Flow field sensing and intelligent control system for tidal current energy water turbine array

Abstract

The invention discloses a tidal current energy water turbine array flow field sensing and intelligent control system, which relates to the technical field of ocean renewable energy sources and comprises a flow field sensing module, an array state monitoring module, an intelligent control module, a data storage and analysis module and an emergency response module. The flow field sensing module comprises a dynamic calibration sub-module and a short-term prediction sub-module, and equipment such as an acoustic Doppler current meter is adopted; the array state monitoring module comprises a fault grading diagnosis sub-module; the intelligent control module comprises a wake flow avoidance sub-module, a yaw self-adaption sub-module and a power optimization sub-module, and adjustment is achieved through combination of a PLC and a PID. The data storage and analysis module adopts a mode of combining local solid-state storage and cloud, and BP and LSTM models are built in the data storage and analysis module; the emergency response module can dispose six types of scenes in a grading manner. The flow field sensing error is effectively reduced, the array generating capacity standard-reaching rate is improved, the fault response time is shortened, the operation and maintenance cost is reduced, the grid-connected stability is enhanced, and the comprehensive energy efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of marine renewable energy technology, and in particular to a tidal current turbine array flow field sensing and intelligent control system. Background Technology

[0002] Tidal energy, as a clean and renewable energy source, boasts advantages such as high energy density and strong predictability, and its large-scale development is of great significance for optimizing the energy structure. Tidal turbine arrays, which achieve power superposition through the coordinated operation of multiple devices, are currently the mainstream mode for tidal energy development. However, ocean current fields are characterized by pulsation, significant turbulence effects, and uneven spatial distribution. Furthermore, the operation of turbine arrays involves complex wake interference, placing extremely high demands on the precision, response speed, and coordination capabilities of the control system. Existing technologies still face numerous bottlenecks.

[0003] Existing tidal power turbine control systems largely borrow from onshore wind power control logic, but they are not fully adapted to the characteristics of ocean flow fields and the operational requirements of the array, resulting in insufficient accuracy and timeliness in flow field sensing. Traditional sensing modules employ a fixed-period calibration mode, which cannot cope with dynamic fluctuations in tidal velocity and direction. Measurement deviations easily occur under pulsating and turbulent flow conditions, leading to inaccurate control parameters. Furthermore, there is a lack of short-term flow field prediction mechanisms for tidal power; control strategies are mostly based on passive adjustments using real-time data, making it difficult to proactively avoid power fluctuations caused by sudden changes in the flow field. This contrasts sharply with the velocity prediction and look-ahead control already implemented in wind power generation. In addition, the turbines within the array mostly operate in independent control modes, lacking a wake interference coordination mechanism. The wake generated by upstream equipment significantly reduces the energy capture efficiency of downstream equipment, hindering the overall array energy efficiency improvement.

[0004] The simplistic control strategies and inefficient fault handling further constrain system operational stability. Existing systems mostly employ simple single-parameter adjustments of rotational speed or angle of attack, lacking a graded adjustment mechanism based on power deviation. This makes them prone to drastic power fluctuations when flow velocity changes significantly, affecting grid connection stability. Fault diagnosis relies heavily on single sensor signals, lacking graded diagnosis and precise source tracing capabilities. Minor faults easily escalate into severe ones, and emergency response often involves direct shutdown without considering the adaptability of fault type to flow field conditions, resulting in energy waste. Furthermore, maintenance relies on manual on-site inspections, and the harsh marine environment leads to high maintenance costs and delayed response times. This represents a technological gap compared to the remote monitoring, intelligent diagnosis, and graded maintenance already implemented in wind power generation, failing to meet the needs of large-scale development of tidal energy arrays. Summary of the Invention

[0005] This invention proposes a tidal current turbine array flow field sensing and intelligent control system to solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a tidal current turbine array flow field sensing and intelligent control system, comprising;

[0007] The flow field sensing module consists of an underwater acoustic Doppler current meter, a three-dimensional flow direction sensor, a pressure sensor, and a data preprocessing unit.

[0008] The array status monitoring module equips each turbine with a speed sensor, blade angle of attack sensor, torque sensor, vibration acceleration sensor, and generator electrical parameter acquisition device;

[0009] The intelligent control module includes a main controller and a distributed turbine controller. The main controller uses an industrial-grade PLC, and the distributed controller is connected one-to-one with each turbine. Based on flow field perception data and status monitoring data, the blade angle of attack and turbine speed are adjusted through the maximum power point tracking algorithm, and PID control is used to achieve real-time correction of yaw angle.

[0010] The data storage and analysis module consists of a local storage server and a cloud database. The local server uses a solid-state drive array; the cloud database synchronizes data via a 4G / 5G module and has a built-in BP neural network model.

[0011] The emergency response module integrates an anomaly detection unit and a handling execution unit. The anomaly detection unit can set thresholds for parameters such as flow rate, vibration acceleration, and generator overvoltage.

[0012] Furthermore, it also includes a short-term flow field prediction submodule, which, based on real-time data from the flow field sensing module and a historical database, calculates the flow field prediction value according to V. t+Δt =V t ×(1+aΔt)+bV t-1 Calculate the future short-term flow velocity; where V t+Δt V represents the flow velocity at a future time Δt. t V represents the current flow velocity. t-1 Δt represents the flow velocity at the previous moment, a is the flow velocity attenuation coefficient, b is the historical data weighting coefficient, and Δt is the prediction time interval. The submodule has a built-in data cleaning unit. Flow velocity prediction can reserve adjustment time for the intelligent control module.

[0013] Furthermore, it also includes a power optimization and adjustment submodule, which is configured according to... Calculate the target power, where P t Let k be the overall efficiency coefficient of the turbine, ρ be the density of seawater, A be the swept area of ​​the turbine, and V be the target output power. t Given the current flow rate; the submodule constructs a power deviation graded adjustment mechanism.

[0014] Furthermore, the flow field sensing module also includes a dynamic calibration submodule. This submodule employs a three-level calibration mechanism: standard signal comparison, deviation compensation, and effect verification. It periodically calls preset standard flow velocity and pressure signals, calculates the deviation rate using δ = (measured value - standard value) / standard value, and generates a linear calibration coefficient k. c al=1 / (1+δ); The calibration cycle is adaptively adjusted according to the flow field stability: calibrate once every 24 hours when the flow velocity fluctuation is ≤0.3m / s; calibrate once every 12 hours when the fluctuation is 0.3-1m / s; calibrate once every 6 hours when the fluctuation is >1m / s; The calibration process adopts online compensation and offline verification mode, and cross-verification is performed through adjacent sensor data after calibration;

[0015] The flow field sensing module also includes a wave field reconstruction and power correction submodule:

[0016] 1. Wave field reconstruction: Based on pressure sensor data, using formulas... ΔP is the pressure fluctuation, ρ is the seawater density, and g is the gravitational acceleration. Calculate the wave height H and wave period T.

[0017] 2. Power Correction: Establish a wave-power correlation model—for example, when H>1.5m, power fluctuations increase by 15-20%, and the target power is dynamically reduced by 10% to avoid sudden changes;

[0018] 3. Collaborative control: Wave parameters are synchronized to the intelligent control module. When waves are violent, the adjustment step size is reduced by 50% and the interval is extended to 10 seconds. Actual results: Power fluctuation is reduced by 25-30% and efficiency is improved by 5-8% under wave conditions.

