the parameters of all yaw motors, the system sorts and groups the motors according to their moment of inertia to generate a segmented startup sequence and optimize the startup process. The process consists of three stages: sorting, segmentation, and sequence generation.

[0053] Sorting Phase: Extract the moment of inertia value of each motor from the initial dataset, such as the first moment of inertia of the first motor, the second moment of inertia of the second motor, etc. Rearrange these moment of inertia values ​​in descending order to form a new motor sequence. Assuming there are three motors in the system with moments of inertia of 5 kg / m², 3 kg / m², and 8 kg / m², the sorted sequence would be: the motor with moment of inertia of 8 is first, the motor with moment of inertia of 5 is second, and the motor with moment of inertia of 3 is third. The motor numbers in the new sequence are adjusted accordingly, denoted as the first sorted motor, the second sorted motor, and so on, up to the nth sorted motor.

[0054] Segmentation Phase: The sorted n motors are divided into several segments, denoted as m, where m is less than or equal to n. Each segment contains one or more motors. To balance energy distribution, the sum of the moments of inertia of all motors is first calculated: the moment of inertia of the first-ordered motor plus the moment of inertia of the second-ordered motor, and so on, up to the moment of inertia of the nth-ordered motor, yielding the total moment of inertia. The total moment of inertia is then divided by the segment number m to calculate the target average moment of inertia for each segment. Next, starting with the first-ordered motor, the motors are sequentially assigned to each segment, ensuring that the sum of the moments of inertia of each segment is as close as possible to the target average. For example, the first segment might contain several motors with the largest moments of inertia, the second segment might contain several motors with the second largest moments, and so on. The moment of inertia of each segment is the sum of the moments of inertia of all motors within that segment, denoted as the moment of inertia of the first segment, the moment of inertia of the second segment, and so on.

[0055] Sequence Generation Stage: Based on the segmentation results, a startup sequence is generated. The sequence is arranged sequentially by segment, with the first segment representing the first group of motors to start, the second segment representing the second group, and so on up to the m-th segment. Motors within each segment start simultaneously, and motors in each segment start sequentially according to time. The final output contains a startup sequence of m segments, used to guide the actual control of the yaw motor.

[0056] This segmented sequencing method prioritizes starting motors with larger rotational inertia and breaks down the starting process into multiple segments, reducing the energy demand for a single start-up, thereby minimizing the impact on system voltage and improving overall operational stability.

[0057] The sequence generation module begins by detecting the deviation between the oncoming wind direction angle and the nacelle orientation angle in real time. By comparing this deviation with preset thresholds, it accurately identifies adjustment needs. Next, the system collects the no-load current and moment of inertia of all yaw motors, forming a complete dataset. Based on this, by sorting and segmenting the moments of inertia, an optimized startup sequence is generated. This process ensures that every step from wind direction change detection to motor startup control is based on accurate data and calculations, providing a scientific basis for nacelle orientation adjustments. Simultaneously, segmented control optimizes energy efficiency and enhances system stability and reliability.

[0058] The sequence generation module has generated a segmented startup sequence based on real-time wind direction changes and motor inertia, laying the foundation for subsequent energy scheduling. However, to achieve smooth motor startup and avoid bus voltage fluctuations, it is necessary to pre-store energy for each segment of the motor and provide initial energy support to the DC bus before startup. The energy pre-charging module is responsible for this critical task. It performs pre-charging of the energy storage capacitors of each segment through an isolated bidirectional converter, and injects an initial energy threshold into the DC bus after pre-charging, laying the energy foundation for the soft start of the first segment of the motor and ensuring that the entire startup process is smooth and controllable.

[0059] The purpose of the energy pre-charge module is to pre-charge the energy storage capacitors in each section before the motor starts and to provide initial energy to the DC bus to support the subsequent soft-start process. The following is a detailed breakdown of the technical logic:

[0060] S2.1: Determine the pre-charge target.

[0061] Starting a yaw motor requires sufficient energy, which is mainly supplied by the DC bus and the energy storage capacitors in each section. In order to meet the starting requirements of different sections of the motor and reduce the instantaneous impact on the DC bus, a reasonable pre-charge energy level must be set for each section of the energy storage capacitor to provide local energy reserves.

[0062] The processing begins with the startup sequence generated by the sequence generation module. The startup sequence is a list containing multiple segments, each containing a group of motors arranged in descending order of motor inertia. For each segment, calculating its startup energy requirement is the primary task. The startup energy requirement depends on the total inertia of all motors within that segment and the target speed. The total inertia is the sum of the inertia of all motors within the segment, while the target speed is the speed the motors need to reach after startup. The startup energy requirement is calculated using a kinetic energy calculation method: multiplying the total inertia of the segment by the square of the target speed, and then taking half of the result, yields the energy value required to start the segment.

[0063] To ensure that the pre-charge energy not only meets the startup requirements but also compensates for losses during energy transmission, the startup energy requirement needs to be multiplied by an energy redundancy coefficient greater than 1 to obtain the target pre-charge energy for each energy storage capacitor segment. The value of this coefficient is determined based on the actual system conditions, typically considering factors such as energy transmission efficiency and system stability. The capacity of each energy storage capacitor segment is a known fixed value, measured in farads. The energy stored in the energy storage capacitor is determined by its capacity and voltage. The calculation method is to multiply the capacity by the square of the voltage and then take half of the result. Based on this relationship, the process of calculating the target pre-charge voltage is as follows: multiply the target pre-charge energy by 2, divide by the capacity of the energy storage capacitor to obtain an intermediate value, and then take the square root of this intermediate value to obtain the target pre-charge voltage.

