Research on method of identifying cable-releasing logic defects in wind turbines

The yaw system of a wind turbine is one of the important components that ensures the normal operation of the turbine and has a significant impact on the power generation of the turbine. Currently, there is little research on the cable-releasing logic in the yaw system. This paper analyzes the possible logical defects in the cable-releasing system and examines the cable-releasing logic of the turbine by using a method based on its operational logic. By subdividing the turbine’s state using information such as power, wind speed, and yaw angle from operational data, abnormal cable-releasing events can be identified through logical filtering to detect whether there are logical defects in the turbine’s cable-releasing logic. The reliability of the method was verified by using examples from turbines that did have cable-releasing logic defects.


Introduction
As the world's demand for energy continues to rise, the scarcity of non-renewable resources such as crude oil and the challenging issue of atmospheric pollution has gradually led people to increase the development and utilization of green and environmentally friendly energy sources like wind and solar power [1] .The advantage of wind energy lies in its widespread distribution and relatively low generation costs.With the continuous advancement of wind power generation technology, the individual capacity of wind turbines has now reached the megawatt level [2] .Wind direction is inherently stochastic and constantly changing, and the key to the efficient and stable operation of the entire wind turbine lies in control technology, with active yaw control systems being an essential component of horizontal-axis wind turbine control systems [4] .
Currently, there has been significant progress in the design of control algorithms for yaw control systems.Mainstream control algorithms include yaw algorithms based on fuzzy controllers [4] , yaw control algorithms based on fixed-time intervals [5] , yaw control algorithms based on PID controllers [6] , and most recently, yaw control systems based on neural networks [11] .These methods, through different designs, reduce the frequent yawing actions caused by random changes in wind direction, demonstrating good static stability and higher reliability.Although the algorithm logic for yaw control systems is continually updated and optimized, the objective presence of factors such as installation errors and accumulated mechanical errors during operation leads to certain system errors in yaw control.The identification method for this error primarily relies on comprehensive data collection and data obtained from the Supervisory Control and Data Acquisition (SCADA) system, which includes yaw and wind direction-related data points.This system error is identified and calculated through data statistics and analysis methods and subsequently corrected [7] .Compared to research on yaw control algorithms and their system errors in wind turbines, research on cable release control algorithms is relatively limited at present.Tan proposed a cable release control algorithm based on cable twist angle and turns [3] .The principle of this algorithm is relatively simple, and the cable release control logic used in mainstream wind turbines on the market is mostly similar to this method.The main difference lies in the cable twist angle threshold that triggers the cable release control logic.However, there has not been a comprehensive analysis of the shortcomings in cable release control logic for wind turbines at present.

Principle of wind turbine yaw system
Figure 1.The flow chart of a typical wind turbine yaw control.
The yaw system is one of the essential subsystems of a wind turbine, which typically adopts the actively yawed form with gear-driven [12] .The principle diagram is shown in Figure 1, with the following three main functions [13] : 1) To control the alignment of the wind turbine's rotor axis (nacelle axis) with the wind direction [8] .According to aerodynamic theory, the expression for the power that a wind turbine can extract from the incoming wind is P = ଵ ଶ ɏC ୮ Sv ଷ cos ଷ Ʌ, where P is the power absorbed by the wind turbine, ɏ is the air density, C ୮ is the power coefficient of the wind turbine, S is the swept area of the rotor, v is the incoming wind speed, and Ʌ is the yaw error angle [5] .From the above formula, it is evident that the power absorbed by the wind turbine is directly proportional to the cube of the cosine of the yaw error angle.The closer the yaw error angle is to 0, the greater the power that the wind turbine can absorb and generate.Therefore, one of the primary functions of the yaw system is to control the orientation of the rotor axis so that the angle between it and the wind direction remains near 0, maximizing the wind turbine's power output.
2) To control the wind turbine's cable release.If the wind turbine continuously yaws in one direction, the cable's twist angle will continuously increase.When the winding angle becomes too large, the cable may break, potentially leading to safety accidents.To prevent such accidents, the wind turbine yaw system needs to monitor the cable's twist angle continuously.When the twist angle reaches a certain threshold, it triggers the automatic cable release action to keep the cable's twist angle within a safe range [9] .
3) To reduce safety risks in extreme wind conditions [14] .When the incoming wind speed exceeds the normal operating range of the wind turbine, the yaw system receives protective commands from the main control system.It promptly adjusts the yaw error angle to ±90 degrees to reduce the power absorbed by the wind turbine, thus protecting the turbine from extreme wind conditions [10] .