[0019] Furthermore, the array status monitoring module also includes a fault classification and diagnosis submodule. This submodule is based on an improved decision tree algorithm and combined with random forest-assisted verification to classify faults into three levels: minor, moderate, and severe, and refine the judgment logic.

[0020] Furthermore, the intelligent control module also includes a wake avoidance coordination submodule; firstly, it combines the turbine diameter D and the current flow velocity V... t Calculation of turbulence intensity I and wake influence radius R w =0.5D + 0.15V t / I, and then determine the wake diffusion direction based on the three-dimensional flow direction sensor data, and construct an array wake distribution heat map.

[0021] Furthermore, the data storage and analysis module also includes an offline modeling submodule. This submodule uses a working condition classification, hybrid modeling, and incremental update technique to build an optimized model. First, typical flow field working conditions are quantitatively classified according to the frequency, amplitude, and turbulence intensity of flow velocity fluctuations. For each type of working condition, an LSTM neural network and a BP neural network are integrated to build a correlation model between flow field working conditions, control parameters, and power generation efficiency. The model uses sliding window incremental learning: every 30 days, the latest data from the past 30 days is extracted, and only the weights of the output layer and hidden layer of the model are updated, without the need to retrain the entire dataset.

[0022] Furthermore, the intelligent control module also includes a yaw angle adaptive adjustment submodule, which adjusts the yaw angle according to θ = arccos(V x / V t Calculate the optimal yaw angle, where θ is the yaw angle and V x The velocity component along the turbine's rotation axis is represented; the submodule introduces a multi-factor correction mechanism: if the turbine's vibration acceleration > 2 m / s² 2 The optimal yaw angle is corrected by ±1° in the direction of vibration reduction; if it is in the wake influence zone of the preceding turbine, it is corrected by ±2° to ±3° in combination with the wake direction; yaw adjustment adopts speed grading and torque closed-loop control: when the deviation is 5°-10°, the yaw motor runs at a low speed of 0.5° / s; when the deviation is 10°-15°, it runs at a medium speed of 1° / s; when the deviation is >15°, it runs at a high speed of 1.5° / s; after the adjustment is in place, the yaw locking torque is detected by the torque sensor, and the locking torque must be ≥90% of the design value to ensure stability; at the same time, energy consumption optimization and error avoidance logic are introduced: when the single adjustment angle is <5°, the yaw adjustment action is not started temporarily; when the adjustment angle is <3° for 3 consecutive times, the adjustment interval is extended to 180s, and the flow field stability must be verified by the flow velocity sensor during the extension period;

[0023] The intelligent control module also includes a global power optimization submodule for the array. This submodule employs a dual-drive mechanism of numerical simulation pre-training and long-term operational data iteration to achieve the maximum power of the entire array (allowing that some turbines are not optimal on a single unit).

[0024] 1. Numerical simulation pre-training: Based on STAR-CCM+, a flow field-power coupling model is established to simulate the "impact of single turbine adjustment on the global flow" under different turbulence intensities. For example, reducing the speed of the preceding turbine by 5% can increase the power of the subsequent turbine by 12%, generating an initial optimization library.

[0025] 2. Long-term data iteration: Extract nearly 90 days of operating data (flow velocity, power of each turbine, wake coefficient) every 15 days, and update the strategy through the LSTM-BP hybrid model. For example, when the optimal power of a single turbine is 100kW, it is adjusted to 92kW to reduce wake interference, so that the power of the adjacent turbines is increased from 75kW to 88kW, and the overall array gain is 11kW.

[0026] 3. Real-time adjustment: The current total power is compared with the model's optimal value every 30 seconds. If the deviation is >2%, an instruction is issued according to the "wake effect priority". After adjustment, the wake change is verified by flow field perception. Actual measurement data: This submodule can increase the array's power generation by 12-15%, which is 8% higher than the single unit optimization.

[0027] Furthermore, the emergency response module also includes a tiered handling submodule, which pre-sets full-process solutions for detection, judgment, handling, and verification for six typical emergency scenarios: Scenario 1: First, adjust the blade angle of attack to the feathering position to reduce turbine resistance, while monitoring flow velocity changes; Scenario 2: First, reduce the speed by 20%, and simultaneously determine the fault type through vibration spectrum analysis. If mechanical resonance occurs, adjust the speed to avoid the resonance range; if the flow field is turbulent, adjust the yaw angle by ±5°; Scenario 3: First, reduce the output power by 30% and start the cooling system's underwater circulation pump in high-flow mode; Scenario 4: Switch to redundant sensor data and enable short-term prediction values ​​from the flow field prediction submodule to assist control; Scenario 5: Attempt to unlock by small-amplitude reciprocating adjustment of the angle of attack. If unlocking fails twice, shut down the turbine; Scenario 6: Immediately reduce the excitation current. If the voltage does not recover within 100ms, cut off the output; all handling processes push progress to the operation and maintenance platform in real time. After handling is completed, the equipment status is checked within 10 minutes. If the restart conditions are met, the turbine is gradually restarted;

[0028] The emergency response module also includes a local emergency control submodule, which is used to handle overall control failures (main controller communication interruption > 10 seconds):

[0029] Local data reliance: Based on real-time data from the turbine's built-in sensors (flow velocity, rotational speed, vibration), combined with a locally stored 30-day historical optimal parameter database (categorized by flow velocity range, such as 1-2 m / s corresponding to a rotational speed of 150-180 r / min);

[0030] Control logic: A simplified PID algorithm is adopted to prioritize safety (automatic feathering when the flow rate is >5m / s), and then output power at 80% of the optimal value for the same historical flow rate to avoid overload;

[0031] Recovery mechanism: After communication is restored, the main controller status is verified by synchronizing data for nearly 10 seconds, and then gradually switched back to global control within 30 seconds (power gradient adjustment to avoid impact); Reliability test: When the overall control fails, the power generation loss is reduced from 40% to less than 10%;

[0032] The tiered response submodule also includes Scenario 7 (Extreme Waves):

[0033] Triggering conditions: The wave field reconstruction and power correction submodule detects H>3m (can be adjusted according to sea area) and T<8s (short-period giant wave);

[0034] Handling procedure: 0-100ms: Adjust blade angle of attack to -15° (maximum feathering) to reduce the frontal area; 100-500ms: Cut off generator output to avoid electrical shock; 500ms-1s: Yaw to be parallel to the wave propagation direction to reduce impact force;

[0035] Restart conditions: H≤1.5m and after 5 minutes, restart at a power gradient of 20% / min; Model test: This process can reduce the stress on the turbine structure by 40-50% and avoid blade damage.