[0064] The energy redundancy coefficient, in the context of wind turbine yaw motor control power supply systems, ensures that the pre-charged energy of the energy storage capacitor is sufficient to meet the motor's starting requirements and cope with energy transmission losses and system uncertainties. When calculating the target pre-charge energy for each segment of the energy storage capacitor, the starting energy requirement of that segment of the motor is multiplied by a coefficient greater than 1. This coefficient not only guarantees the basic energy required for motor starting but also provides redundancy for energy losses and unpredictable factors during system operation, thereby improving system reliability and stability. Its practical basis comes from energy management strategies in actual engineering projects. In wind power systems, various losses occur during energy transmission, including efficiency losses when converting electrical energy to mechanical energy, resistance losses in circuits, and delays and errors in the control system. If the pre-charged energy is calculated only based on theoretical requirements, insufficient energy may occur due to losses or extreme conditions, leading to starting failure or system instability. Therefore, the introduction of the energy redundancy coefficient aims to ensure the normal operation of the system under various working conditions through engineering practice, and its necessity has been verified in practical applications. The determination of the numerical value relies on the comprehensive results of system simulation, historical data analysis, and field testing. Specifically, by statistically analyzing wind power system operating data, the energy loss distribution under different wind conditions and motor loads can be quantified, thereby deriving a reasonable redundancy range. Simultaneously, the system design considers worst-case energy demands, such as extreme wind speeds or sudden load changes, to ensure system robustness. In practice, the value of this coefficient is adjusted based on factors such as system configuration, motor characteristics, and wind farm environment, typically ranging from 1.1 to 1.5. This range strikes a balance between energy utilization efficiency and system stability, avoiding excessive redundancy that leads to resource waste, or insufficient redundancy that negatively impacts performance. The introduction of the energy redundancy coefficient provides a buffer for energy fluctuations during motor startup, enhancing the smooth operation and high efficiency of the wind turbine yaw motor control system. This meets the requirement of full disclosure of the invention under patent law and avoids uncertainty in the claims.

[0065] Calculating the energy redundancy factor typically requires first determining the basic energy demand, then setting the redundancy ratio based on losses and uncertainties, and finally deriving the factor. Let's illustrate this with a hypothetical example: Assume a wind turbine yaw motor has an inertia of 10 kg·m² and a target rotational speed of 10 rad / s. The basic energy demand is calculated using the formula... Calculated as Next, considering the system characteristics, assuming an energy transfer efficiency of 90% (i.e., a loss of 10%), we need to divide by the efficiency to obtain the adjustment factor. Additionally, to cope with wind speed fluctuations, an extra 20% redundancy is added, which is multiplied by 1.2. Combining these two parts, the redundancy ratio is... Finally, the pre-charge target energy is... The energy redundancy factor is 1.333. This process calculates the base energy and adjusts it according to losses and uncertainties to ensure that the pre-charged energy meets actual needs and has a margin.

[0066] In this way, the pre-charge target energy and voltage of each energy storage capacitor can be accurately determined, providing a clear control basis for subsequent energy transfer, while ensuring that energy distribution matches actual needs and improving the system's operating efficiency.

[0067] S2.2: Perform pre-charging of the energy storage capacitor.

[0068] After determining the pre-charge target for each energy storage capacitor segment, energy needs to be transferred from the DC bus to each segment. If all segments are pre-charged simultaneously, it may cause excessive load on the DC bus. Therefore, pre-charging must be completed step by step according to the startup sequence to ensure process control and optimize energy use.

[0069] In practice, following the startup sequence generated in the sequence generation module, energy is sequentially supplied to the energy storage capacitors of each segment, starting from the first segment. Energy transfer is achieved through an isolated bidirectional converter, which can directionally deliver energy from the DC bus to the designated energy storage capacitors. The pre-charge power setting must comprehensively consider the converter's power limit and the DC bus's stability requirements to ensure that energy delivery is both rapid and does not cause voltage fluctuations. During the pre-charge process, the voltage value of the energy storage capacitors is monitored in real time. When the voltage of a certain segment's energy storage capacitor reaches or exceeds its pre-charge target voltage, energy delivery to that segment is stopped, and pre-charging begins for the next segment's energy storage capacitors. This process is repeated until all segments' energy storage capacitors in the startup sequence have reached their respective pre-charge target voltages.

[0070] By pre-charging sequentially, each energy storage capacitor obtains independent energy reserves before startup. This decentralized energy support method effectively reduces the load pressure on the DC bus during motor startup, while improving energy utilization efficiency and providing stable conditions for subsequent motor control.

[0071] S2.3: Initial energy threshold for injecting into the DC bus.

[0072] The startup of the first stage motor will directly consume energy from the DC bus, which may cause a sudden voltage drop and affect system stability. To address this, after pre-charging all the energy storage capacitors, it is necessary to increase the energy level of the DC bus to provide additional energy support for the startup of the first stage motor.

[0073] First, calculate the initial energy threshold required for the DC bus. The initial energy threshold is calculated by multiplying the starting energy requirement of the first motor segment by an amplification factor greater than 1. The starting energy requirement of the first motor segment has already been calculated in sub-step one, i.e., multiplying the total inertia of that segment by the square of the target speed, and then taking half of the result. The specific value of the amplification factor is determined based on the system design, typically considering the energy surge during startup and the bus stability requirements, ensuring that the injected energy not only offsets the energy consumption during startup but also retains a certain margin. Energy injection is accomplished through an isolated bidirectional converter or other energy source (such as a backup power supply), transferring the calculated initial energy threshold to the DC bus.

[0074] The DC bus capacitor's capacitance is a known value, and the current bus voltage is obtained through real-time monitoring. To calculate the voltage increase after energy injection, the initial energy threshold is first multiplied by 2, then divided by the DC bus capacitor's capacitance to obtain an intermediate value. The square root of this intermediate value is then taken to obtain the theoretical voltage value after injection. Subtracting the theoretical voltage value from the current bus voltage gives the voltage increase. The isolated bidirectional converter is then used to adjust the DC bus voltage, increasing it from its current value to the sum of the current value and the voltage increase.

[0075] By increasing the energy level of the DC bus, the system can effectively resist the energy surge during the first stage of motor startup, ensure voltage stability, provide reliable energy support for the subsequent soft-start process, and enhance the overall operational stability.