Possible defections of cable-releasing logic
According to the design of the controlling system by various turbine manufacturers, there may be two types of detection in the cable-releasing control logic: 1) During the startup process, the yaw movement does not consider the triggering conditions of cable-releasing logic.After the wind turbine has been shut down for a period and the wind conditions are suitable for restarting into power generation, the wind direction may have changed significantly.To ensure the safe transition of the wind turbine into the power generation state, yaw error correction IOP Publishing doi:10.1088/1742-6596/2771/1/0120253 is required at this point.The turbine's controlling system will automatically perform yaw movements based on the turbine's orientation and incoming wind direction to align the turbine with the wind direction.In tightly integrated control systems, the yawing during startup generally involves the following key steps: determining clockwise/anticlockwise yaw movements based on turbine orientation and incoming wind direction, assessing whether the yaw movement in that direction will trigger cable-releasing based on the turbine's twist angle and cable release threshold, and if triggering is likely, performing a reverse yaw movement.For the yaw movements during the operation of the wind turbine, each key step in the above process contains comprehensive logical judgments.Considering different wind conditions and the turbine's state, it aims to minimize the duration of the yaw movements while ensuring the safety of turbine operation, thereby further enhancing the turbine's power generation.However, during the startup process of the wind turbine, some turbine manufacturers, for various reasons, have simplified the above process, which defaults to yawing in the direction corresponding to the shortest yaw path.This strategy results in an initial ineffective yaw movement during the turbine startup yawing process.Subsequently, the cable release threshold is triggered, and the cable release is initiated under conditions where the wind is suitable for the turbine to be connected to the grid.Finally, the startup preparation is completed through a secondary yawing process reaching for wind alignment, allowing the wind turbine to enter the grid-connected power generation state.This process is illustrated in Figure 3. 2) During standby, the cable release threshold is reached but not activated.In the wind turbine's main control system, typically two cable release triggering thresholds are set: one is the dynamic cable release threshold, which is used to control the cable release logic triggering of the wind turbine during normal grid-connected operation.When the wind turbine reaches this threshold during normal operation, it should immediately shut down and initiate cable-releasing until the twist angle returns to a lower range (usually within ±360°); the other is the static cable release threshold, which is used to control the cable release logic triggering of the wind turbine during standby.A cable release is initiated in advance when the wind speed is low and does not meet the power generation conditions, reducing the loss of power generation caused by the frequent triggering of the dynamic cable release threshold due to continuous unidirectional yaw movements.
In certain situations, when the ambient wind speed drops below the wind turbine's cut-in speed and the turbine transitions from grid-connected operation to standby mode, the twist angle of the wind turbine may be higher than the static cable release threshold, the wind turbine should trigger the static cable release logic to initiate cable-releasing in advance, aiming to increase the wind turbine's power generation.Depending on the design by different wind turbine manufacturers, the waiting time to trigger the static cable release logic in low wind conditions varies.Some turbine manufacturers set a relatively high waiting time value for triggering the static cable release logic in low wind conditions, which can reduce the self-energy consumption in wind turbines during low wind conditions to some extent.However, this setting may also lead to situations where the wind turbine triggers the static cable release logic during the startup process when the wind conditions are suitable for grid-connected operation, as shown in Figure 4.