[0036] Furthermore, it includes a remote operation and maintenance and energy efficiency assessment sub-module. This sub-module is based on a cloud platform to build an integrated system of visual monitoring, intelligent assessment, and precise operation and maintenance. Remote visualization uses WebGL technology to generate a 3D interface, displaying the array layout, flow field heat map, turbine status dashboard, and fault location map in real time. Energy efficiency assessment generates multi-dimensional reports on a daily, weekly, and monthly basis: core indicators include total array power generation, power generation per unit flow velocity, equipment utilization rate, and wake influence coefficient; comparison indicators include deviation rates from historical best values ​​and design values; operation and maintenance suggestions are generated using rule reasoning and case matching: based on equipment runtime, fault frequency, and energy efficiency decline trend, combined with a historical operation and maintenance case library, the optimal solution is output; remote control adopts a dual authentication, command rehearsal, and feedback confirmation security mechanism: operation and maintenance personnel must be authenticated by fingerprint and password, and the effect is rehearsed in a virtual simulation environment before the command is issued. The command is executed only after the rehearsal is passed.

[0037] Compared with existing technologies, the beneficial effects of this invention are:

[0038] This invention addresses the core pain points of existing tidal power turbine array control systems—namely, inaccurate sensing, lagging control, insufficient coordination, and extensive operation and maintenance—through a full-chain innovation in flow field perception, intelligent control, and operation and maintenance systems, thereby achieving a comprehensive improvement in array operation performance.

[0039] The accuracy and timeliness of flow field sensing are significantly enhanced, laying the foundation for precise control. A dynamic calibration submodule is employed, adaptively adjusting the calibration cycle based on flow field stability. Through standard signal comparison and deviation compensation mechanisms, combined with cross-validation and interference filtering algorithms, sensor measurement errors are effectively reduced, ensuring that parameters such as flow velocity and direction maintain high accuracy over the long term. The short-term flow field prediction submodule constructs a predictive model by fusing real-time and historical data, enabling it to capture flow field change trends in advance. This provides the control module with adjustment time, avoiding power fluctuations caused by traditional passive control, and making the control response more forward-looking and adaptable to the dynamic characteristics of ocean currents.

[0040] The synergistic optimization of wake interference and power control significantly improves the overall energy efficiency of the array. The wake avoidance coordination submodule constructs a wake distribution heatmap of the array by calculating the wake influence range and diffusion direction, and adjusts the yaw angle and speed of subsequent turbines based on a priority strategy to effectively avoid the wake core area, solving the problem of array energy efficiency degradation in the traditional independent control mode. The power optimization and regulation submodule constructs a hierarchical regulation mechanism, dynamically corrects the target power by combining flow field parameters and equipment status, and links the speed and blade angle of attack adjustment. At the same time, it coordinates with the generator converter to ensure smooth output, so that the actual power generation efficiency approaches the theoretical optimal level, significantly improving the energy capture capability under different flow field conditions.

[0041] The refined fault handling and operation and maintenance management significantly reduces operational risks and costs. The fault classification and diagnosis submodule uses an improved decision tree algorithm combined with multi-parameter cross-validation to achieve accurate fault classification. Differentiated handling strategies are implemented for minor, moderate, and severe faults, avoiding losses caused by "one-size-fits-all" downtime. At the same time, fault tracing is achieved through spectrum analysis and parameter comparison, providing precise guidance for maintenance. The remote operation and maintenance and energy efficiency assessment submodule significantly shortens operation and maintenance response time and reduces reliance on on-site inspections through 3D visualization monitoring, multi-dimensional energy efficiency analysis, and intelligent operation and maintenance suggestion generation. Combined with a dual-authentication remote control mechanism, it balances operation and maintenance convenience with operational safety, effectively reducing the operation and maintenance costs throughout the array's life cycle.

[0042] The grid adaptability and operational stability are comprehensively improved. The emergency response module has preset graded handling procedures for typical scenarios such as extreme flow velocities, equipment vibration, and electrical anomalies. The handling process is coordinated with grid parameters to avoid voltage fluctuations caused by power surges and improve the array's grid-connection compatibility. The yaw angle adaptive adjustment submodule dynamically corrects the optimal yaw angle by combining flow field parameters and equipment vibration status. Through velocity grading and torque closed-loop control, it ensures that the equipment always captures energy in the optimal posture, while introducing energy consumption optimization logic to extend equipment life. Overall, this invention constructs an integrated "sensing-control-diagnosis-operation and maintenance" system through multi-module collaboration, effectively solving the core technical bottlenecks in the operation of tidal power arrays and providing strong support for their large-scale development. Attached Figure Description

[0043] Figure 1 This is a schematic block diagram of a tidal energy turbine array flow field sensing and intelligent control system proposed in this invention;

[0044] Figure 2 This is a comparison curve of power fluctuations under different control modes under power grid interference.

[0045] Figure 3 A bar chart comparing the velocity prediction errors configured for different upstream turbine units;

[0046] Figure 4A radar chart showing the energy efficiency indicators for multi-module collaborative control;

[0047] Figure 5 These are comparison curves showing the voltage recovery characteristics after grid interference. Detailed Implementation

[0048] 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.

[0049] 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," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not 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.

[0050] 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.

[0051] Reference Figures 1 to 5 A tidal current turbine array flow field sensing and intelligent control system includes a flow field sensing module, an array status monitoring module, an intelligent control module, a data storage and analysis module, and an emergency response module. Each module achieves data interaction and collaborative operation through industrial Ethernet communication.

[0052] The flow field sensing module consists of an underwater acoustic Doppler current meter, a three-dimensional flow direction sensor, a pressure sensor, and a data preprocessing unit. The acoustic Doppler current meter is an ADCP-S300 model, with a measurement range of 0.2-5 m / s, an accuracy of ±0.01 m / s, and a sampling frequency of 10 Hz. The three-dimensional flow direction sensor has a measurement range of 0-360° and an accuracy of ±1°. The pressure sensor monitors water pressure changes within a water depth range of 0-100 m. The data preprocessing unit uses a Kalman filter algorithm to eliminate noise and standardizes the flow velocity, flow direction, water depth, and water pressure data before transmitting them to subsequent modules.

[0053] The array-based condition monitoring module equips each turbine with a speed sensor, blade angle-of-attack sensor, torque sensor, vibration acceleration sensor, and generator electrical parameter acquisition unit. The speed sensor has a measurement range of 0-300 r / min and an accuracy of ±1 r / min; the blade angle-of-attack sensor has an adjustment range of -15° to +15° and a detection accuracy of ±0.5°; the vibration acceleration sensor has a monitoring frequency of 0-500 Hz and a measurement range of ±10 m / s². 2 The electrical parameter acquisition device collects generator voltage, current, and power factor in real time at a sampling frequency of 50Hz, and simultaneously uploads the operating status data of each turbine.

[0054] The intelligent control module includes a main controller and distributed turbine controllers. The main controller uses an industrial-grade PLC, and the distributed controllers are connected one-to-one with each turbine. Based on flow field perception data and status monitoring data, the blade angle of attack and turbine speed are adjusted through a maximum power point tracking algorithm. PID control is used to achieve real-time yaw angle correction, ensuring that the swept surface is aligned with the main flow direction. At the same time, it coordinates the operating parameters of the turbines in the array to avoid wake interference.

[0055] The data storage and analysis module consists of a local storage server and a cloud database. The local server uses a solid-state drive array with a storage capacity of 1TB, retaining nearly 90 days of raw data. The cloud database synchronizes data via a 4G / 5G module and has a built-in BP neural network model that predicts the flow rate change trend for the next 1-6 hours based on historical flow field data and operating parameters, and outputs optimal control parameter suggestions.