[0076] The energy pre-charging module's processing is closely integrated with the sequence generation module's output, namely the segmented startup sequence. Based on this sequence, the startup energy requirement for each motor segment is first calculated, and the pre-charging target energy and voltage for each energy storage capacitor segment are determined accordingly. Subsequently, following the sequence order, each energy storage capacitor segment is pre-charged sequentially via an isolated bidirectional converter, ensuring sufficient local energy reserves before motor startup. Finally, by injecting an initial energy threshold into the DC bus, the bus voltage is increased, providing additional energy support for the startup of the first motor segment. The entire process, through precise energy calculation and orderly pre-charging operations, optimizes energy allocation and utilization efficiency, ensuring the smooth operation and high-efficiency performance of the wind turbine yaw motor control system during startup, creating favorable conditions for subsequent steps.

[0077] The energy pre-charging module utilizes an isolated bidirectional converter to pre-charge the energy storage capacitors in each section and injects an initial energy threshold into the DC bus, laying the energy reserve for the soft start of the first motor. However, the current surge during motor startup can still affect the stability of the bus voltage. To ensure a smooth startup process, the dynamic current limiting module introduces ramp-up current limiting logic and real-time monitoring technology to dynamically adjust the startup strategy, controlling the bus surge within a manageable range and providing stable operating conditions for the subsequent energy reinjection module's motor startup and energy reinjection.

[0078] The purpose of the dynamic current limiting module is to constrain the starting current through ramp current limiting logic when the first stage of motor startup is triggered, and to monitor the voltage drop rate and peak width and duration characteristics of the bus voltage in real time. Based on the analysis results of the preset model, it determines whether to adjust the current limiting slope and delay the next stage of startup command to ensure that the bus voltage fluctuation is within a controllable range. The following is a detailed breakdown of the technical logic:

[0079] S3.1: Trigger the first stage of the motor and use ramp current limiting logic to constrain the starting current.

[0080] The start-up of the yaw motor causes a surge in instantaneous current demand, which in turn severely impacts the DC bus voltage. To smooth this process, the system employs ramp-up current limiting logic when triggering the first stage of motor start-up, gradually increasing the current to mitigate the sudden change in energy demand. The servo driver, as the core component of current control, can precisely adjust the output to ensure the controllability of current changes.

[0081] In practice, the first stage of motor startup is triggered, and the rising rate of the starting current is controlled by the servo driver. The starting current exhibits a linear characteristic over time, meaning it gradually increases at a fixed slope, starting from the motor's no-load current. The motor's no-load current is the current value when the motor is running without load, which is read and recorded in real time by sensors in the sequence generation module. The starting current is calculated by adding the motor's no-load current to the result of multiplying time by the current-limiting slope. The current-limiting slope is a preset fixed value, measured in amperes per second, representing the rate at which the current increases over time. Its value is determined during system design based on motor characteristics and the bus voltage tolerance range. Once the current rises to the stable value required for motor operation, the servo driver adjusts its output based on the motor's feedback signal to maintain current stability.

[0082] By gradually increasing the current, the energy demand changes during motor startup were effectively mitigated, and the drastic disturbance of the bus voltage caused by instantaneous high current was avoided, creating stable initial conditions for subsequent monitoring and adjustment.

[0083] S3.2: Synchronously acquire voltage drop rate characteristics and peak width duration characteristics.

[0084] The bus voltage fluctuations and current peak characteristics caused by motor startup are key indicators for assessing the impact on the system. Real-time extraction of these features provides quantitative data support for precise decision-making and adjustments. High-frequency data acquisition technology plays a crucial role here, ensuring the real-time nature and accuracy of feature extraction.

[0085] In practice, before the first stage of motor startup, the bus voltage curve data is continuously collected at fixed time intervals. After startup, the lowest point of the bus voltage and its occurrence time, as well as the time it takes for the voltage to recover to a stable value, are identified. The lowest point of the bus voltage is the minimum voltage drop during startup, and the recovery time is the period from the lowest point to the voltage recovering to a stable value. The stable value is the value when the bus voltage tends to be stable after motor startup. The system calculates the instantaneous voltage attenuation ratio to evaluate the combined impact of the bus voltage drop rate and recovery time. The instantaneous voltage attenuation ratio is calculated as follows: subtract the lowest point voltage from the bus voltage at startup to obtain the voltage drop amplitude, and then divide the voltage drop amplitude by the recovery time to obtain the instantaneous voltage attenuation ratio, in volts per second. The bus voltage at startup is the voltage value when the initial energy threshold is injected after pre-charging in the energy pre-charge module, and it has been recorded as the initial state parameter of the system.

[0086] Simultaneously, the peak current segment during motor startup is captured, the peak current is identified, and the time span from half the peak current to the peak current and then back to half the peak current is measured; this is called the half-peak width. For ease of comparison, the system normalizes the half-peak width to the motor's rated current. The normalization is calculated by dividing the half-peak width by the ratio of the peak current to the motor's rated current, yielding the current surge peak width in seconds. The motor's rated current is an inherent parameter, already recorded in the sequence generation module by reading data from the motor nameplate or sensors. The current surge peak width reflects the duration of the current surge, facilitating system analysis of its potential impact on the bus voltage.

[0087] By extracting the instantaneous voltage attenuation ratio and the peak width of the current wave, the transient impact of motor startup on bus voltage and current was quantified, providing an accurate data foundation for subsequent model analysis and ensuring the pertinence and effectiveness of the adjustment strategy.

[0088] S3.3: Model Analysis and Decision Making.

[0089] After obtaining the instantaneous pressure attenuation ratio and wave peak width, a pre-trained random forest regression model is used to evaluate the stability of the startup process. Based on the evaluation results, the control strategy is dynamically adjusted to adapt to the operational requirements under different working conditions. The random forest regression model is suitable for analyzing complex working conditions because it can handle multidimensional features and output continuous predictions.

