Method for predicting power and adjusting rotating speed of tidal current energy water turbine

Abstract

The invention discloses a tidal current energy water turbine power prediction and rotation speed adjustment method, and relates to the technical field of offshore tidal current energy power generation, and the method comprises the following steps: a tidal current data collection step: obtaining parameters such as tidal current speed and flow direction, and storing historical data; a power prediction step: constructing a model based on the improved LSTM neural network, and outputting a power prediction value in the future 5-30 min; a rotating speed adjustment decision step: receiving a predicted value and a real-time parameter, and formulating an adjustment instruction; executing a driving step, executing an instruction, and realizing rotating speed control through a hydraulic actuator and the like to form a closed loop; and a monitoring and early warning step: monitoring the state of each step, giving an alarm and generating a report when the state is abnormal. The method improves the power prediction precision and the rotating speed adjustment response speed, reduces the equipment loss, enhances the collaboration with the power grid, guarantees the stable operation of the system, and improves the tidal current energy utilization efficiency.

Description

Technical Field

[0001] This invention relates to the field of nearshore tidal power generation technology, and in particular to a method for predicting the power and regulating the speed of a tidal turbine. Background Technology

[0002] With the global trend of energy structure transitioning towards clean and renewable energy, tidal energy, as a marine renewable energy source with abundant reserves and high predictability, has become a research hotspot in the field of new energy. The core equipment of a tidal power generation system is the tidal turbine, whose operating efficiency and power output stability directly determine the economic benefits of the entire power generation system. However, the operating environment of tidal turbines is complex. Nearshore tidal current speeds are affected by various factors such as astronomy, meteorology, and topography, exhibiting frequent fluctuations. This leads to unstable power output for the turbines—when the tidal current speed increases sharply, the turbine power easily exceeds the rated range, causing grid frequency fluctuations; when the tidal current speed decreases sharply, the power output is insufficient, making it difficult to meet the grid load demand.

[0003] Traditional tidal current turbines often employ simple speed control strategies, such as open-loop control based on real-time tidal current velocity or delayed closed-loop feedback control, lacking the ability to predict future tidal current changes. These control methods exhibit significant lag; when tidal current velocity changes rapidly, speed regulation often lags behind actual demand. This not only fails to effectively stabilize power output but may also cause excessive impact loads on the turbine's transmission system and critical blade components due to frequent speed fluctuations, accelerating equipment wear and shortening its service life. Furthermore, traditional systems do not consider error factors in the tidal current data acquisition process, such as spatial differences between sensor measurement points and the actual operating position of the turbine, and data transmission delays, resulting in insufficient accuracy of basic data and further affecting the accuracy of power prediction and speed regulation.

[0004] Furthermore, under extreme weather conditions such as typhoons and cold waves, tidal current parameters can experience abrupt changes. Traditional power prediction models, not optimized for extreme conditions, are prone to prediction failures, failing to provide reliable data for speed regulation and potentially leading to turbine malfunctions due to overload or underload operation. Simultaneously, existing systems lack a coordinated matching mechanism with grid load demand, resulting in a disconnect between turbine power output and grid load requirements. This not only wastes energy but may also impact grid stability. These issues collectively restrict the large-scale application and economic benefits of tidal current power generation systems, necessitating a technical solution with accurate power prediction, rapid speed regulation, and multi-scenario adaptability to overcome these technical bottlenecks. Summary of the Invention

[0005] The present invention proposes a highly efficient encrypted tablet data protection system and implementation method to solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for predicting the power and regulating the speed of a tidal current turbine, comprising the following steps:

[0007] The tidal flow data acquisition step is used to obtain tidal flow parameters of the turbine operating environment in real time, including tidal flow velocity, direction, water depth and seawater density. This step integrates a Doppler current meter, a three-dimensional compass, a pressure level gauge and a density sensor. The data is sent to the data processing unit via 4G / 5G wireless transmission and the historical tidal flow data of the past 30 days is stored for model training.

[0008] The power prediction step constructs a prediction model based on an improved long short-term memory neural network, which includes an input layer, a hidden layer, and an output layer. An adaptive learning rate is used during model training. The speed regulation decision step, as the core control unit, receives the predicted power value from the power prediction step and formulates a speed regulation strategy based on the current operating parameters of the turbine. This step has a built-in safety threshold, and generates a speed regulation command when the predicted power exceeds the threshold or the real-time speed deviates from the rated value.

[0009] The execution drive step is used to execute the instructions for the speed regulation decision step, including a hydraulic actuator, a gearbox and a speed feedback sensor; the hydraulic actuator changes the water flow channel area by adjusting the opening of the turbine guide vanes, the gearbox adjusts the transmission ratio, and the speed feedback sensor collects the adjusted speed data in real time to form a closed-loop control;

[0010] The monitoring and early warning system monitors the operational status of each step in real time, including sensor failures in the power flow data acquisition step, prediction deviations in the power prediction step, command execution in the speed regulation decision step, and equipment temperature in the drive execution step. When an anomaly is detected, an audible and visual alarm is triggered, a fault report is generated, and the report is sent to the remote operation and maintenance platform.

[0011] Furthermore, it includes a tidal current velocity spatiotemporal correction step. This step optimizes the real-time tidal current velocity acquired in the tidal current data acquisition step, eliminating the impact of spatial location differences and time lag on data accuracy. The step establishes a tidal current velocity correction model, combining the horizontal distance and vertical height between the turbine installation location and the current meter measurement point, as well as the tidal current propagation speed, to calculate the corrected real-time tidal current velocity in the following manner: Among them, v c orr is the corrected real-time power flow velocity, v m EAS is the raw tidal current velocity measured by the current meter, θ is the angle between the tidal current direction and the horizontal distance x, L is the rotation radius of the turbine blades, k is the spatial correction factor, Δt is the data transmission delay time, and v p ro represents the speed at which trends spread.

[0012] Furthermore, it also includes a dynamic speed regulation coefficient calculation step, which is used to dynamically adjust the speed regulation sensitivity based on the deviation between the predicted power value and the real-time power. The speed regulation coefficient is calculated in the following manner: Among them, K r eg is the dynamic speed adjustment coefficient, K0 is the basic adjustment coefficient, α is the attenuation coefficient, and P is the dynamic speed adjustment coefficient. p red represents the predicted power output from the power prediction step, P. r eal represents the real-time output power of the turbine, P r ated is the rated power of the turbine, K m in is the minimum adjustment coefficient; when the deviation between the predicted power and the real-time power is small, K... r For example, when the speed is close to K0, the speed adjustment is sensitive and responds quickly to small fluctuations; when the deviation is large, K... r eg drops to near K m The speed adjustment is smooth, extending the service life of the turbine.

[0013] Furthermore, the power prediction step also includes an extreme tidal current condition identification unit, which is used to identify abnormal tidal current conditions caused by extreme weather such as typhoons and cold waves. The unit sets thresholds for sudden changes in tidal current velocity, fluctuations in flow direction, and abnormal seawater density. When the tidal current parameters collected in real time exceed any of these thresholds, it is determined to be an extreme condition. At this time, the unit automatically switches to the extreme condition prediction sub-model. This sub-model is trained based on historical tidal current-power datasets under extreme weather conditions and uses a gradient boosting decision tree algorithm to shorten the prediction step size to 1-5 minutes and control the prediction error within 12%. At the same time, it sends an extreme condition warning to the monitoring and early warning step to remind maintenance personnel to strengthen equipment monitoring.