[0056] The emergency response module integrates an anomaly detection unit and a response execution unit. The anomaly detection unit is set to a flow velocity > 5 m / s and a vibration acceleration > 5 m / s². 2 When thresholds such as generator overvoltage are reached, the handling execution unit can trigger actions such as blade feathering, emergency shutdown, and yaw to avoid current, while simultaneously pushing alarm information to the operation and maintenance terminal via SMS and platform.

[0057] This invention also includes a short-term flow field prediction submodule, which, based on real-time data from the flow field sensing module and a historical database, calculates the flow field prediction value according to V. t+Δt =V t×(1+aΔt)+bV t-1 Calculate the future short-term flow velocity. Where V... t+Δt V represents the flow velocity at a future time Δt. t V represents the current flow velocity. t-1 denoted as , where 'a' is the velocity at the previous moment, and 'a' is the velocity decay coefficient (valued at 0.002-0.003 s during high tide). -1 The value during the ebb tide phase is 0.004-0.005s. -1 ), where b is the historical data weighting coefficient (0.2-0.3 for steady flow and 0.3-0.4 for fluctuating flow), and Δt is the prediction time interval (60s for normal operating conditions and 10s for turbulent operating conditions). The submodule has a built-in data cleaning unit that first removes outliers with velocity mutations exceeding 2m / s, and then smooths the data using a sliding window averaging method. After prediction, the measured value is compared with the predicted value every 5s, and the coefficients a and b are dynamically corrected when the deviation exceeds 0.3m / s. Velocity prediction allows for adjustment time for the intelligent control module, avoiding power fluctuations caused by sudden flow field changes, improving the foresight of the control response, and reducing power fluctuation amplitude by more than 40% under fluctuating flow conditions.

[0058] This invention also includes a power optimization and adjustment submodule, which is configured according to... Calculate the target power, where P t Let k be the overall efficiency coefficient of the turbine (0.42-0.45 when the blade angle of attack is 0°, and 0.35-0.38 when the angle of attack is ±10°), and ρ be the density of seawater (dynamically compensated for by water depth, with ρ increasing by 0.1 kg / m³ for every 10 m increase in water depth). 3 A is the swept area of ​​the turbine (corrected according to the real-time blade speed; when the speed exceeds 200 r / min, A is corrected by 1.02 times), V t The current flow velocity is used. The submodule constructs a power deviation graded adjustment mechanism: when the deviation is <5%, the speed is finely adjusted in steps of 0.5 r / min; when the deviation is 5%-15%, the speed is adjusted in steps of 1 r / min, and the blade angle of attack is adjusted in conjunction (adjustment range ±1°); when the deviation is >15%, a rapid adjustment mode is activated, with a step size of 2 r / min and an angle of attack adjustment range of ±2°. During the adjustment process, it works in conjunction with the generator converter to ensure smooth changes in output power and avoid voltage fluctuations. In the flow velocity range of 0.5-3 m / s, the actual power generation efficiency can approach more than 95% of the theoretical optimal value.

[0059] In this invention, the flow field sensing module also includes a dynamic calibration submodule, which employs a three-level calibration mechanism of "standard signal comparison - deviation compensation - effect verification". Preset standard flow velocity signals (three standard values: 0.5 m / s, 2 m / s, and 4 m / s) and pressure signals (pressure values ​​corresponding to water depths of 10 m, 50 m, and 100 m) are periodically invoked and simulated input to the sensor via a high-precision signal generator. The deviation between the measured value and the standard value is compared, and the deviation rate is calculated using δ = (measured value - standard value) / standard value, generating a linear calibration coefficient k. c al = 1 / (1+δ). The calibration cycle is adaptively adjusted according to the flow field stability: calibrate every 24 hours when the flow velocity fluctuation is ≤0.3m / s; calibrate every 12 hours when the fluctuation is 0.3-1m / s; calibrate every 6 hours when the fluctuation is >1m / s. The calibration process adopts the "online compensation + offline verification" mode, without interrupting the sensing operation. After calibration, cross-verification is performed using data from adjacent sensors to ensure that the error rate is ≤0.5%. To address the issue of acoustic Doppler current meters being susceptible to bubble interference, a bubble filtering algorithm is automatically enabled after calibration to further reduce measurement errors, ensuring that the flow velocity measurement accuracy remains within ±0.01m / s over a long period. The flow field sensing module also includes a wave field reconstruction and power correction submodule.

[0060] 1. Wave field reconstruction: Based on pressure sensor data (sampling frequency 10Hz), wave height H and wave period T are calculated using the formula (where is the pressure fluctuation, is the seawater density, and is the gravitational acceleration).

[0061] 2. Power Correction: Establish a wave-power correlation model—for example, when H>1.5m, power fluctuations increase by 15-20%, and the target power is dynamically reduced by 10% to avoid sudden changes;

[0062] 3. Collaborative Control: Wave parameters are synchronized to the intelligent control module. When waves are violent, the adjustment step size is reduced by 50%, and the interval is extended to 10 seconds. Actual results: Power fluctuation is reduced by 25-30% and efficiency is improved by 5-8% under wave conditions.

[0063] In this invention, the array status monitoring module also includes a fault classification and diagnosis submodule. This submodule, based on an improved decision tree algorithm and combined with random forest-assisted verification, classifies faults into three levels: minor, moderate, and severe, and refines the judgment logic. Minor faults include: static deviation of blade angle of attack of 1-2° or dynamic fluctuation of ±1°, and vibration acceleration of 2-5 m / s². 2(Frequency < 50Hz), generator power factor fluctuation ±0.02, trigger parameter fine-tuning (angle of attack correction ±0.5°, speed fine-tuning ±1r / min), diagnostic response time ≤100ms; Moderate fault: speed fluctuation ±10-20r / min, abnormal torque change 10%-20%, generator voltage fluctuation ±5%, activate backup control logic (switch to redundant sensor data, enable preset power curve), diagnostic response time ≤50ms, synchronously record vibration spectrum and current waveform during the fault period; Severe fault: sensor signal loss If the failure duration is >1s, generator overcurrent (current exceeds rated value by 1.2 times), or blade jamming (no response to angle of attack adjustment), immediately execute an emergency shutdown (feather blades to -15°, cut off generator output, lock yaw mechanism). The diagnostic response time is ≤30ms. At the same time, initiate fault tracing: differentiate mechanical faults (spectral peaks concentrated in 100-200Hz) from flow field interference (no obvious peaks in the spectrum) through vibration spectrum analysis, locate circuit faults through electrical parameter comparison, and the fault information includes occurrence time, parameter curves, and tracing results. The storage period is ≥1 year.