[0090] In practice, the instantaneous voltage attenuation ratio and the peak width of the surge wave are used as input features and fed into a random forest regression model. This pre-trained model can predict the smoothness of the startup process based on the input features and output a dimensionless startup smoothness coefficient. The magnitude of the startup smoothness coefficient reflects the impact of the startup process on the bus voltage; a lower value indicates a smaller impact, and vice versa. The system presets a safety threshold to determine whether the startup smoothness coefficient is within an acceptable range. This safety threshold is determined based on the tolerable range of the bus voltage during system design and is a dimensionless fixed value.

[0091] During startup, the system calculates the startup smoothness coefficient in real time and compares it with a safety threshold. If the startup smoothness coefficient is less than the safety threshold, it indicates that the impact of the current startup process on the bus voltage is within a controllable range. The system maintains the current current limiting slope and continues to start the first stage of the motor, recording the bus voltage curve for reference in the energy reinjection module. If the startup smoothness coefficient is greater than or equal to the safety threshold, it indicates that the current startup process has an excessive impact on the bus voltage, and the system needs to adjust the control parameters. The specific adjustment method is to divide the current limiting slope by an adjustment factor greater than 1 to obtain a new current limiting slope, thereby slowing down the current rise rate. The magnitude of the adjustment factor is set based on the empirical value determined during system design. Simultaneously, the system calculates a delay time based on the degree to which the startup smoothness coefficient exceeds the safety threshold and delays the startup command of the next stage of the motor by this time. The delay time is calculated by multiplying the difference between the startup smoothness coefficient and the safety threshold by a preset time scaling factor, in seconds. The adjusted current limiting slope will be reapplied to the starting current control of the first stage motor until the starting smoothness coefficient meets the requirements.

[0092] By conducting real-time assessments and dynamic adjustments, the bus voltage fluctuations are kept within a safe range, thereby improving the adaptability and stability of the wind turbine yaw motor control system under different operating conditions.

[0093] The dynamic current limiting module's processing begins with the pre-charged energy storage capacitor from the energy pre-charge module and the DC bus injecting initial energy. The system first triggers the start-up of the first stage of the motor, employing ramp current limiting logic to smooth the starting current and mitigate sudden changes in energy demand. Next, by real-time monitoring of the instantaneous voltage decay ratio of the bus voltage and the peak width characteristics of the starting current, the transient impact of the start-up process on the system is quantified. A random forest regression model is used to evaluate the start-up stability coefficient, and the current limiting slope and the start-up sequence of the next stage of the motor are dynamically adjusted based on the results to ensure the stability of the bus voltage. This entire process, through data-driven analysis and adaptive control, optimizes the motor start-up process, providing stable operating conditions for the energy reinjection module to trigger the next stage of the motor and perform energy reinjection, thus improving the continuous operation capability of the wind power generation system under extreme conditions.

[0094] The following is a detailed description of the pre-training and optimization process of the random forest regression model:

[0095] 1) Data preparation stage;

[0096] Data collection;

[0097] Before pre-training the random forest regression model, time-series data on the motor startup process needs to be collected from actual operation or simulation experiments of the wind power system. This data includes parameters such as starting current, DC bus voltage, and motor speed. To quantify the impact of the startup process on the DC bus voltage, two core features are extracted: instantaneous voltage decay ratio and current peak width. The instantaneous voltage decay ratio is calculated as follows: first, the DC bus voltage value at the moment of motor startup is recorded as an initial reference value; then, voltage data is continuously collected during startup, and the voltage change between adjacent time points is calculated, i.e., the voltage value at the later time point minus the voltage value at the previous time point. This change is then divided by the time interval between the two time points to obtain the instantaneous rate of voltage drop; finally, the absolute value of this instantaneous rate is divided by the voltage value at the moment of startup to obtain the instantaneous voltage decay ratio. The current peak width is calculated as follows: the time-series waveform of the starting current is analyzed to determine the time point when the current reaches its maximum value. Then, the time length from when the current value reaches its maximum value to when it drops to 90% of its maximum value is measured in seconds, which is taken as the current peak width. In addition, to assess start-up smoothness, the maximum amplitude of voltage fluctuations—that is, the absolute value of the difference between the minimum DC bus voltage and the rated voltage during startup—and the time required for the voltage to recover from the minimum to a stable state are recorded. A ratio is obtained by dividing the maximum amplitude of voltage fluctuations by the rated voltage, and this ratio is then multiplied by the voltage recovery time to calculate the start-up smoothness coefficient, which serves as the model's output target. This data acquisition method comprehensively captures the dynamic characteristics of motor startup, providing a reliable foundation for subsequent model training.

[0098] Data cleaning and preprocessing;

[0099] The collected data may contain outliers or noise, therefore cleaning is necessary. The cleaning process involves iterating through the transient voltage attenuation ratio and flow wave peak width of each data set, calculating the mean and standard deviation of these features across all samples. For any given feature value, if the absolute value of its difference from the mean is greater than three times the standard deviation, it is identified as an outlier and removed. After cleaning, the features are normalized to eliminate the influence of different dimensions. The normalization calculation process is as follows: for the transient voltage attenuation ratio, find the maximum and minimum values ​​among all samples, subtract the minimum value from the transient voltage attenuation ratio of each sample, and then divide by the difference between the maximum and minimum values ​​to obtain the normalized value; the same operation is repeated for the flow wave peak width. After cleaning and normalization, the quality and consistency of the data are improved, facilitating the model's accurate learning of the relationship between the features and the start-up smoothness coefficient.

[0100] 2) Model building stage;

[0101] Feature selection;

[0102] When constructing the random forest regression model, the instantaneous voltage attenuation ratio and current peak width are used as the main input features because these two features directly reflect the dynamic changes in voltage and current during startup. In addition, motor inertia and starting slope can be added as optional features according to actual needs. The motor inertia is calculated based on the motor mass and rotation radius, while the starting slope is obtained by dividing the initial speed change by the corresponding time interval. These features are chosen because they are closely related to the physical process of motor startup and can enhance the model's predictive ability for startup smoothness coefficients.