[0014] Furthermore, the speed regulation decision-making step also includes a multi-objective optimization unit. When formulating the speed regulation strategy, this unit simultaneously considers three objectives: power output stability, equipment loss rate, and power generation efficiency. The unit establishes a multi-objective optimization function, using power fluctuation, turbine torque fluctuation, and gearbox wear coefficient as optimization indicators. It then uses a non-dominated sorting genetic algorithm to solve for the optimal speed regulation scheme, generating 3-5 candidate schemes. Each scheme includes the target speed, guide vane opening adjustment, and gearbox transmission ratio adjustment value. It also labels the power stability score, equipment loss score, and efficiency score of each scheme, allowing the control system to select the optimal strategy based on current operating requirements.

[0015] Furthermore, the execution drive steps also include a hydraulic oil condition monitoring unit, which is used to monitor the hydraulic oil performance of the hydraulic actuator in real time; the unit integrates an oil contamination sensor, a viscosity sensor and a moisture content sensor; when the oil contamination exceeds the standard, the hydraulic oil filtration system is automatically activated.

[0016] Furthermore, the monitoring and early warning process also includes an equipment remaining life prediction unit. This unit predicts the remaining service life of key turbine components based on equipment operating data and degradation models. The unit collects stress data from the blades, vibration data from the gearbox, and wear data from the seals of the hydraulic actuators. Combining the initial performance parameters and life curves of the components, the remaining life is calculated as follows: Where L_rem is the remaining lifetime of the component, L t otal represents the total design life of the component, σ represents the actual stress of the component at time t, and σ m ax represents the maximum allowable stress of the component, v represents the actual operating speed of the component at time t, and v r ated is the rated operating speed of the component, and t is the current operating speed.

[0017] Furthermore, it includes a grid load coordination step, which is used to achieve coordinated matching between turbine power output and grid load demand. This step communicates with the grid dispatch center via Ethernet to obtain the grid load demand curve for the next 1-2 hours in real time. The grid load demand is compared with the turbine's predicted power output output from the power prediction step. When the predicted power exceeds the grid load demand by more than 15%, a power reduction command is sent to the speed regulation decision step to reduce power output by lowering the turbine speed. When the predicted power is lower than the grid load demand by more than 10%, a power increase command is sent to increase power output by raising the speed, while the speed is always within a safe threshold range. After coordinated regulation, the matching degree between turbine power output and grid load is improved to over 90%, reducing energy curtailment and grid peak-shaving pressure.

[0018] Furthermore, the power flow data acquisition step also includes a data anomaly repair unit, which is used to repair missing or abnormal power flow data caused by sensor failure or signal interference. The unit adopts an interpolation algorithm based on Kalman filtering. When missing or abnormal data is detected, the algorithm is called: first, an initial estimate is established by using historical data from the same period and data from adjacent time points, and then iteratively corrected by Kalman filtering. The error of the repaired data is ≤5%. At the same time, the occurrence time, type and repair method of the abnormal data are recorded, a data quality report is generated, and the data integrity rate is maintained above 99.5%.

[0019] Furthermore, the execution drive step also includes a speed regulation buffer unit, which is used to mitigate the impact during the speed regulation process and protect the turbine transmission system. The unit has a built-in buffer spring and damper, which are installed in series at the connection between the hydraulic actuator and the guide vane. When the speed regulation command requires the guide vane opening to change rapidly, the buffer spring absorbs part of the impact force, and the damper slows down the movement speed. At the same time, the unit collects the compression of the buffer spring and the oil temperature of the damper in real time and sends a deceleration command to the speed regulation decision step.

[0020] Compared with existing technologies, the beneficial effects of this invention are as follows: In terms of power prediction, by improving the LSTM neural network to construct the prediction model and combining multi-dimensional power flow parameters with historical data for training, the prediction accuracy is greatly improved. At the same time, a special prediction sub-model is designed for extreme power flow conditions to avoid prediction failure under extreme weather conditions, providing a reliable predictive basis for speed regulation and effectively reducing regulation errors caused by power prediction deviations.

[0021] In terms of speed regulation performance, the system uses a dynamic regulation coefficient calculation module to dynamically adjust the regulation sensitivity based on power deviation, avoiding excessive wear and tear on the equipment caused by frequent speed fluctuations. A multi-objective optimization unit considers power stability, equipment losses, and power generation efficiency, generating regulation schemes adapted to different operating requirements, ensuring power output meets grid requirements while extending equipment lifespan. The buffer unit and hydraulic oil monitoring unit of the drive module further improve the smoothness and reliability of speed regulation, reducing damage to the transmission system from impact loads and lowering the risk of equipment failure.

[0022] The system also boasts excellent environmental adaptability and collaborative capabilities. The tidal current velocity spatiotemporal correction module and data anomaly repair unit eliminate the impact of measurement errors and data anomalies on the system, ensuring accurate and reliable basic data. The grid load coordination module achieves precise matching between turbine power output and grid load demand, reducing energy curtailment and grid peak-shaving pressure, and improving energy utilization efficiency. Furthermore, the equipment remaining life prediction unit and monitoring and early warning module realize health management and anomaly early warning throughout the equipment's entire lifecycle, facilitating preventative maintenance planning by operation and maintenance personnel, reducing unplanned downtime, and ensuring long-term stable system operation. This lays a solid foundation for the large-scale and commercial application of tidal current power generation systems. Attached Figure Description

[0023] Figure 1 This is a schematic block diagram of the tidal current turbine power prediction and speed regulation method proposed in this invention;

[0024] Figure 2 A line graph showing the comparison of power prediction errors;

[0025] Figure 3 A bar chart showing the speed regulation response time;

[0026] Figure 4 This is a line graph showing the degree of grid load coordination and matching. Detailed Implementation

[0027] 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.

[0028] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0029] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. Furthermore, the terms "installed," "connected," and "linked" should be interpreted broadly; for example, they may refer to a fixed connection, a detachable connection, or an integral connection; they may refer to a mechanical connection or an electrical connection; they may refer to a direct connection or an indirect connection through an intermediate medium; and they may refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. The invention will now be described in further detail with reference to the accompanying drawings.

[0030] Reference Figures 1 to 4 A method for predicting the power and regulating the speed of a tidal current turbine, comprising the following steps:

[0031] The core function of the tidal current data acquisition step is to acquire multi-dimensional tidal current parameters in the turbine operating environment in real time and with high accuracy, specifically covering tidal current velocity, tidal current direction, seawater depth, and seawater density. This step integrates multiple types of high-precision sensors. A Doppler current meter is used to measure tidal current velocity, with a measurement range set from 0.2 m / s to 5 m / s and a measurement accuracy controlled within ±0.02 m / s. The sampling frequency is set to 1 Hz to ensure the capture of subtle changes in tidal current velocity. A three-dimensional compass is used to measure tidal current direction, covering a range from 0 degrees to 360 degrees with a measurement accuracy of ±1 degree, accurately reflecting dynamic adjustments in tidal current direction. A pressure level gauge is used to measure seawater depth, with a measurement range from 0 meters to 50 meters and an accuracy of ±0.05 meters, adapting to different nearshore water depth conditions. A density sensor is used to acquire seawater density, with a measurement range set from 1020 kg / m³ to 1030 kg / m³ and an accuracy of ±0.5 kg / m³, avoiding the influence of seawater density differences on power calculations. All data collected by the sensors is transmitted to the data processing unit via 4G / 5G wireless transmission. The signal latency of this transmission is controlled within 50 milliseconds to ensure real-time data transmission. Simultaneously, the system automatically stores nearly 30 days of historical power flow data, with a data sampling interval of 5 minutes. This historical data will be used for the training and optimization of the subsequent power prediction model, providing data support for improving model accuracy.