[0064] In this invention, the intelligent control module also includes a wake avoidance coordination submodule, which achieves precise wake avoidance based on a "flow velocity attenuation model + array topology analysis". First, it combines the turbine diameter D and the current flow velocity V... t Calculation of turbulence intensity I and wake influence radius R w =0.5D + 0.15V t / I (I value ranges from 0.05 to 0.2), and then the wake diffusion direction is determined based on the three-dimensional flow direction sensor data to construct an array wake distribution heat map. When the downstream turbine is in the wake region and the flow velocity decreases by >0.5 m / s, a graded coordination strategy is initiated: Priority 1 (yaw adjustment), the yaw angle of the downstream turbine is adjusted by ±5° to ±15° according to the wake offset angle to ensure that the swept surface avoids the wake core region; Priority 2 (speed adjustment), if the yaw angle cannot be adjusted due to mechanical limitations, the ratio k of the wake velocity to the mainstream velocity is used. v =V_{wake} / V_{mainstream}\Adjusting speed, speed correction value = rated speed × k v This ensures that the power output of the subsequent turbines remains above 85% of the normal level; Priority 3 (array coordination): if the adjustment effect of a single turbine is insufficient, coordinate with the preceding turbines to fine-tune their speed (reducing it by 5%-10%) to weaken the wake intensity. The wake data is updated every 20 seconds during the coordination process, and the adjustment is verified through the flow field sensing module to ensure that the flow velocity recovery rate of the subsequent turbines is ≥80%, and the overall power generation efficiency of the array is improved by 10%-15%.

[0065] In this invention, the data storage and analysis module also includes an offline modeling submodule, which uses a "condition classification-hybrid modeling-incremental update" technical approach to construct an optimization model. First, typical flow field conditions are quantitatively classified according to flow velocity fluctuation frequency (f≤0.1Hz for steady flow, 0.1-1Hz for pulsating flow, f>1Hz for turbulent flow), amplitude (A≤0.5m / s for low amplitude, 0.5-1m / s for medium amplitude, A>1m / s for high amplitude), and turbulence intensity (I≤0.1 for weak turbulence, I>0.1 for strong turbulence), resulting in a total of 9 basic conditions. For each condition, an "flow field condition-control parameter-power generation efficiency" correlation model is constructed by integrating an LSTM neural network (capturing temporal features) and a BP neural network (fitting static mapping). The input parameters include flow velocity, flow direction, water depth, and turbulence intensity, while the output parameters are the optimal blade angle of attack, rotational speed, and yaw angle. The model employs sliding window incremental learning: every 30 days, it extracts the latest data from the past 30 days (including newly added operating condition samples) and updates only the weights of the output and hidden layers, eliminating the need to retrain the entire dataset. Update time is ≤2 hours. Offline model validation uses a 5-fold cross-validation method, requiring a validation accuracy of ≥95% before deployment. After deployment, the model's predicted parameters are compared with the actual optimal parameters every 7 days. If the deviation exceeds 3%, a local weight correction is triggered, providing accurate data support for optimizing control strategies under different flow field conditions. This improves the array's power generation efficiency in complex flow fields by more than 20% compared to fixed parameter control.

[0066] In this invention, the intelligent control module further includes a yaw angle adaptive adjustment submodule, which adjusts the yaw angle according to θ = arccos(V x / V t Calculate the optimal yaw angle, where θ is the yaw angle and V x The velocity component along the turbine's rotation axis is calculated by fusing data from three measuring points of a three-dimensional flow direction sensor, with a spacing of 0.5m between the measuring points, and the average value is taken as V. x V t The flow velocity in the current field is measured by an acoustic Doppler current meter. The submodule introduces a multi-factor correction mechanism: if the turbine vibration acceleration > 2 m / s²... 2The optimal yaw angle is corrected by ±1° in the direction of vibration reduction; if it is in the wake influence zone of the preceding turbine, it is corrected by ±2°-±3° in combination with the wake direction. Yaw adjustment employs speed gradation and torque closed-loop control: when the deviation is 5°-10°, the yaw motor operates at a low speed of 0.5° / s; when the deviation is 10°-15°, it operates at a medium speed of 1° / s; when the deviation is >15°, it operates at a high speed of 1.5° / s. After adjustment, the yaw locking torque is detected by a torque sensor, and the locking torque must be ≥90% of the design value to ensure stability. Simultaneously, energy consumption optimization and error avoidance logic are introduced: when a single adjustment angle is <5°, yaw adjustment is not initiated (referencing actual measurement data: this angle deviation has a <1% impact on power generation efficiency, but adjustment increases motor losses by 15%, mostly due to sensor vibration errors); when three consecutive adjustment angles are <3°, the adjustment interval is extended to 180s, and during this extended period, the flow field stability must be verified by a flow velocity sensor (if the flow velocity fluctuation is <0.3m / s, no adjustment is needed). This ensures the turbine captures tidal energy at its optimal attitude while reducing the energy consumption of the yaw system and extending the equipment's service life.

[0067] The intelligent control module also includes a global power optimization submodule for the array. This submodule employs a dual-drive mechanism of numerical simulation pre-training and long-term operational data iteration to achieve the maximum power of the entire array (allowing that some turbines are not optimal on a single unit).

[0068] 1. Numerical simulation pre-training: Based on STAR-CCM+, a flow field-power coupling model is established to simulate the "impact of single turbine adjustment on the global flow" under different turbulence intensities. For example, reducing the speed of the preceding turbine by 5% can increase the power of the subsequent turbine by 12%, generating an initial optimization library.

[0069] 2. Long-term data iteration: Extract nearly 90 days of operating data (flow velocity, power of each turbine, wake coefficient) every 15 days, and update the strategy through the LSTM-BP hybrid model. For example, when the optimal power of a single turbine is 100kW, it is adjusted to 92kW to reduce wake interference, so that the power of the adjacent turbines is increased from 75kW to 88kW, and the overall array gain is 11kW.

[0070] 3. Real-time adjustment: The current total power is compared with the model's optimal value every 30 seconds. If the deviation is >2%, an instruction is issued according to the "wake effect priority". After adjustment, the wake change is verified by flow field perception. Actual measurement data: This submodule can increase the array's power generation by 12-15%, which is 8% higher than the single unit optimization.

[0071] In this invention, the emergency response module also includes a tiered handling submodule, which pre-sets a full-process solution of "detection-judgment-handling-verification" for six typical emergency scenarios. Scenario 1 (extreme flow velocity > 5 m / s): First, adjust the blade angle of attack to the feathering position (-15°) to reduce turbine resistance, while monitoring flow velocity changes. If the flow velocity does not decrease within 10 seconds, shut down the turbine. After shutdown, yaw to a direction perpendicular to the flow direction to avoid the current. Scenario 2 (vibration acceleration > 5 m / s²) 2 Scenario 1 (Generator overheating > 85℃): First, reduce the speed by 20%, and simultaneously determine the fault type through vibration spectrum analysis. If it is mechanical resonance (fixed peak value in the spectrum), adjust the speed to avoid the resonance range (±10r / min). If it is turbulent flow (no fixed peak value in the spectrum), adjust the yaw angle ±5°. If the vibration does not subside within 30s, shut down the generator. Scenario 2 (Generator overheating > 85℃): First, reduce the output power by 30%, and start the cooling system's underwater circulation pump in strong flow mode (forced seawater circulation to remove the generator's heat). Scenario 3 (Sensor communication interruption): Switch to redundant sensor data, and enable the short-term prediction value of the flow field prediction submodule to assist control. If the interruption lasts for > 5min, shut down the generator. Scenario 4 (Blade jamming): Attempt to unlock the blade by adjusting the angle of attack (±2°) with a small amplitude. If unlocking fails twice, shut down the generator to avoid damage to the blades. Scenario 5 (Generator overvoltage > 1.1 times the rated value): Immediately reduce the excitation current. If the voltage does not recover within 100ms, cut off the output. All treatment processes are pushed to the operation and maintenance platform in real time (in detection - in treatment - treatment completed - effect verification). Within 10 minutes after treatment is completed, the equipment status is checked and it meets the restart conditions (flow rate ≤ 4m / s, vibration ≤ 2m / s). 2 If the electrical parameters are normal, restart gradually, increasing the power at a rate of 20% / min during restart to avoid impacting the power grid.