[0103] Model parameter settings;

[0104] The random forest regression model consists of multiple decision trees, each independently predicting the initial stability coefficient, with the final result being the average of all trees. The initial number of decision trees is set between 100 and 500, with the specific value determined through subsequent validation. The depth of each tree and the minimum number of samples per leaf node need optimization. The depth controls the complexity of the tree, while the minimum number of samples per leaf node limits the minimum number of samples contained in each leaf, avoiding overly fine segmentation. These parameter settings aim to balance the model's prediction accuracy and computational efficiency.

[0105] Training process;

[0106] Before training, the dataset is divided into a training set (80%) and a validation set (20%). The training process involves randomly sampling a subset of samples and features from the training set to construct the first decision tree. For each sample, starting from the root node, the model is branched to the left or right child node based on the feature value, until a predetermined depth is reached or the number of samples in a leaf node is insufficient. Each leaf node outputs the average startup stability coefficient of the samples within that node. This process is repeated to construct all decision trees. After training, the model is tested using the validation set. The sum of squares of the differences between the predicted and actual startup stability coefficients is calculated and divided by the number of samples in the validation set to obtain the mean squared error, which serves as the evaluation metric. This random sampling training method increases the model's diversity, enabling it to adapt to data variations under different operating conditions.

[0107] 3) Model optimization stage;

[0108] Feature importance analysis;

[0109] To improve model efficiency, it is necessary to evaluate the contribution of each feature to the prediction of the startup stability coefficient. The calculation process is as follows: for each decision tree, calculate the total reduction in prediction error for each feature across all split nodes, i.e., the sum of the absolute values ​​of the difference between the sum of the variances of the startup stability coefficients of the child nodes and the variance of the parent node after each split; sum the statistical results of all trees to obtain the contribution value of each feature; divide each feature's contribution value by the total contribution value to obtain the relative importance percentage. If the importance of a feature is below a set threshold (e.g., 5%), that feature is removed. This analysis simplifies input features, reduces redundant calculations, and improves the model's running speed and prediction accuracy.

[0110] Hyperparameter tuning;

[0111] To optimize the number and depth of decision trees, five-fold cross-validation is used. The process involves dividing the training set into five equal parts, using four parts to train the model, and leaving one part for validation. For each training set, multiple values ​​for the number of decision trees (ranging from 100 to 500) and the depth (ranging from 5 to 20) are tested, and the mean squared error of the validation set is calculated. The mean squared errors of the five validations are averaged, and the parameter combination with the smallest average error is selected. This method ensures the stability of the parameters across different data subsets through multiple validations, enabling the model to have strong generalization ability while fitting the training data.

[0112] Model integration;

[0113] To further improve prediction accuracy, a random forest regression model can be integrated with a support vector regression model. The integration process involves first training both the random forest and support vector regression models separately; then, for the validation set samples, recording the predicted values ​​of both models; calculating the average error between the two predicted values ​​and the actual values, assigning a higher weight (e.g., 60%) to the model with the smaller error and a lower weight (e.g., 40%) to the other model; finally, the predicted value is a weighted average of the two models' predictions. This integration method combines the robustness of random forests with the ability of support vector regression to capture nonlinear relationships, thus improving the reliability of the prediction results.

[0114] 4) Model validation and application stage;

[0115] Offline verification;

[0116] After optimization, the model performance was evaluated using an independent test dataset. The testing process involved inputting the instantaneous pressure attenuation ratio and wave peak width into the test set, calculating the start-up smoothness coefficient of the model output; subtracting the predicted value from the actual value, taking the square of the difference, and averaging the squared differences across all samples as the mean square error; specifically for extreme conditions such as high wind speeds or motor failures, the deviation between the predicted and actual values ​​was statistically analyzed to ensure the error remained within acceptable limits. This validation confirmed the model's adaptability and stability under diverse operating conditions.

[0117] Online applications;

[0118] The model was deployed into the wind turbine yaw motor control system and ran in real time. The application process involved the system collecting the instantaneous voltage attenuation ratio and current wave peak width during motor startup, inputting these data into the model, and outputting a startup smoothness coefficient. This coefficient was compared with a preset safety threshold. If the threshold was exceeded, adjustments were made by reducing the current limiting slope or delaying the next stage of motor startup until the DC bus voltage stabilized. This real-time application ensured that the impact of the motor startup process on the power supply system was minimized, maintaining reliable system operation.

[0119] Through the above pre-training and optimization process, the random forest regression model can accurately predict the starting stability coefficient based on features such as instantaneous voltage attenuation ratio and wave peak width, providing intelligent support for the wind power yaw motor control system and ensuring the stability of DC bus voltage under complex operating conditions.

[0120] In the dynamic current limiting module, ramp-up current limiting logic and real-time monitoring technology are used to adaptively regulate the first stage of motor startup, ensuring that bus voltage surges are controllable. These steps lay the foundation for energy reserves and voltage stability for multi-stage motor startup. However, during the sequential startup of multiple motors, the timing of bus voltage recovery and the effective utilization of regenerated energy become crucial. The energy reinjection module needs to further optimize the startup sequence and energy distribution within this context to reduce subsequent voltage fluctuations and improve system stability.

[0121] The goal of the energy reinjection module is to trigger the next stage of motor startup when the bus voltage recovers to the upper edge of the preset recovery threshold, and to inject the regenerative energy generated during the previous motor acceleration phase into the energy storage capacitor to reduce the second voltage fluctuation. The following is a detailed breakdown of the technical logic:

[0122] S4.1: Monitor bus voltage and trigger the next motor segment.

[0123] After the first stage of motor starts, the DC bus voltage will drop due to the starting current demand, and then gradually recover. If the next stage of motor is triggered before the voltage has fully recovered, the voltage may drop further, affecting system stability. Therefore, it is necessary to accurately determine the timing of voltage recovery to ensure the orderly and continuous start-up process. Real-time voltage monitoring technology plays a crucial role here, providing high-precision dynamic data.