[0032] The power prediction step utilizes an improved Long Short-Term Memory (LSTM) neural network to construct a dedicated power prediction model for accurate prediction of turbine output power. The model's network structure is specifically designed, comprising three core layers: an input layer, hidden layers, and an output layer. The input layer has eight neurons, corresponding to eight key feature parameters including tidal current velocity, tidal current direction, seawater depth, and seawater density, ensuring the model comprehensively receives tidal environment information. The hidden layer has three layers, each containing 64 neurons, using the ReLU activation function, which effectively solves the gradient vanishing problem and improves model training efficiency and prediction accuracy. The output layer has one neuron specifically for outputting the predicted turbine power value for a future time period. An adaptive learning rate strategy is employed during model training. The initial learning rate is set to 0.01, and it automatically decays by 10% after every 100 iterations to balance training speed and convergence accuracy. Model accuracy is evaluated using the root mean square error (RMSE) to ensure the trained model meets practical application requirements. After the trained model is loaded into the real-time prediction unit, it will continuously receive the real-time power flow parameters transmitted from the power flow data acquisition step, and then output the predicted turbine power values ​​for the next 5 to 30 minutes. According to actual tests, the prediction error of the model can be controlled within 8%, providing a reliable predictive basis for subsequent speed regulation.

[0033] The speed regulation decision-making step, as the core control unit of the entire system, undertakes the critical tasks of receiving power prediction information, analyzing operating status, and formulating speed regulation strategies. This step receives the power prediction value output by the power prediction step in real time, and simultaneously collects the current operating parameters of the turbine, including the turbine's real-time speed, actual output power, and transmission system torque. The step has a pre-set set of comprehensive safe operating thresholds, with an upper speed limit of 150 rpm, a lower speed limit of 30 rpm, and a power fluctuation tolerance of ±10%. These thresholds are determined based on the performance parameters and safety requirements of different turbine models. When the predicted power output by the power prediction step exceeds the set power fluctuation tolerance range, or when the turbine's real-time speed deviates from the rated speed range (set to 50 rpm to 100 rpm depending on the turbine model), the step immediately initiates the speed regulation strategy formulation process. By analyzing the deviation between predicted power and real-time power, and the deviation between real-time speed and rated speed, and combining the inertial characteristics of the turbine and the load-bearing capacity of the transmission system, specific speed adjustment commands are generated. The delay time from the generation of the command to its transmission to the actuator is controlled within 100 milliseconds to ensure rapid response of the adjustment action.

[0034] The execution drive step's main function is to precisely execute the speed regulation commands issued in the speed regulation decision step, achieving dynamic adjustment of the turbine's speed. This step comprises three core components: a hydraulic actuator, a gearbox, and a speed feedback sensor. The hydraulic actuator, as the power source for the actuator, has a rated thrust of 50 kN and a response time controlled within 200 milliseconds, enabling rapid driving of the turbine's guide vanes. The gearbox adjusts the transmission ratio, which can be flexibly adjusted between 5 and 20, maintaining a transmission efficiency above 95% to minimize energy loss during power transmission. The speed feedback sensor is a photoelectric sensor with a measurement range covering 0 to 200 rpm and a measurement accuracy of ±0.1 rpm, capable of real-time acquisition of adjusted turbine speed data. During the specific adjustment process, the hydraulic actuator changes the water flow channel area by adjusting the opening of the turbine guide vanes. The guide vane opening adjustment range is set from 0 to 80 degrees, affecting the rotational speed by changing the force and angle of the water flow impacting the vanes. The gearbox works in conjunction with the guide vane adjustment to synchronously adjust the transmission ratio, further optimizing the speed control accuracy. The speed feedback sensor continuously feeds back the real-time collected speed data to the speed adjustment decision-making process, forming a complete closed-loop control to ensure that the speed adjustment accuracy meets the design requirements.

[0035] The monitoring and early warning step is responsible for real-time monitoring and anomaly warning of the operational status of each step in the entire system, ensuring stable and reliable system operation. This step continuously monitors the working status of each sensor in the power flow data acquisition step, including whether the sensors are acquiring data normally and whether data transmission is interrupted; it also monitors the prediction deviation in the power prediction step, promptly identifying when the prediction error exceeds a set threshold; it also monitors the command generation and transmission in the speed regulation decision step, as well as the operating temperature of the equipment executing the drive step, with gearbox oil temperature controlled below 80 degrees Celsius and hydraulic oil temperature controlled below 65 degrees Celsius. When the step detects an anomaly, such as sensor data interruption, power prediction error exceeding 15%, or equipment operating temperature exceeding the set threshold, it immediately triggers an audible and visual alarm. The sound pressure level of the alarm is set to above 85 decibels to ensure clear detection by maintenance personnel, and the alarm light flashing frequency is set to 2 Hz to enhance the alarm effect through visual cues. Simultaneously, the step automatically generates a fault report, detailing the fault type, fault occurrence time, and preliminary fault troubleshooting suggestions, and sends it wirelessly to the remote maintenance platform to provide guidance for maintenance personnel to handle faults promptly.

[0036] This invention also includes a tidal current velocity spatiotemporal correction step. The core function of this step is to optimize and correct the real-time tidal current velocity obtained from the tidal current data acquisition step. This eliminates the impact of spatial differences between the measurement point and the actual operating location of the turbine, as well as time lag in data transmission, on the accuracy of the tidal current velocity data, providing more accurate basic data for subsequent power prediction. This step achieves data optimization by establishing a dedicated tidal current velocity correction model. During model calculation, key parameters such as the horizontal distance and vertical height between the turbine installation location and the current meter measurement point, as well as the tidal current propagation speed in the sea area, are considered to calculate the corrected real-time tidal current velocity as follows:

[0037]

[0038] Among them, v c orr represents the corrected real-time power flow velocity, measured in meters per second; v m EAS represents the raw tidal velocity measured by the current meter, in meters per second; θ represents the angle between the tidal flow direction and the horizontal distance x, in degrees; L represents the rotation radius of the turbine blades, in meters, this parameter is determined according to the specific turbine model; k represents the spatial correction coefficient, ranging from 0.05 to 0.1, the specific value needs to be determined based on the marine topographic features, such as the presence of reefs, seabed slope, etc.; Δt represents the delay time from data acquisition by the sensor to transmission to the data processing unit, in seconds; v pro represents the tidal current propagation speed in this sea area, measured in meters per second (m / s). This parameter is obtained by analyzing historical tidal current data over the past year, typically ranging from 0.3 m / s to 0.8 m / s. After optimizing the original tidal current speed data using this correction model, the measurement error of the tidal current speed decreased from ±0.1 m / s before correction to ±0.03 m / s after correction, significantly improving the accuracy of the tidal current speed data. This provides more accurate input data for the power prediction step, further reducing the power prediction error and ensuring that the power prediction results better reflect actual operating conditions.