[0072] The emergency response module also includes a local emergency control submodule, which is used to handle overall control failures (main controller communication interruption > 10 seconds):

[0073] Local data reliance: Based on real-time data from the turbine's built-in sensors (flow velocity, rotational speed, vibration), combined with a locally stored 30-day historical optimal parameter database (categorized by flow velocity range, such as 1-2 m / s corresponding to a rotational speed of 150-180 r / min);

[0074] Control logic: A simplified PID algorithm is adopted to prioritize safety (automatic feathering when the flow rate is >5m / s), and then output power at 80% of the optimal value for the same historical flow rate to avoid overload;

[0075] Recovery mechanism: After communication is restored, the main controller status is verified by synchronizing data for nearly 10 seconds, and then gradually switched back to global control within 30 seconds (power gradient adjustment to avoid impact); Reliability test: When the overall control fails, the power generation loss is reduced from 40% to less than 10%;

[0076] The tiered response submodule also includes Scenario 7 (Extreme Waves):

[0077] Triggering conditions: The wave field reconstruction and power correction submodule detects H>3m (can be adjusted according to sea area) and T<8s (short-period giant wave);

[0078] Handling procedure: 0-100ms: Adjust blade angle of attack to -15° (maximum feathering) to reduce the frontal area; 100-500ms: Cut off generator output to avoid electrical shock; 500ms-1s: Yaw to be parallel to the wave propagation direction to reduce impact force;

[0079] Restart conditions: H≤1.5m and after 5 minutes, restart at a power gradient of 20% / min; Model test: This process can reduce the stress on the turbine structure by 40-50% and avoid blade damage.

[0080] This invention also includes a remote operation and maintenance and energy efficiency assessment submodule, which builds an integrated system of "visualized monitoring - intelligent assessment - precise operation and maintenance" based on a cloud platform. Remote visualization uses WebGL technology to generate a 3D interface, displaying in real time the array layout, flow field heat map (distinguished by flow velocity color: blue ≤1m / s, green 1-3m / s, red >3m / s), turbine status dashboard (including 12 core parameters such as speed, angle of attack, and power), and fault location map (marking the location and type of faulty equipment). It supports timeline backtracking (data from the last 7 days) and parameter curve comparison (comparison of single / multiple devices). Energy efficiency assessment generates multi-dimensional reports daily, weekly, and monthly: core indicators include total array power generation, power generation per unit flow velocity (total power generation / Σ(flow velocity × operating time)), equipment utilization rate (actual operating time / available time), and wake impact coefficient (power generation of wake-affected equipment / power generation of normal equipment); comparison indicators include the deviation rate from historical best values ​​and design values, automatically marking abnormal items when the deviation exceeds 10%. The maintenance recommendations utilize rule-based reasoning and case matching to generate optimal solutions: based on equipment runtime (>180 days prompting lubrication), failure frequency (more than 2 monthly failures prompting component inspection), and energy efficiency degradation trend (a continuous 5% power decrease prompting blade cleaning), combined with a historical maintenance case library, the optimal solution is output (including operation steps, required tools, and estimated man-hours). Remote control employs a "dual authentication - command rehearsal - feedback confirmation" security mechanism: maintenance personnel must authenticate via fingerprint and password; commands are rehearsed in a virtual simulation environment before being issued; execution is only performed after successful rehearsal; and equipment status feedback is received within 10 seconds of execution to ensure accurate command execution. This submodule can reduce maintenance response time by 60%, equipment failure rate by 30%, and array annual maintenance costs by more than 25%.

[0081] The following two examples further illustrate the specific implementation of this system:

[0082] Example 1: Nearshore wind-powered complementary tidal current turbine array control system (application scenario in Zhoushan, Zhejiang)

[0083] This embodiment focuses on a wind and tidal energy complementary power generation project in the nearshore waters of Zhoushan, Zhejiang Province. The water depth in this area is 20-30m, the tidal current velocity is 0.5-3m / s, and there is a significant pulsating current (fluctuation frequency 0.2-0.8Hz) during high and low tides. The array consists of six 500kW horizontal axis tidal turbines (spaced 50m apart, arranged in a straight line). Traditional control systems suffer from lagging flow field perception and coarse power regulation, resulting in the array's annual power generation reaching only 65% ​​of the design value. The present invention achieves precise control, and the specific implementation is as follows.

[0084] 1. Selection and overall configuration of core system components

[0085] The flow field sensing module uses an ADCP-S300 acoustic Doppler current meter (measurement range 0.2-5 m / s, accuracy ±0.01 m / s, sampling frequency 10 Hz), equipped with a three-dimensional flow direction sensor (0-360°, ±1° accuracy) and a deep-water pressure sensor (0-100 m, ±0.1 MPa). The data preprocessing unit uses an STM32F407 microcontroller with a built-in Kalman filter algorithm. The array status monitoring module equips each turbine with a Hall effect speed sensor (0-300 r / min, ±1 r / min), a magnetostrictive angle-of-attack sensor (-15° to +15°, ±0.5°), and a piezoelectric vibration acceleration sensor (0-500 Hz, ±10 m / s²). 2 ) and electrical parameter acquisition device (voltage 0-1000V, current 0-1000A, sampling frequency 50Hz).

[0086] The intelligent control module's main controller uses a Siemens S7-1500 PLC, while the distributed controller is a Schneider M258 PLC for each turbine, communicating via Profinet industrial Ethernet. The data storage and analysis module's local server uses a four-panel 256GB solid-state drive array (RAID5 redundancy), and the cloud database is based on an Alibaba Cloud ECS server, incorporating a built-in Python-based BP neural network prediction model. The emergency response module integrates an Omron E5CC temperature controller and a customized relay module, linking the blade feathering actuator and yaw motor.

[0087] 2. Core Module Operation Mechanism and Technical Details

[0088] 2.1 Flow Field Sensing and Short-Term Prediction

[0089] The short-term flow field prediction submodule is based on formula V t+Δt =V t ×(1+aΔt)+bV t-1Calculations show that during the high tide phase, a = 0.002 s. -1 b = 0.2, Δt = 60s; ebb tide phase a = 0.004s -1 b = 0.3, Δt = 40s. The data cleaning unit removes outliers with sudden velocity changes exceeding 2 m / s (such as instantaneous fluctuations caused by the sensor being entangled in fishing nets), and smooths the data using a 5-point sliding window averaging method. After prediction, the measured value is compared with the predicted value every 5 seconds. When the deviation exceeds 0.3 m / s, the coefficient 'a' is adjusted by 0.0001s. -1 Gradient correction, the b coefficient is adjusted by a gradient of 0.01. For example, the current flow velocity V t = 2m / s, V at the previous moment t-1 =1.8m / s, V at high tide t+60 =2×(1+0.002×60)+0.2×1.8=2.24+0.36=2.6m / s, the measured value is 2.5m / s, the deviation is 0.1m / s, no correction is needed.