Method for identifying defections in wind turbine cable release logic
Based on the analysis of the two types of flaws in wind turbine cable release logic mentioned above, this paper proposes a wind turbine cable release logic flaw identification method grounded in the operational principles of wind turbines.The flowchart is illustrated in Figure 5.The wind turbine cable release logic flaw identification method proposed in this paper mainly consists of the following key steps: 1) Data acquisition, organization, and cleaning; feature precomputation: the data required for this method includes wind turbine SCADA operational data and certain fundamental information about the wind turbine.In the wind turbine SCADA operational data, essential measurements such as wind speed ‫,ݒ‬ active power ܲ ௧ , cable twist angle ߮ , geographical wind direction ߮ ௪ௗ , and turbine orientation ߮ ௧௨ must be included.The fundamental information to be obtained about the wind turbine includes the cut-in wind speed ‫ݒ‬ , rated yaw rate ο߮, and the wind speed-power curve of the turbine.The time interval for wind turbine SCADA operational data is typically ten minutes, and there might be some missing or anomalous values.Therefore, after the acquisition, data cleaning operations such as the imputation of missing values and the replacement of anomalous values are necessary.
This paper defines two key intermediate variables: the changing rate of twist and twist direction.The changing rate of twist reflects the change in the twist angle of the wind turbine within the next ten minutes.Its calculation formula at moment ‫ݐ‬ is: IOP Publishing doi:10.1088/1742-6596/2771/1/012025 From the formula, it can be seen that the changing rate of twist does not contain directional information about the yaw.Therefore, the variable of twist direction ‫ݎ݅݀‬ ‫)ݐ(‬ is introduced, and its calculation formula is: When the twist direction is 1, it indicates that the wind turbine's yaw movement is approaching the cable release threshold at that moment.When the twist direction is 0, the wind turbine is not yawing. 2) Traversal of SCADA data for abnormal cable release identification: by traversing the preprocessed SCADA data at each timestamp, an approach based on the operational principles of wind turbines is utilized to identify timestamps with abnormal cable releases.This specifically involves recognizing various states as follows: Startup State Identification: If the wind speed is greater than the cut-in wind speed and the mean active power in the past ten minutes is less than 50 kW, it is considered that the wind turbine is in a startup state; otherwise, traverse to the next timestamp.This state aims to avoid frequent start-stop cycles of the wind turbine due to wind speed fluctuation near the cut-in wind speed.
Threshold State Identification: If the wind turbine's twist angle exceeds 450° (1.25 rotations), it is considered that the wind turbine may reach the static twist threshold during the startup state.Depending on the settings of different wind turbine manufacturers, the static cable release threshold typically ranges from 540° to 650° (1.5 to 1.8 rotations).
Cable Release State Identification: If the twist rate exceeds the wind turbine's rated yaw rate multiplied by 5 minutes, and the twist direction is greater than 0 (yawing towards a twist angle of 0°), it is considered that the wind turbine is in a cable release state.
Redundant Recognition Exclusion: If the current timestamp is identified as the moment of an abnormal cable release, the time interval between the current timestamp and the last correctly identified occurrence of abnormal cable release by the wind turbine is compared.If the interval is greater than 30 minutes, the current timestamp is considered the latest occurrence of abnormal cable release.Otherwise, traverse to the next timestamp.By employing a 30-minute time window to monitor the last occurrence of abnormal cable release for redundancy, the frequency of redundant recognition can be effectively reduced.
3) Cable release abnormal record output: after identifying all timestamps where abnormal cable releases occurred, we calculate the power generation loss caused by each instance of abnormal cable release and output the records of abnormal cable releases along with the corresponding power generation losses.This paper provides an estimation method for the loss of generated power, and the steps are as follows: for each identified timestamp of abnormal cable release, we consider the 10 minutes before and 20 minutes after it as the period of abnormal cable release activity; based on the wind speed from the SCADA data and the wind speed-power curve of the turbine, we calculate the wind turbine's power generation capacity during this period ܲ ௦௧ = ‫;)ݒ(݂‬ and we calculate the power generation loss of the wind turbine during this period ‫ܮ‬ = σ (ܲ ௦௧ െ ܲ ௧ ) ‫כ‬ ο‫ݐ‬ ௧ .

Case study
The case study in this paper is based on the SCADA operational data of 40 units of 2.0 MW wind turbines, spanning from January 2021 to October 2021.The data was collected at 10-minute intervals, approximately 1.75 million SCADA data records.By utilizing the abnormal cable release identification method described earlier in this paper, a total of 289 abnormal cable release records were identified.Subsequently, visual validation was conducted based on 1-minute interval SCADA data for one hour before and after each abnormal cable release occurrence.Out of the 289 abnormal cable release records, 276 were verified to be correct, yielding an accuracy rate of 95.5%.The total IOP Publishing doi:10.1088/1742-6596/2771/1/0120257 loss of electricity was approximately 128, 000 kWh, equivalent to 1.5 hours of generation.The economic loss was estimated to be around 76, 800 RMB.

Conclusion
This paper analyzes potential flaws in the cable release control logic of wind turbines and proposes an identification method based on the operational principles of wind turbines to address these flaws.Through the analysis of operational data from 40 wind turbines over 10 months, we effectively identified cable release logic flaws in the turbines, offering guiding insights for subsequent technical improvement efforts in the wind farm.Based on the identification method and its effectiveness presented in this paper, we demonstrate the research value of identifying cable release logic flaws in wind turbines, aiming to enhance the awareness of issues caused by the cable release logic in wind turbine operations and maintenance within the wind energy production sector.

Figure 2 .
Figure 2. Control flow chart of yawing during startup.

Figure 3 .
Figure 3. Cable-releasing moves in the wrong direction.

Figure 4 .
Figure 4.The cable release threshold is reached but not activated.