[0039] This invention also includes a dynamic speed regulation coefficient calculation step. The main function of this step is to dynamically adjust the sensitivity during speed regulation based on the deviation between the predicted power value output from the power prediction step and the actual output power of the turbine. This prevents excessive wear on key components such as the turbine transmission system and blades due to frequent and drastic speed fluctuations, thereby extending the overall service life of the equipment. This step calculates the real-time speed regulation coefficient by establishing a regulation coefficient calculation model as follows:

[0040]

[0041] Among them, K r eg represents the dynamically calculated speed regulation coefficient, ranging from 0.2 to 1.5. A larger coefficient value indicates higher speed regulation sensitivity. K0 represents the basic regulation coefficient, ranging from 1.0 to 1.2. The specific value needs to be determined based on the turbine's inertial parameters, such as moment of inertia and transmission system damping. α represents the attenuation coefficient, ranging from 2 to 5. This coefficient controls the attenuation rate of the regulation coefficient as the power deviation changes; a larger coefficient results in a faster attenuation rate as the deviation increases. P p red represents the predicted power of the turbine over the future time period output by the power prediction step, in kilowatts; P r eal represents the current real-time output power of the turbine, in kilowatts; P r ated represents the rated output power of the turbine, in kilowatts, and is determined based on the turbine's design parameters; K m `in` represents the minimum adjustment coefficient, ranging from 0.2 to 0.3. This parameter is set to prevent the speed regulation from failing due to an excessively small adjustment coefficient, ensuring that basic speed regulation functions can still be achieved even with large power deviations. When the deviation between the predicted power value and the real-time power is small, for example, less than or equal to 5%, the calculated K... r For example, when the speed regulation coefficient is close to the baseline regulation coefficient K0, the speed regulation system maintains high sensitivity and can quickly respond to small power fluctuations, ensuring stable power output; when the power deviation is large, such as greater than 10%, K... r e.g., it will drop to near the minimum adjustment coefficient K.m In this way, the speed adjustment becomes smoother, avoiding severe impacts on the equipment due to excessive adjustment, effectively reducing component wear, and extending the operation and maintenance cycle and service life of the turbine.

[0042] In this invention, the power prediction step also includes an extreme tidal current condition identification unit. The core function of this unit is to accurately identify abnormal tidal current conditions caused by extreme weather events such as typhoons, cold waves, and storm surges. This prevents conventional power prediction models from failing due to abnormal data distribution under extreme conditions, ensuring the accuracy and reliability of power predictions during extreme weather. This unit identifies the conditions by pre-setting multi-dimensional extreme condition judgment thresholds, specifically including tidal current velocity mutation thresholds, flow direction fluctuation thresholds, and seawater density anomaly thresholds. The tidal current velocity mutation threshold is set at a change in tidal current velocity greater than 1 meter per second within 5 minutes. When a drastic change in tidal current velocity is detected within a short period, it is preliminarily determined that extreme conditions may have occurred. The flow direction fluctuation threshold is set at a change in tidal current direction greater than 30 degrees within 10 minutes. Significant fluctuations in tidal current direction are a typical characteristic of extreme weather effects. The seawater density anomaly threshold is set at a deviation of the actual measured seawater density from the normal density range of the sea area within ±2 kg / m³. Extreme weather may cause intense seawater mixing, leading to density anomalies. When the real-time collected tidal current parameters exceed any of the above thresholds, the unit immediately determines that the current condition is an extreme tidal current condition. At this point, the unit automatically switches to a pre-trained extreme condition prediction sub-model. This sub-model is trained on a large dataset of historical tidal currents and power data under extreme weather conditions. The dataset contains over 500 sets of tidal current parameters and corresponding turbine power data under different extreme conditions, ensuring the model has good generalization ability. The extreme condition prediction sub-model uses the Gradient Boosting Decision Tree (GBDT) algorithm, which has advantages in handling nonlinear and outlier data. To improve the real-time performance of predictions, the prediction step size is shortened from the usual 5 minutes to 1-5 minutes. Field tests have verified that the power prediction error of this sub-model under extreme conditions can be controlled within 12%, meeting the basic prediction requirements under extreme weather conditions. Furthermore, while switching to the extreme condition prediction sub-model, the unit sends an extreme condition warning signal to the monitoring and early warning system, reminding maintenance personnel to strengthen the monitoring of turbine operating status and prepare for equipment protection and emergencies in advance.

[0043] In this invention, the speed regulation decision-making step also includes a multi-objective optimization unit. When formulating the speed regulation strategy, this unit no longer solely pursues power output stability, but simultaneously considers three core objectives: power output stability, equipment operating loss rate, and power generation efficiency, achieving synergistic optimization of multi-dimensional performance. This unit constructs an optimization model by establishing a multi-objective optimization function, using power fluctuation, turbine torque fluctuation, and gearbox wear coefficient as key optimization indicators. The power fluctuation control target is set to not exceed 5% of the turbine's rated power, i.e., ΔP ≤ 5%P. r The power output stability is ensured to meet the grid connection requirements; the turbine torque fluctuation control target is set to not exceed 8% of the rated torque, i.e., ΔT≤8%T. r To prevent drastic torque fluctuations from causing excessive load on the transmission system, the gearbox wear coefficient is controlled to be no more than 0.02 (W≤0.02), reducing the wear rate of the gearbox and extending its service life. To solve this multi-objective optimization problem, the unit employs a non-dominated sorting genetic algorithm (NSGA-II). This algorithm can find a set of Pareto optimal solutions under multi-objective conflict, ultimately generating 3 to 5 candidate speed regulation schemes. Each candidate scheme includes detailed execution parameters such as the target speed, guide vane opening adjustment, and gearbox transmission ratio adjustment. Simultaneously, each scheme's power stability, equipment loss level, and power generation efficiency are quantitatively scored, ranging from 1 to 10 points, with higher scores indicating better performance. The control system can select the optimal scheme based on current operational needs. For example, when the grid has high power stability requirements, schemes with a power stability score of at least 8 points are prioritized; when in the later stages of the equipment maintenance cycle, and equipment loss needs to be controlled, schemes with an equipment loss score of at least 8 points are prioritized, ensuring the speed regulation strategy better suits the actual application scenario.

[0044] In this invention, the execution drive step also includes a hydraulic oil condition monitoring unit. The main function of this unit is to monitor various performance indicators of the hydraulic oil in the hydraulic actuator in real time, promptly detect problems such as oil contamination, aging, and excessive moisture, and prevent the hydraulic actuator from experiencing reduced adjustment accuracy, motion jamming, or even malfunction due to hydraulic oil performance degradation, ensuring long-term stable operation of the hydraulic actuator. This unit integrates multiple types of oil monitoring sensors. Among them, the oil contamination sensor detects the impurity content in the hydraulic oil, with a measurement range covering levels 10 to 20 of the ISO4406 standard and a measurement accuracy of ±1 level, accurately identifying the level of particulate contaminants in the oil; the viscosity sensor measures the viscosity of the hydraulic oil, with a measurement range set from 20 centistokes to 200 centistokes and an accuracy controlled within ±5 centistokes; viscosity is a key parameter affecting the fluidity and transmission efficiency of hydraulic oil; and the moisture content sensor detects the moisture content in the hydraulic oil, with a measurement range from 0 ppm to 1000 ppm and an accuracy of ±50 ppm. Excessive moisture can lead to hydraulic oil emulsification and corrosion of internal equipment components. The unit's monitoring frequency is set to collect data every 10 minutes. The collected data is compared in real time with preset standard thresholds. The standard thresholds for oil contamination are set to not exceed ISO 440616 / 13 level, hydraulic oil viscosity to 30-100 centistokes, and moisture content to not exceed 300 ppm. When the oil contamination level exceeds the standard threshold, the unit automatically activates the hydraulic oil filtration system. This system has a filtration accuracy of 10 microns and a flow rate of 10 liters per minute, quickly filtering impurities and reducing contamination. When the hydraulic oil viscosity exceeds the standard range or the moisture content exceeds the standard, the unit immediately sends an oil replacement reminder signal to the monitoring and early warning system. It also automatically records the cumulative usage time of the hydraulic oil. When the cumulative usage time exceeds 3000 hours, a mandatory oil replacement reminder is sent to ensure the hydraulic oil is always in good working condition. Through continuous monitoring and maintenance by this unit, the adjustment accuracy of the hydraulic actuator can be maintained above 98% for a long period, avoiding adjustment failures caused by oil problems.