[0124] In practice, the system continuously collects DC bus voltage curve data and records its changes over time. The system presets a bus recovery threshold in volts, slightly higher than the bus voltage during normal motor operation, to ensure the DC bus has sufficient energy reserves to support the next motor start-up. During monitoring, the system compares the current bus voltage with the recovery threshold in real time. When the current bus voltage reaches or exceeds the recovery threshold, the system issues a start command for the second motor. After the start command is issued, the system uses the ramp current limiting logic applied in the dynamic current limiting module to control the rise rate of the second motor's starting current via a servo driver, maintaining smooth current changes, and continues to monitor the dynamic changes in the bus voltage to ensure the controllability of the subsequent start-up process.

[0125] By triggering the next motor start when the bus voltage recovers to a safe level, the system achieves orderly connection of segmented starting, avoids transient impact on the DC bus caused by the simultaneous starting of multiple motors, ensures voltage stability, and provides a stable operating environment for subsequent motor starting and energy management.

[0126] S4.2: Capture and inject regenerative energy.

[0127] After the first stage motor accelerates to the target speed, it enters regenerative braking mode, generating regenerative energy. If not utilized, this energy can cause abnormal fluctuations in the bus voltage and even trigger protection mechanisms. By injecting the regenerative energy into the energy storage capacitor of the next stage motor, local energy support can be provided for subsequent startups, while simultaneously optimizing the overall energy utilization efficiency of the system. The isolated bidirectional converter, as the core device for energy transfer, can efficiently and controllably complete energy transfer.

[0128] In practice, once the first motor accelerates to the target speed, the servo driver detects that the speed has reached the preset value and automatically switches to regenerative braking mode to generate regenerative energy. The total amount of regenerative energy depends on the motor's operating state and braking duration. The system transfers the regenerative energy from the DC bus to the energy storage capacitor of the second motor via an isolated bidirectional converter. The capacity of the energy storage capacitor is a known fixed value, measured in farads. The energy injection process requires control of the injection power to avoid secondary disturbances to the bus voltage. The injection power is calculated by dividing the regenerative energy by the injection time and then multiplying it by an energy transfer efficiency coefficient. The energy transfer efficiency coefficient is a dimensionless value, ranging from 0 to 1, reflecting the efficiency loss of the converter during energy transfer, and is usually determined by equipment parameters. The injection time is a time span preset by the system based on the recovery speed of the bus voltage and the expected timing of the second motor's start-up. After injection, the voltage of the energy storage capacitor in the second motor rises, increasing the stored energy and providing additional energy reserves for the second motor's start-up.

[0129] By recovering the regenerative energy of the first stage motor and injecting it into the energy storage capacitor of the second stage motor, the system not only achieves efficient energy utilization, but also reduces direct dependence on the DC bus by dispersing energy sources, providing stable energy support for the start-up of the next stage motor and improving the system's operating efficiency.

[0130] S4.3: Reduce the second voltage fluctuation.

[0131] When the second-stage motor starts, its energy demand may again cause a drop in the DC bus voltage, affecting system stability. To mitigate this issue, the system prioritizes utilizing the energy stored in the second-stage motor's energy storage capacitor to meet the starting requirements, thereby relieving the power supply pressure on the DC bus. The rapid response characteristics of the energy storage capacitor enable it to provide energy promptly, reducing the fluctuation range of the bus voltage.

[0132] In practice, before starting the second-stage motor, the system calculates the extractable energy from the energy storage capacitor. The extractable energy is calculated as follows: multiply the capacitor's capacity by the square of the current voltage, subtract the square of the minimum allowable voltage, and take half of the result. The unit is joules. The current voltage is the capacitor's voltage after regenerative energy injection, in volts, determined by the pre-charge voltage of the energy pre-charge module and the regenerative energy injection in this step. The minimum allowable voltage is the minimum operating voltage determined during the capacitor's design, in volts, ensuring the capacitor can still operate normally after energy is provided. The starting energy requirement of the second-stage motor has been calculated in the energy pre-charge module: multiply the motor's total inertia by the square of the target speed, and take half of the result, in joules. The system compares the extractable energy with the second-stage motor's starting energy requirement. If the extractable energy is greater than or equal to the starting energy requirement, the starting energy of the second-stage motor is entirely provided by the energy storage capacitor; if the extractable energy is less than the starting energy requirement, the shortfall is supplemented by the DC bus. The extraction of supplemental energy is controlled by a servo driver to ensure that the current rise conforms to the ramp current limiting logic. During startup, the system prioritizes extracting energy from the energy storage capacitor, reducing the energy demand on the DC bus and thus reducing the fluctuation range of the bus voltage.

[0133] By prioritizing the use of the energy storage capacitor, the system effectively alleviates the power supply pressure on the DC bus, significantly reduces voltage fluctuations caused by the start-up of the second-stage motor, and enhances the smoothness of the multi-stage motor start-up process and the overall reliability of the system.

[0134] The energy reinjection module's processing begins with the dynamic current limiting module's monitoring of the first-stage motor startup and bus voltage curve. By comparing the bus voltage with a preset recovery threshold in real time, it precisely determines the startup timing of the second-stage motor, ensuring the orderly startup of each stage. Subsequently, the system captures the regenerative energy generated by the acceleration of the first-stage motor and injects it into the energy storage capacitor of the second-stage motor through an isolated bidirectional converter, providing local energy support for startup. During the startup of the second-stage motor, the system prioritizes the use of the energy storage capacitor, reducing energy extraction from the DC bus and mitigating voltage fluctuations. The entire process, through precise voltage monitoring, energy recovery, and distribution, optimizes the connection and energy management of multi-stage motor startup, providing stable operating conditions for the braking energy recovery of the cyclic replenishment module and auxiliary load power supply, ensuring the continuous operation and high efficiency of the wind power generation system under sudden crosswind scenarios.