[0090] The dynamic calibration submodule is calibrated every 12 hours (current velocity fluctuation in this sea area is 0.5 m / s). A signal generator simulates standard current velocity signals of 0.5 m / s, 2 m / s, and 4 m / s, with measured values ​​of 0.502 m / s, 2.005 m / s, and 4.010 m / s, respectively. The deviation rates δ are 0.4%, 0.25%, and 0.25%, and the calibration coefficients k_cal are 0.996, 0.9975, and 0.9975, respectively. After calibration, a bubble filtering algorithm is enabled to filter out bubble interference caused by wave breaking, and the current velocity measurement error is stabilized within ±0.008 m / s.

[0091] 2.2 Power Optimization and Yaw Adjustment

[0092] The power optimization and regulation submodule is based on the formula. Calculations show that k = 0.43 when the blade angle of attack is 0°, and ρ = 1027.5 kg / m when the water depth is 25 m. 3 When the turbine diameter is 5m (A=19.635m2) and the flow velocity is 2m / s, P t =0.43×1027.5×19.635×8≈698kW (considering losses, the target power is set at 650kW). When the measured power is 617.5kW (deviation 5%), the speed is adjusted from 180r / min to 185r / min in 0.5r / min increments, and the power increases to 648kW; when the flow velocity suddenly drops to 1.5m / s, the power deviation is 15%, the speed is adjusted to 160r / min and the angle of attack is adjusted from 0° to +1°, and the power stabilizes at 487kW (target 490kW). During the adjustment process, in coordination with the ABB converter, the output voltage fluctuation is controlled within ±2%.

[0093] The yaw angle adaptive adjustment submodule is configured according to θ = arccos(Vx / V t ) Calculations show that V is measured at three measuring points (0.5m apart) of a three-dimensional flow direction sensor. x The speeds were 1.98 m / s, 2.00 m / s, and 2.02 m / s respectively, with an average of 2.00 m / s. V t = 2.05 m / s², θ = arccos(2.00 / 2.05) ≈ 14°. Because the vibration acceleration is 1.5 m / s². 2 (<2m / s) 2 No correction is needed. The yaw motor adjusts at a speed of 1° / s. After reaching the position, the torque sensor detects a locking torque of 1200 N·m (design value 1300 N·m, ≥90%), confirming that the locking is effective.

[0094] 2.3 Emergency Response and Data Storage

[0095] The emergency response tiered handling submodule detected a sudden increase in flow velocity to 5.2 m / s (extreme flow velocity scenario), immediately adjusted the blade angle of attack to -15° (feathering), and after 10 seconds the flow velocity dropped to 4.8 m / s, requiring no shutdown; when the vibration acceleration of turbine #1 reached 5.5 m / s², the submodule was activated. 2 Reducing the rotational speed by 20% (from 180 r / min to 144 r / min) revealed that the peak frequency was concentrated at 150 Hz (mechanical resonance). Adjusting the rotational speed to 150 r / min avoided resonance, and the vibration decreased to 1.8 m / s². 2 .

[0096] The data storage and analysis module stores raw data on the local server every 10 seconds, while the cloud database is synchronized every minute via a 4G module. The offline modeling submodule updates the "flow field condition-control parameters-power generation efficiency" model every 30 days, using incremental learning to update only the weights corresponding to the data of the last 30 days, with an update time of 1.5 hours and a model validation accuracy of 96%.

[0097] 3. Performance data representation

[0098] Table 1: Performance Comparison between Traditional Control Systems and the System of This Invention

[0099]

[0100] Explanation: The data in Table 1 comes from 180 days of continuous operation statistics. Traditional systems, due to fixed-cycle calibration (24 hours), have flow velocity measurement errors of ±0.05 m / s, power fluctuations of 25% under pulsating flow conditions, and array power generation only reaching 65% of the design value. Fault diagnosis relies on a single sensor with a response time of 500 ms, minor faults are easily escalated, and annual maintenance costs account for 20%. This invention reduces the error to ±0.008 m / s through dynamic calibration and bubble filtering, and reduces the fluctuation amplitude to 8% through flow field prediction and graded power adjustment. Fault graded diagnosis has a fast response, maintenance is achieved through remote monitoring, the cost ratio is reduced to 8%, and the power generation compliance rate is increased by 27 percentage points, perfectly adapting to the pulsating flow characteristics of nearshore and nearshore tidal energy complementary scenarios.

[0101] Example 2: Tidal Energy Array Control System for Complex Flow Fields Around Islands (Application Scenario in Pingtan Waters, Fujian)

[0102] This embodiment addresses a tidal power generation project around Pingtan Island in Fujian Province. The water depth in this area is 15-25m, with strong turbulence (turbulence intensity 0.15-0.2). The array consists of eight 300kW turbines (arranged in a diamond shape with a spacing of 15m). Traditional systems suffer from severe wake interference and crude fault handling, resulting in a 30% lower power generation efficiency for subsequent turbines compared to preceding ones. The present invention employs a scheme to achieve coordinated control, and the specific implementation is as follows.

[0103] 1. Selection and overall configuration of core system components

[0104] The flow field sensing module deploys six ADCP-S300 units around the perimeter and center of the array. The three-dimensional flow direction sensor is the Saimo Electric SMS-LX type (measurement frequency 15Hz), and the pressure sensor is the Kunlun Coast JYB-KO-HAG type. The array status monitoring module adds a vibration spectrum analyzer (RIONSA-77), and the electrical parameter acquisition unit integrates a power factor transmitter (accuracy ±0.01).

[0105] The intelligent control module uses a Rockwell CompactLogix L24ER as its main controller, with distributed controllers equipped with AB1756 series I / O modules. The wake avoidance coordination submodule features a GPU accelerator card for heatmap generation. The data storage and analysis module has a 2TB local server storage capacity, and its cloud database is connected to the Huawei Cloud IoT platform. The remote operation and maintenance submodule uses a WebGL 3D visualization engine. The emergency response module adds a generator overcurrent protector (Schneider GV2-ME series) and a blade unlocking actuator.

[0106] 2. Core Module Operation Mechanism and Technical Details

[0107] 2.1 Wake avoidance coordination and flow field calibration

[0108] Wake avoidance coordination submodule (press R)w =0.5D + 0.15V t Calculations show that the turbine diameter is 4m (D=4), the flow velocity is 2.5m / s, the turbulence intensity is I=0.15, and R... w =0.5×4+0.15×2.5 / 0.15=2+2.5=4.5m. The three-dimensional flow direction sensor measured a wake diffusion angle of 30°. The constructed heat map showed that turbine #3 was located in the wake region of turbine #1 (flow velocity reduced to 1.8m / s). Priority 1 adjustment was initiated: the yaw angle of turbine #3 was adjusted from 0° to +12° to avoid the wake core region, and the flow velocity recovered to 2.3m / s. When the yaw of turbine #5 was restricted and could not be adjusted (priority 1 failed), press k. v =1.9 / 2.5=0.76 Adjust the speed from 160r / min to 122r / min, and maintain the power at 220kW (originally 180kW).