[0045] In this invention, the monitoring and early warning step also includes an equipment remaining life prediction unit. This unit, based on real-time operating data and degradation models of key turbine components, accurately predicts the remaining service life of the components, providing data support for maintenance personnel to formulate preventative maintenance plans and reducing unplanned downtime caused by sudden component failures. This unit continuously collects operating status data of key turbine components. For turbine blades, real-time stress data is collected using strain gauges attached to the blade surface. The strain gauge measurement range is set to 0 MPa to 500 MPa, with a measurement accuracy of ±5 MPa. The stress data directly reflects the load borne by the blades. For the gearbox, vibration data is collected using accelerometers installed on the gearbox housing. The accelerometer measurement range is set to 0 g to 50 g, with an accuracy of ±0.1 g. The vibration data can be used to analyze the wear and failure status of gears and bearings inside the gearbox. For hydraulic actuators, wear data of the seals is collected using displacement sensors. The displacement sensor measurement range is set to 0 mm to 5 mm, with an accuracy of ±0.02 mm. The wear of the seals directly affects the sealing performance and adjustment accuracy of the hydraulic actuator. The unit combines this real-time operational data with the initial performance parameters and preset life curves of the components to calculate the remaining life of each key component in the following manner:

[0046]

[0047] Among them, L r em represents the remaining lifespan of a component, measured in hours; L t `otal` represents the total design life of the component, expressed in hours. It is determined based on the component's material, structure, and operating environment. Specifically, the total design life of the blade is set at 20,000 hours, the gearbox at 15,000 hours, and the hydraulic actuator at 8,000 hours. `σ(t)` represents the actual stress the component experiences at time `t`, expressed in megapascals (MPa). m ax represents the maximum allowable stress of the component, measured in megapascals (MPa), and is determined by the mechanical properties of the component material; v(t) represents the actual operating speed of the component at time t, measured in revolutions per minute (rpm) for blades and gearboxes, and in meters per second (m / s) for hydraulic actuators; v r "ated" represents the rated operating speed of the component, with the unit being the same as the actual operating speed; "t" represents the cumulative time the component has been running, in hours. When the calculated remaining lifespan of a component is less than 10% of its total design lifespan, the unit will immediately send a component replacement reminder signal to the remote operation and maintenance platform, reminding operation and maintenance personnel to arrange spare parts procurement and maintenance personnel deployment in advance to ensure that the replacement is completed before the component's lifespan expires, avoiding equipment downtime due to component failure and ensuring the continuous and stable operation of the tidal power generation system.

[0048] This invention also includes a grid load coordination step. The core function of this step is to achieve coordinated matching between the power output of the hydro turbine and the load demand of the grid, avoiding energy waste caused by excessive turbine output power or impacts on grid frequency and voltage stability due to insufficient power, thus improving the compatibility of the tidal power generation system with the grid. This step establishes a stable communication connection with the grid dispatch center via Ethernet, using the Modbus-TCP protocol with a baud rate set to 1 Mbps to ensure high-speed and reliable data transmission. The step obtains the grid load demand curve for the next 1 to 2 hours from the grid dispatch center in real time. The data interval for this curve is set to 15 minutes, and the curve contains detailed information such as the peak load, valley load, and load fluctuation range for the future time period. The process involves real-time comparison and analysis of the acquired grid load demand data with the turbine future power prediction data output by the power prediction step. When the predicted turbine power exceeds the grid load demand by more than 15%, a power reduction command is immediately sent to the speed regulation decision step. Upon receiving the command, the speed regulation decision step formulates a corresponding speed reduction strategy, decreasing the turbine speed to reduce power output, with the maximum speed reduction controlled within 20% to ensure the speed remains within the safe operating threshold. Conversely, when the predicted turbine power is more than 10% lower than the grid load demand, the process sends a power increase command to the speed regulation decision step. The speed regulation decision step then formulates a speed increase strategy, increasing the turbine speed to boost power output, with the maximum speed increase controlled within 15% to avoid overloading the equipment due to excessive speed. Through this coordinated regulation method, the matching degree between turbine power output and grid load demand can be improved to over 90%, significantly reducing energy curtailment during tidal power generation, lowering the grid's peak-shaving pressure, and improving the overall stability of the power system.

[0049] In this invention, the tidal current data acquisition step also includes a data anomaly repair unit. The main function of this unit is to repair tidal current data loss or anomalies caused by factors such as temporary sensor malfunctions or interference from marine environmental signals, ensuring the continuity and integrity of the tidal current data and providing reliable input data support for the power prediction and speed regulation decision-making steps. This unit uses a Kalman filter-based interpolation algorithm for data repair. When the unit detects missing data (i.e., no valid data received from three consecutive sampling points) or anomalies (i.e., the acquired data exceeds the normal physical range by more than twice, such as a tidal current speed exceeding 10 meters per second or a seawater density below 1000 kilograms per cubic meter), it immediately initiates the data repair process. The repair process first establishes an initial estimate by calling historical data from the same period within the past three months under similar weather conditions, combined with valid data from adjacent times before and after the missing or anomaly. Then, iterative correction is performed using the Kalman filter algorithm, with the number of iterations set to 5 to 10. Each iteration adjusts the data value based on the new error estimate, ultimately obtaining highly accurate repaired data. Through actual testing and verification, the error between the repaired data and the actual data can be controlled within 5%, which can meet the needs of subsequent steps. Simultaneously, the unit records in detail the occurrence time and type of abnormal data (e.g., missing data, exceeding limits), as well as the specific repair methods and repair errors. It regularly generates data quality reports, including key indicators such as data integrity rate, data anomaly rate, and data repair success rate. Through continuous optimization of the repair algorithm, it ensures that the data integrity rate of the power flow data acquisition step is maintained above 99.5% over the long term, laying a data foundation for the stable operation of the entire system.

[0050] In this invention, the execution drive step also includes a speed regulation buffer unit. The core function of this unit is to alleviate the impact load between the hydraulic actuator and the turbine guide vanes and transmission system during speed regulation, protecting precision components such as gears and bearings in the turbine transmission system, reducing the risk of equipment failure, and extending the service life of components. This unit incorporates two core components: a buffer spring and a damper. The buffer spring is made of high-strength alloy material with a stiffness coefficient set at 5000 N / m and a maximum deformation controlled within 20 mm, effectively absorbing the impact force generated during regulation. The damper uses a hydraulic damping structure with a damping coefficient set at 100 N·s / m, which can slow down the speed change of moving parts through damping. The buffer spring and damper are installed in series at the connection between the hydraulic actuator and the turbine guide vanes, forming a complete buffer structure. When the speed regulation decision-making step issues a command requiring a rapid change in the guide vane opening, such as a change exceeding 10 degrees within one second, the buffer spring will first absorb part of the impact force, preventing it from being directly transmitted to the guide vane and transmission system. Simultaneously, the damper will act as a damper, slowing the guide vane's movement speed and reducing the rate of change of the guide vane opening from 0.2 degrees per millisecond without buffering to 0.05 degrees per millisecond, significantly reducing motion impact. Furthermore, the unit will collect real-time data on the compression of the buffer spring and the operating temperature of the damper. A displacement sensor monitors the spring compression, and a temperature sensor monitors the damper oil temperature. When the spring compression exceeds 15 mm or the damper oil temperature exceeds 70 degrees Celsius, the unit will immediately send a deceleration command to the speed regulation decision-making step. Upon receiving the command, the speed regulation decision-making step will further reduce the guide vane opening adjustment speed to prevent overload damage to the buffer components. Through the function of this buffer unit, the impact load on the turbine transmission system during speed regulation can be reduced by more than 60%. Long-term operation has verified that the fatigue life of the gearbox can be extended by 30%, significantly reducing equipment failures caused by impact loads and improving the operational reliability of the entire drive process.