[0135] From the sequence generation module to the energy reinjection module, real-time wind direction detection, segmented start-up sequence generation, pre-charging of the energy storage capacitor, adaptive control of the starting current, and triggering of the next motor segment have been implemented, ensuring the smoothness of multi-segment motor start-up and the stability of the bus voltage. However, after yaw adjustment, the motor enters the braking phase, generating a large amount of regenerative energy. If not properly handled, this may lead to energy waste or abnormal fluctuations in the bus voltage. The cyclic replenishment module focuses on the takeover and distribution of braking regenerative energy and the resetting of the energy storage capacitor, providing energy reserves for the next round of yaw adjustment, realizing closed-loop energy utilization and continuous stable system operation.

[0136] S5.1: Regenerative braking energy.

[0137] After the yaw motor completes yaw adjustment, it needs to stop rotating. If the regenerative braking energy generated during this process is not properly handled, it may cause the DC bus voltage to rise, affecting system stability. Therefore, the system uses a bidirectional converter as an energy regulation device, taking advantage of its ability to flexibly control the energy flow direction and transmission rate to capture regenerative braking energy and maintain the stability of the DC bus voltage.

[0138] In practice, the system continuously monitors the speed of the yaw motor through a servo driver. When the speed drops to zero, it confirms that the yaw motor has entered the braking phase, and the bidirectional converter switches to regenerative mode to begin capturing regenerative braking energy.

[0139] Regenerative braking energy originates from the conversion of kinetic energy during motor deceleration. Its magnitude is determined by the motor's inertia and deceleration rate, and is measured in joules. The captured regenerative braking energy is transmitted to the DC bus via a bidirectional converter, and then distributed according to the power demands of the auxiliary loads within the cabin and the energy status of the energy storage capacitors. To prevent excessive regenerative braking energy from causing the DC bus voltage to exceed the safe range, a pre-set upper limit for the DC bus voltage is implemented. If the real-time monitored DC bus voltage reaches or exceeds this upper limit, the bidirectional converter suspends the feedback of regenerative braking energy to the DC bus until the voltage falls back within the safe range. Throughout this process, the bidirectional converter dynamically adjusts the energy transmission rate based on the real-time value of the DC bus voltage to ensure that the voltage remains within the predetermined safe operating range.

[0140] In this way, the regenerative braking energy is safely captured and converted into usable electrical resources, the stability of the DC bus voltage is maintained, providing a reliable energy foundation for subsequent energy distribution, while avoiding system risks caused by abnormal voltage fluctuations.

[0141] S5.2: Supply auxiliary loads within the cabin according to priority.

[0142] The wind power system nacelle contains various auxiliary loads, such as lighting and communication equipment, each with different power requirements and importance. To improve energy efficiency and reduce dependence on external power sources, the system utilizes captured regenerative braking energy and allocates it according to a preset priority sequence, ensuring that critical loads receive priority power support.

[0143] In practice, the system pre-establishes a priority sequence for auxiliary loads, with loads arranged from highest to lowest importance, and the highest priority load at the top of the sequence. The system monitors the power demand of each auxiliary load in real time and adds up the power demands of all auxiliary loads to calculate the total power demand. The bidirectional converter feeds back regenerative energy to the DC bus at a certain power; this power is called regenerative energy power. The allocation process starts with the first load in the priority sequence. If the regenerative energy power is greater than or equal to the power demand of the first load, the power demand of the first load is fully met, and the remaining value after subtracting the power demand of the first load from the regenerative energy power is used for the allocation of the next priority load. If the regenerative energy power is less than the power demand of the first load, all regenerative energy power is supplied to the first load, and the allocation of subsequent loads stops. This process continues until the regenerative energy power is allocated or the demands of all auxiliary loads are met. During the allocation process, the auxiliary loads extract the required energy from the DC bus through the bidirectional converter, while the system continuously monitors the DC bus voltage to ensure it remains within a safe range, preventing the voltage from exceeding the predetermined range due to energy allocation.

[0144] By allocating regenerative braking energy according to priority sequence, the system achieves optimized utilization of power resources, ensures stable operation of high-priority loads, maintains the stability of DC bus voltage, reduces power input from external power sources, and enhances the system's energy self-sufficiency.

[0145] S5.3: Reset the energy storage capacitor threshold.

[0146] After the regenerative braking energy distribution is completed, the system needs to adjust the energy level of the energy storage capacitor to a preset target value to ensure sufficient energy support for the next yaw adjustment. A bidirectional converter is used to dynamically control the energy flow direction, and the energy level is precisely reset by supplementing energy from the DC bus to the energy storage capacitor.

[0147] In practice, the system first assesses the current energy level of each energy storage capacitor segment. The energy level is calculated by measuring its voltage: multiplying the capacitor's capacity by the square of the current voltage, then dividing the result by two, yielding the current energy level in joules. The capacitor's capacity is a known fixed value in farads, while the voltage is monitored in real-time by sensors in volts. Each energy storage capacitor segment has a preset target energy threshold in joules, determined based on the pre-charge energy required for initial startup in the energy pre-charge module. If the current energy level is lower than the target threshold, the system replenishes energy to the energy storage capacitor from the DC bus via a bidirectional converter. The replenishment energy is calculated by subtracting the current energy level from the target threshold, yielding the required replenishment energy in joules. The bidirectional converter transmits energy to the energy storage capacitor at a certain power; this power is called the replenishment power in watts. During replenishment, the system monitors the DC bus voltage in real-time to ensure it does not fall below a preset safety lower limit in volts. If the DC bus voltage approaches the lower limit, the supplementary power should be appropriately reduced to maintain overall system stability. After supplementation, the energy level of each energy storage capacitor will reach the target energy threshold.

[0148] Through dynamic energy replenishment, the energy level of the energy storage capacitor is precisely adjusted to the preset target value, providing sufficient energy reserves for the next yaw adjustment, while ensuring the stability of the DC bus voltage and enhancing the system's energy management capability during continuous operation.