[0051] The following two examples further illustrate the specific implementation of this system:

[0052] Example 1: Power prediction and speed regulation of tidal current turbine under normal offshore operating conditions (turbine model: TL-50, rated power 50kW, rated speed 70r / min)

[0053] 1. Detailed configuration and parameter settings of system modules

[0054] The tidal flow data acquisition module is mounted on a fixed bracket 10m upstream of the turbine. The Doppler current meter is an acoustic Doppler current profiler with a sampling frequency of 1Hz and a measurement range of 0.2-5m / s. Regular calibration ensures an accuracy of ±0.02m / s. The three-dimensional compass is a high-precision magnetoresistive compass, outputting flow direction data every 1 second, with a measurement range of 0-360° and an accuracy of ±1°. The pressure level gauge is installed 2m underwater on the bracket, converting water pressure to water depth, with a measurement range of 0-50m and an accuracy of ±0.05m. The density sensor is a vibrating tube density meter, collecting seawater density data every 2 seconds, with a measurement range of 1020-1030 kg / m³. 3 Accuracy ±0.5kg / m 3 Data transmission uses a 5G module, with latency controlled within 50ms. Historical data storage period is 30 days, sampling interval is 5 minutes, and stored data includes tidal current velocity, direction, water depth, density, and corresponding turbine power and speed at each moment.

[0055] Power prediction module: An improved LSTM model was built using the Python TensorFlow framework. The input layer has 8 neurons corresponding to tidal current velocity, direction, water depth, density, average velocity over the past 10 minutes, average direction over the past 10 minutes, water temperature, and wave height. There are 3 hidden layers, each with 64 neurons, using the ReLU activation function and a Dropout layer (dropout rate = 0.2) to prevent overfitting. The output layer has 1 neuron that outputs power prediction values ​​for 5-30 minutes. The model training dataset consists of historical data from the past 30 days (8640 sets in total), divided into training and testing sets in a 7:3 ratio. The initial learning rate is 0.01, decreasing by 10% every 100 iterations. Training stopped when the RMSE on the validation set stabilized below 3%, and the final prediction error was ≤8%.

[0056] Speed ​​regulation decision module: The safety threshold is set at a speed of 30-150 r / min, power fluctuation ±10% (i.e., 45-55 kW), and rated speed of 70 r / min. The module uses an STM32H743 microcontroller as the control core. It receives the power prediction value and real-time operating parameters (speed, power, torque) every 100 ms. When the predicted power is >55 kW or <45 kW, or the real-time speed is >77 r / min (+10%) or <63 r / min (-10%), the regulation strategy is initiated. After the instruction is generated, it is transmitted to the execution driver module via the CAN bus with a delay of ≤100 ms.

[0057] The actuator drive module uses a double-acting hydraulic cylinder with a rated thrust of 50kN and a response time of ≤200ms. The guide vane opening is adjusted from 0-80° by controlling the oil flow through a proportional valve. The gearbox has a transmission ratio of 5-20, employs a helical gear structure, and has an efficiency of ≥95%. The speed feedback sensor is a photoelectric encoder with a resolution of 1000 lines / revolution, a measurement range of 0-200r / min, and an accuracy of ±0.1r / min, outputting speed data every 1ms. The buffer unit has a buffer spring with a stiffness coefficient of 5000N / m, a maximum deformation of 20mm, and a damper with a damping coefficient of 100N·s / m, and is installed in series between the hydraulic cylinder and the guide vane connecting rod.

[0058] Auxiliary Modules: In the tidal current velocity spatiotemporal correction module, the turbine blade rotation radius L = 2.5m, horizontal distance x = 10m, vertical height y = 2m, spatial correction coefficient k = 0.08 (set based on the flat seabed topography of this sea area), data transmission delay Δt = 0.05s, and tidal current propagation velocity v_pro = 0.5m / s (derived from data collected over the past year); In the dynamic speed regulation coefficient calculation module, the basic regulation coefficient K0 = 1.1, attenuation coefficient α = 3, minimum regulation coefficient Kmin = 0.25, and rated power Prated = 50kW; In the equipment remaining life prediction unit, the blade design total life is 20000h, gearbox 15000h, hydraulic actuator 8000h, blade maximum allowable stress σmax = 450MPa, gearbox rated speed 70r / min, and hydraulic actuator rated operating speed 0.1m / s.

[0059] 2. System operation process and formula application

[0060] Current tidal data acquisition and correction: Sensors acquire data in real time, and the current velocity spatiotemporal correction module calculates the correction value according to the formula: Suppose at some moment v meas =2.5m / s, θ =30°, substituting these values ​​gives v corr =2.5×(1+(10×0.866+2×0.5) / 2.5×0.08)-(0.025 / 2.5)×100=2.5×(1+9.66 / 2.5×0.08)-1=2.5×1.309-1=2.27m / s. After correction, the speed error is reduced from ±0.1m / s to ±0.03m / s.

[0061] Power prediction and speed regulation: The corrected data is input into the power prediction module, which outputs the predicted power P for the next 10 minutes. pred =58kW (exceeding the 55kW threshold). The dynamic speed regulation coefficient calculation module calculates K according to the formula. reg : (At this time, the real-time power P) real =52kW), K reg With a sensitivity of approximately 1.02 (high sensitivity), the speed regulation decision module formulates a strategy: the guide vane opening is reduced from the current 60% to 50%, the gearbox transmission ratio is adjusted from 10 to 11, and the target speed is reduced to 65 r / min. After receiving the command, the drive module executes the action, the hydraulic cylinder pushes the guide vane, and the buffer unit reduces the opening change rate from 0.2° / ms to 0.05° / ms. The speed feedback sensor provides real-time speed feedback, forming a closed-loop control. After 1.5s, the speed stabilizes at 65 r / min, and the power is reduced to 54kW (within ±10%).

[0062] Equipment Monitoring and Early Warning: The equipment remaining life prediction unit calculates the remaining life of the blades according to the formula: (Assuming the current running time is t = 3000h, σ(t) average 200MPa, n(t) average 70r / min), the integral result ≈ 3000 × (200 / 450) × (70 / 70) = 1333h, therefore L rem ≈20000-1333=18667h, no need to replace; the hydraulic oil condition monitoring unit collects data every 10 minutes, with contamination level ISO440615 / 12 (≤16 / 13), viscosity 60cSt (30-100cSt), moisture 200ppm (≤300ppm), and the condition is normal.

[0063] 3. Performance Comparison Data

[0064] Table 1: Comparison of System Performance under Normal Operating Conditions

[0065] index This system Traditional system Differences in performance Power prediction error ≤8％ 15％-20％ Accuracy improved by 40%-60% Speed ​​adjustment response time ≤1.5s 5-8s Response speed improved by 70%+ Power fluctuation range ±10％ ±20％-25％ Stability improved by 50%+ Equipment failure rate <0.5 times / month 2-3 times / month Failure rate reduced by 75%+

[0066] Table 1 shows data based on 30 days of continuous operation and testing, demonstrating the significant advantages of this system. Power prediction error has been reduced from 15%-20% in traditional systems to ≤8%, thanks to the improved LSTM model combined with spatiotemporal correction data, resulting in more accurate capture of power flow patterns. Speed ​​regulation response time has been shortened from 5-8s to ≤1.5s, benefiting from rapid command transmission and buffer regulation, avoiding power overshoot. Power fluctuation range has been narrowed from ±20%-25% to ±10%, meeting grid connection requirements. Equipment failure rate has been significantly reduced due to dynamic adjustment coefficients reducing impact, buffer units, and oil level monitoring protecting key components, minimizing unplanned downtime, and improving power generation efficiency and economic benefits.

[0067] Example 2: Power prediction and speed regulation of tidal current turbine under extreme typhoon conditions (same model TL-50 turbine)

[0068] 1. Customized configuration of system modules

[0069] Tidal data acquisition module: During typhoons, the sampling frequency is increased to 5Hz, the sensor is equipped with a wave-proof protective cover, the Doppler current meter range is extended to 0-10m / s, the density sensor is equipped with an impact-resistant shell, and data transmission adopts 5G+satellite dual-mode backup to avoid single communication interruptions. The data anomaly repair unit monitors once every 100ms. When data from three consecutive sampling points is missing or exceeds the normal range (velocity > 10m / s, flow direction change > 30° / 10min), Kalman filter interpolation repair is immediately initiated.

[0070] Power prediction module: The threshold for the extreme power flow condition identification unit is set as follows: 5-minute velocity change > 1 m / s, 10-minute flow direction change > 30°, and density deviation ± 2 kg / m³. 3 The extreme condition prediction sub-model is trained based on 500 sets of historical typhoon data (including three typhoon events in this sea area from 2021 to 2023). It adopts the GBDT algorithm, adds wind speed, wind direction, and air pressure as input features, and has a prediction step size of 1-5 minutes. The parameters are optimized through grid search (learning rate 0.05, tree depth 8, estimators=200), and the final prediction error is ≤12%.

[0071] Speed ​​regulation decision module: Under extreme operating conditions, the safety threshold is adjusted to 30-120 r / min (lowering the upper limit to avoid overload), with power fluctuations of ±15% (42.5-57.5 kW). The multi-objective optimization unit prioritizes solutions with equipment loss scores ≥9 to reduce impact. The module communicates with the remote operation and maintenance platform in real time, uploading operating data every 5 seconds and receiving emergency commands.

[0072] Execution drive module: The control accuracy of the proportional valve of the hydraulic actuator is improved, the adjustment step is reduced from 0.5° to 0.1°, the gearbox is equipped with overload protection (automatic disconnection when torque > 120% of rated value), the monitoring frequency of spring compression of the buffer unit is increased to 1kHz, and the forced cooling fan is activated when the oil temperature exceeds 70℃.

[0073] 2. Extreme operating conditions and formula application

[0074] Operating Condition Identification and Forecasting Switching: Two hours before typhoon landfall, the tidal current velocity rapidly increased from 3 m / s to 4.2 m / s (a 1.2 m / s change over 5 minutes, exceeding the 1 m / s threshold), and the flow direction changed from 180° to 215° (a 35° change over 10 minutes, exceeding the 30° threshold). The extreme operating condition identification unit determined that extreme operating conditions had been entered, and the power prediction module automatically switched to the GBDT sub-model, outputting the predicted power P for the next 5 minutes. pred =62kW (exceeding the 57.5kW threshold).

[0075] Dynamic adjustment and safety control: The dynamic speed adjustment coefficient calculation module calculates K according to the formula. reg : (Real-time power P_real = 55kW), K reg ≈0.98 (slightly reduced sensitivity), multi-objective optimization unit generates the following scheme: guide vane opening is reduced from 65% to 40%, transmission ratio is adjusted from 10 to 15, target speed is 60 r / min (equipment loss score 9.2). When the drive module is executed, the buffer unit reduces the guide vane opening change rate to 0.03° / ms to avoid impact. The speed feedback display stabilizes at 60 r / min after 5s, and the power is reduced to 56kW (within ±15%).

[0076] Remaining life prediction and early warning: The equipment remaining life prediction unit calculates the gearbox's remaining life according to the formula: (Maximum allowable stress σ of gearbox) max =350MPa, during the typhoon the average σ(t) is 280MPa, n(t) average is 60r / min), the integral result ≈3000×(280 / 350)×(60 / 70)=3000×0.8×0.857≈2057h, therefore L rem ≈15000-2057=12943h, still within the safe range, but a reminder was sent to the operation and maintenance platform that "gear wear needs to be checked after the typhoon".

[0077] 3. Performance comparison data under extreme operating conditions

[0078] Table 2: Comparison of System Performance under Extreme Typhoon Conditions

[0079]

[0080]

[0081] Table 2 data is based on 24-hour operational measurements during a typhoon, demonstrating the system's significant advantages under extreme conditions. Traditional systems, lacking dedicated extreme weather prediction models, suffer from errors of 30%-40%, leading to inaccurate adjustments and equipment overload rates of 30%-40%. This system, through its GBDT sub-model, controls errors to ≤12%, and its dynamic adjustment coefficient and buffer unit reduce impact loads by over 60%, completely avoiding overload. Emergency response success rate is 100%, thanks to dual-mode communication and real-time early warning ensuring timely command transmission and fault handling. In contrast, traditional systems, due to communication interruptions and prediction failures, have an emergency success rate of only 50%-60%. This proves that this system can ensure the safe and stable operation of the turbine under extreme weather conditions, significantly improving the environmental adaptability of the tidal power generation system.

[0082] Reference Figure 2This line graph visually illustrates the differences in prediction accuracy among different models. The error of the traditional LSTM model increases significantly with increasing sample number, reaching a maximum of 19.1%, due to its failure to consider the spatiotemporal characteristics of power flow and adaptation to extreme operating conditions. The improved LSTM model of this invention maintains an error of ≤7.8%, thanks to multi-feature input and adaptive learning rate optimization, resulting in more accurate capture of power flow change patterns. Although the error of the extreme operating condition GBDT sub-model is slightly higher than that of the improved LSTM, it remains stable within 11.5% under complex operating conditions, demonstrating its ability to handle abnormal data. The graph verifies the technical advantages of the prediction module of this invention, providing a reliable data foundation for speed regulation.

[0083] Reference Figure 3 This bar chart clearly reflects the balance between speed regulation speed and stability. Traditional open-loop control response times exceed 6.8 seconds, exhibiting significant lag due to a lack of real-time feedback and optimization algorithms. The unbuffered closed-loop control of this invention reduces the response time to 1.5-2.1 seconds, thanks to rapid command transmission and dynamic adjustment coefficients, resulting in a response speed improvement of over 70%. The buffered closed-loop control response time only increases by 0.2-0.3 seconds, yet the buffer unit significantly reduces impact loads, demonstrating that rapid response is combined with equipment protection. The chart illustrates that the execution drive module of this invention achieves an optimal balance between response speed and operational safety.

[0084] Reference Figure 4 This line graph visually illustrates the synergistic effect between power output and grid demand. Traditional systems exhibit significant deviations between output and load demand; for example, outputting 58kW (+16%) when the load is 50kW, and 65kW (+8.3%) when the load is 60kW, due to the lack of a coordinated adjustment mechanism. The system of this invention consistently outputs close to the demand curve, with deviations ≤ ±2kW, achieving a matching degree exceeding 90%. This is because the grid load coordination module adjusts the rotational speed in real time, dynamically matching load changes while ensuring safety. This reduces energy curtailment and grid peak-shaving pressure, improving the energy utilization efficiency and commercial value of tidal power.