[0149] The regenerative energy replenishment module's process begins with the stabilization of the DC bus voltage and the energy status of the energy storage capacitor after the multi-stage motors in the energy reinjection module start up. It captures the regenerative energy generated by the yaw motor during braking through a bidirectional converter and distributes it to auxiliary loads within the nacelle according to priority, achieving optimized utilization of power resources. Subsequently, the system assesses the current energy level of the energy storage capacitor and replenishes energy from the DC bus to the target energy threshold through the bidirectional converter, ensuring sufficient energy reserves. This entire process, through precise energy capture, distribution, and replenishment, achieves closed-loop energy management for the yaw motor control, enhancing the wind power generation system's adaptability and stability under complex operating conditions.

[0150] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0151] It should be noted that the system of the present invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting a variety of hardware environments and usage requirements.

[0152] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0153] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely to distinguish one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0154] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A power supply system for controlling a wind turbine yaw motor, characterized in that, include: Sequence generation module: Real-time monitoring of wind direction changes, acquisition of motor no-load current and inertia, and generation of segmented start-up sequences in descending order of inertia; Energy pre-charge module: Utilizes an isolated bidirectional converter to pre-charge the energy storage capacitors of each section to the target energy level according to the startup sequence, and injects the initial energy threshold into the DC bus to ensure energy preparation before the motor soft start; Dynamic current limiting module: Triggers the first stage of the motor and uses ramp current limiting logic to constrain the starting current. During startup, it synchronously acquires the voltage drop rate characteristics and peak width duration characteristics, and uses the results of preset model analysis to dynamically adjust the current limiting slope and the startup sequence of the next stage. Energy recovery module: When the bus voltage recovers to the upper edge of the recovery threshold, the next stage of the motor is triggered, and the regenerative energy generated in the previous acceleration stage is injected into the energy storage capacitor through the bidirectional converter to reduce the second voltage fluctuation; Cyclic Replenishment Module: After yaw adjustment, the isolated bidirectional converter takes over the brake regenerative energy and supplies it to the cabin auxiliary loads according to priority. Then, the energy storage capacitor is reset to the target energy threshold to retain a stable energy reserve for the next round of startup. The sequence generation module includes the following: monitoring wind direction and nacelle orientation, calculating the deviation between the two and comparing it with a preset threshold to determine whether yaw adjustment is needed, obtaining the no-load current and inertia parameters of each yaw motor when the necessity of adjustment is confirmed, arranging all yaw motors in descending order of inertia value and dividing them into multiple segments to balance the total inertia of each segment, generating a startup sequence to activate each segment in sequence, starting from the segment containing the motor with the highest inertia value; The energy pre-charge module includes the following components: calculating the starting energy requirement of each motor segment based on the motor inertia and target speed; adjusting the starting energy requirement through the energy redundancy coefficient to set the pre-charge target energy of each energy storage capacitor segment; calculating the pre-charge target voltage of each energy storage capacitor segment based on the energy storage capacitor capacity and the pre-charge target energy; pre-charging each energy storage capacitor segment according to the starting sequence and monitoring the energy storage capacitor voltage in real time until the pre-charge target voltage is reached; calculating the initial energy threshold required by the DC bus based on the starting energy requirement of the first motor segment and the amplification factor; injecting energy into the DC bus through an isolated bidirectional converter and adjusting the DC bus voltage to support motor starting. The dynamic current limiting module includes the following: It evaluates the motor's starting smoothness by real-time monitoring of the instantaneous voltage attenuation ratio of the bus voltage and the peak width of the starting current. The instantaneous voltage attenuation ratio is obtained by acquiring the bus voltage curve and calculating the difference between the lowest point and the starting point, divided by the rise time. The peak width of the starting current is obtained by truncating the peak segment and measuring the half-peak width, then normalizing it to the motor's rated current. The instantaneous voltage attenuation ratio and the peak width of the starting current are input into a pre-trained random forest regression model to calculate the starting smoothness coefficient. Based on the comparison between the start-up stability coefficient and the preset safety threshold, the current limiting slope and the start-up sequence of the next motor are dynamically adjusted to ensure that the bus voltage fluctuation is controlled within the predetermined range.

2. The power supply system for wind turbine yaw motor control according to claim 1, characterized in that, The energy redundancy factor is a value greater than 1, used to calculate the pre-charge target energy of each segment of the energy storage capacitor by multiplying the starting energy requirement of each segment of the motor by the energy redundancy factor.

3. A power supply system for controlling a wind turbine yaw motor according to claim 2, characterized in that, The energy reinjection module includes the following components: The DC bus voltage is monitored, and the next motor is triggered to start when the DC bus voltage recovers to the upper edge of the preset recovery threshold. At the same time, the regenerative energy generated by the acceleration of the previous motor is injected into the energy storage capacitor through the isolated bidirectional converter to support the start of the next motor and reduce the fluctuation of DC bus voltage.

4. A power supply system for controlling a wind turbine yaw motor according to claim 3, characterized in that, The energy reinjection module also includes the following: The injected power of regenerative energy is determined by the energy transmission efficiency and the injection time, and the starting energy provided by the energy storage capacitor takes priority over the DC bus in order to reduce the extraction of energy from the DC bus.

5. A power supply system for controlling a wind turbine yaw motor according to claim 4, characterized in that, The cyclic supply module includes the following: After the yaw adjustment is completed, the regenerative energy generated by the yaw motor during the braking phase is captured by the bidirectional converter and transmitted to the DC bus through the bidirectional converter. At the same time, the DC bus voltage is monitored to prevent the voltage from exceeding the preset upper limit. Then, according to the preset auxiliary load priority sequence and the real-time monitored auxiliary load power demand, the regenerative energy is distributed to the auxiliary loads in the cabin in priority order to ensure that high priority loads receive power support first.

6. A power supply system for controlling a wind turbine yaw motor according to claim 5, characterized in that, The cyclic supply module also includes the following: After allocation, the current energy level of each energy storage capacitor is assessed, and energy is replenished from the DC bus to the preset target energy threshold through a bidirectional converter to ensure that the energy storage capacitor provides sufficient energy reserves for the next yaw adjustment.

Images