[0085] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for predicting the power and regulating the speed of a tidal current turbine, characterized in that, Includes the following steps: The tidal flow data acquisition step is used to obtain tidal flow parameters of the turbine operating environment in real time, including tidal flow velocity, direction, water depth and seawater density. This step integrates a Doppler current meter, a three-dimensional compass, a pressure level gauge and a density sensor. The data is sent to the data processing unit via 4G / 5G wireless transmission and the historical tidal flow data of the past 30 days is stored for model training. The power prediction step involves building a prediction model based on an improved long short-term memory neural network. This model includes an input layer, hidden layers, and an output layer; an adaptive learning rate is used during model training. The speed regulation decision-making step, as the core control unit, receives the predicted power value from the power prediction step and, in conjunction with the current operating parameters of the turbine, formulates a speed regulation strategy. This step incorporates a safety threshold, which generates a speed adjustment command when the predicted power exceeds the threshold or the real-time speed deviates from the rated value. The execution drive step is used to execute the instructions for the speed regulation decision step, including a hydraulic actuator, a gearbox and a speed feedback sensor; the hydraulic actuator changes the water flow channel area by adjusting the opening of the turbine guide vanes, the gearbox adjusts the transmission ratio, and the speed feedback sensor collects the adjusted speed data in real time to form a closed-loop control; The monitoring and early warning system monitors the operational status of each step in real time, including sensor failures in the power flow data acquisition step, prediction deviations in the power prediction step, command execution in the speed regulation decision step, and equipment temperature in the drive execution step. When an anomaly is detected, an audible and visual alarm is triggered, a fault report is generated, and the report is sent to the remote operation and maintenance platform.

2. The method for predicting the power and regulating the speed of a tidal current turbine according to claim 1, characterized in that, It also includes a tidal current velocity spatiotemporal correction step, which optimizes the real-time tidal current velocity acquired in the tidal current data acquisition step and eliminates the impact of spatial location differences and time lag on data accuracy. This step establishes a tidal current velocity correction model, combining the horizontal distance and vertical height between the turbine installation location and the current meter measurement point, as well as the tidal current propagation speed, to calculate the corrected real-time tidal current velocity in the following manner: Among them, v c orr is the corrected real-time power flow velocity, v m EAS is the raw tidal current velocity measured by the current meter, θ is the angle between the tidal current direction and the horizontal distance x, L is the rotation radius of the turbine blades, k is the spatial correction factor, Δt is the data transmission delay time, and v p ro represents the speed at which trends spread.

3. The method for predicting the power and regulating the speed of a tidal current turbine according to claim 1, characterized in that, It also includes a dynamic speed regulation coefficient calculation step, which is used to dynamically adjust the speed regulation sensitivity based on the deviation between the predicted power value and the real-time power. The speed regulation coefficient is calculated in the following manner: Among them, K r eg is the dynamic speed adjustment coefficient, K0 is the basic adjustment coefficient, α is the attenuation coefficient, and P is the dynamic speed adjustment coefficient. p red represents the predicted power output from the power prediction step, P. r eal represents the real-time output power of the turbine, P r ated is the rated power of the turbine, K m in is the minimum adjustment coefficient; when the deviation between the predicted power and the real-time power is small, K... r For example, when the speed is close to K0, the speed adjustment is sensitive and responds quickly to small fluctuations; when the deviation is large, K... r eg drops to near K m The speed adjustment is smooth, extending the service life of the turbine.

4. The method for predicting the power and regulating the speed of a tidal current turbine according to claim 1, characterized in that, The power prediction step also includes an extreme tidal current condition identification unit, which is used to identify abnormal tidal current conditions caused by extreme weather such as typhoons and cold waves. The unit sets thresholds for sudden changes in tidal current velocity, fluctuations in flow direction, and abnormal seawater density. When the tidal current parameters collected in real time exceed any of these thresholds, it is determined to be an extreme condition. At this time, the unit automatically switches to the extreme condition prediction sub-model. This sub-model is trained based on historical tidal current-power datasets under extreme weather conditions and uses a gradient boosting decision tree algorithm to shorten the prediction step size to 1-5 minutes and control the prediction error within 12%. At the same time, it sends an extreme condition warning to the monitoring and early warning step to remind maintenance personnel to strengthen equipment monitoring.

5. The method for predicting the power and regulating the speed of a tidal current turbine according to claim 1, characterized in that, The speed regulation decision-making process also includes a multi-objective optimization unit. When formulating a speed regulation strategy, this unit considers three objectives simultaneously: power output stability, equipment loss rate, and power generation efficiency. The unit establishes a multi-objective optimization function, using power fluctuation, turbine torque fluctuation, and gearbox wear coefficient as optimization indicators. It then uses a non-dominated sorting genetic algorithm to solve for the optimal speed regulation scheme, generating 3-5 candidate schemes. Each scheme includes the target speed, guide vane opening adjustment, and gearbox transmission ratio adjustment value. It also labels each scheme with power stability score, equipment loss score, and efficiency score, allowing the control system to select the optimal strategy based on current operating requirements.

6. The method for predicting the power and regulating the speed of a tidal current turbine according to claim 1, characterized in that, The execution drive steps also include a hydraulic oil condition monitoring unit, which is used to monitor the hydraulic oil performance of the hydraulic actuator in real time; the unit integrates an oil contamination sensor, a viscosity sensor and a moisture content sensor; when the oil contamination exceeds the standard, the hydraulic oil filtration system is automatically activated.

7. The method for predicting the power and regulating the speed of a tidal current turbine according to claim 1, characterized in that, The monitoring and early warning process also includes an equipment remaining life prediction unit. This unit predicts the remaining service life of key turbine components based on equipment operating data and degradation models. The unit collects stress data from the blades, vibration data from the gearbox, and wear data from the seals of the hydraulic actuators. Combining the initial performance parameters and life curves of the components, the remaining life is calculated as follows: Where L_rem is the remaining lifetime of the component, L t otal represents the total design life of the component, σ represents the actual stress of the component at time t, and σ m ax represents the maximum allowable stress of the component, v represents the actual operating speed of the component at time t, and v r ated is the rated operating speed of the component, and t is the current operating speed.

8. The method for predicting the power and regulating the speed of a tidal current turbine according to claim 1, characterized in that, It also includes a grid load coordination step, which is used to achieve coordinated matching between turbine power output and grid load demand. This step communicates with the grid dispatch center via Ethernet to obtain the grid load demand curve for the next 1-2 hours in real time. The grid load demand is compared with the turbine's predicted power output from the power prediction step. When the predicted power exceeds the grid load demand by more than 15%, a power reduction command is sent to the speed regulation decision step to reduce power output by lowering the turbine speed. When the predicted power is lower than the grid load demand by more than 10%, a power increase command is sent to increase power output by increasing the speed, while the speed is always within a safe threshold range.

9. The method for predicting the power and regulating the speed of a tidal current turbine according to claim 1, characterized in that, The power flow data acquisition process also includes a data anomaly repair unit, which is used to repair missing or abnormal power flow data caused by sensor failure or signal interference. The unit uses an interpolation algorithm based on Kalman filtering. When missing or abnormal data is detected, the algorithm is called: first, an initial estimate is established by using historical data from the same period and data from adjacent time points, and then iteratively corrected by Kalman filtering. The error of the repaired data is ≤5%. At the same time, the occurrence time, type and repair method of the abnormal data are recorded, and a data quality report is generated.

10. The method for predicting the power and regulating the speed of a tidal current turbine according to claim 1, characterized in that, The execution drive step also includes a speed regulation buffer unit, which is used to mitigate the impact during speed regulation and protect the turbine transmission system. The unit has a built-in buffer spring and damper, which are installed in series at the connection between the hydraulic actuator and the guide vane. When the speed regulation command requires the guide vane opening to change rapidly, the buffer spring absorbs part of the impact force, and the damper slows down the movement speed. At the same time, the unit collects the compression of the buffer spring and the oil temperature of the damper in real time and sends a deceleration command to the speed regulation decision step.

Images