[0109] The dynamic calibration submodule, due to the 1.2 m / s current fluctuation in this sea area, is calibrated every 6 hours, simulating a 10 m water depth pressure signal (98 kPa). The measured value is 98.5 kPa, with a deviation rate of 0.51%. The calibration coefficient k... c al = 0.9949, and the cross-validation deviation between adjacent ADCP data is 0.006 m / s, which meets the error requirements. To address the instantaneous velocity fluctuations caused by turbulence, a filtering algorithm with a sliding window width of 10 was enabled after calibration to further smooth the data.

[0110] 2.2 Fault Classification Diagnosis and Yaw Adjustment

[0111] The fault classification and diagnosis submodule detected a static deviation of 1.8° in the angle of attack of the blades of turbine #4 (minor fault), and immediately corrected the angle of attack to -0.3° (from 1.5°), reducing the vibration acceleration from 2.2 m / s². 2 Reduced to 1.5 m / s 2 ; The speed of turbine #7 fluctuated by ±18 r / min (moderate fault). Switch to redundant speed sensor data and activate the preset power curve (flow velocity 2 m / s corresponds to speed 150 r / min); The angle of attack sensor signal of turbine #2 was lost (severe fault). Emergency shutdown was triggered within 30ms: the blades feathered to -15°, the generator output was cut off, the yaw mechanism was locked, and the spectrum analysis showed no mechanical fault peak. It was determined that the sensor line was broken, and the maintenance command was pushed to the operation and maintenance terminal.

[0112] The adaptive yaw angle adjustment submodule calculated θ = 18°. Since turbine #7 is located in the wake region of turbine #6, the yaw angle was corrected to 21°, with a deviation of 3°-5°. The yaw motor operated at 0.5° / s, and the locking torque detected after reaching the target position was 1050 N·m (design value 1100 N·m, ≥90%). After three consecutive adjustments of 2°, the adjustment interval was extended to 120s, and the yaw motor power decreased from 1.5kW to 1.05kW (energy saving 30%).

[0113] 2.3 Remote Operation and Maintenance and Emergency Response

[0114] The remote operation and maintenance submodule's 3D interface displays the array layout and flow field heat map in real time (blue ≤ 1.5 m / s, green 1.5-2.5 m / s, red > 2.5 m / s). The power curve of turbine #7 shows an abnormal drop, and the energy efficiency assessment report shows that the power generation per unit flow velocity is 12% lower than the historical best value. A blade cleaning suggestion (including operating steps and an estimated 2-hour working time) is pushed out. When remotely adjusting parameters, operation and maintenance personnel are authenticated via fingerprint and password. Commands are executed only after a successful virtual simulation, and equipment status feedback is received within 10 seconds.

[0115] The emergency response tiered handling submodule detected that the generator was overheated to 88℃, reduced the output power by 30% (from 300kW to 210kW), and started the cooling system's underwater circulation pump in high-flow mode. After 45 seconds, the temperature dropped to 78℃. When blade jamming occurred (no response to angle of attack adjustment), it attempted to adjust back and forth by ±2° twice, which successfully unlocked the blades and restored the angle of attack to normal, thus avoiding downtime losses.

[0116] 3. Performance data representation

[0117] Table 2: Comparison of wake and fault handling performance between traditional control systems and the system of this invention

[0118] Performance indicators Traditional control system This invention system Subsequent turbine flow velocity recovery rate 40％ 85％ Array wake influence coefficient 0.7 0.92 Mean time to resolve faults 45min 8min Number of unplanned downtimes 12 times / year 2 times / year Remote maintenance coverage 0％ 100％

[0119] Explanation: The data in Table 2 comes from 200 days of operational testing. Traditional systems lack wake avoidance mechanisms, resulting in a turbine flow velocity recovery rate of only 40% and a wake impact coefficient of 0.7 (power generation is only 70% of that without wake). Faults rely on on-site troubleshooting, with a response time of 45 minutes, 12 unplanned outages per year, and no remote maintenance capabilities. This invention, through wake thermal mapping and priority adjustment, increases the recovery rate to 85% and the impact coefficient to 0.92. Fault classification diagnosis and remote guidance reduce response time to 8 minutes and unplanned outages to 2. Remote maintenance enables full-process visualization, significantly reducing on-site inspection needs and perfectly solving wake interference and maintenance challenges in complex flow fields, resulting in a 28% increase in overall array power generation efficiency.

[0120] Reference Figure 2This diagram visually demonstrates the advantages of this invention in handling power grid interference. In traditional methods, power fluctuates drastically after the interference disappears (within 2 seconds), with deviations reaching ±30%, easily impacting the power grid. This invention maintains power stability during the interference period (0-2 seconds) through a control phase, and initiates a stabilization phase after the interference disappears, with power gradually returning to its rated value, reaching full stability after 7 seconds. This demonstrates the stabilization phase's role in "providing smooth transients," solving the problem of sudden power changes after interference in traditional single-control modes, and meeting the power grid specifications for voltage stability.

[0121] Reference Figure 3 This figure highlights the innovative value of multi-dimensional upstream turbine unit optimization. Traditional single-turbine reference errors reach 8.5%, while random turbine units, although reduced to 5.2%, still exceed the threshold. This invention, through "central power generation unit selection + high / low / horizontal turbine similarity comparison (≥80% / 90%)", combined with surface feature correction, reduces the error to 1.8%. This verifies the effectiveness of dynamic upstream turbine unit combination, solves the prediction deviation problem caused by unreasonable single-turbine references or combinations in traditional schemes, and provides a precise basis for subsequent power regulation.

[0122] Reference Figure 4 The figure shows that in the traditional non-coordinated mode, the various indicators are scattered, and wake avoidance and grid connection compatibility are less than 55%; some coordination only links the power generation and transmission and distribution units, with limited improvement. This invention, through the coordination of the "tidal power turbine array unit - energy management unit - transmission and distribution unit" within the intelligent control module, combined with feedback from the data monitoring and analysis module, achieves a comprehensive breakthrough of over 85% in indicators such as power stability and wake avoidance. This confirms the role of coordinated control in optimizing the overall efficiency of the tidal power array and solves the "each fighting their own battle" problem of traditional decentralized control.

[0123] Reference Figure 5 The figure shows that voltage fluctuations reach ±20% without intervention, with a long recovery period. Although the control phase accelerates the initial recovery, fluctuations still occur by ±10% after the interference disappears. This invention injects reactive power to raise the voltage during the interference period (0-3s) through the control phase. After the interference disappears, a stabilization phase is initiated (3-8s), where power is finely adjusted at finite time intervals to smoothly return the voltage to its rated value, with fluctuations ≤5%. This reflects the core value of the stabilization phase in "suppressing transient impacts," meeting the stringent requirements of the power grid for voltage stability.

[0124] 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 tidal current turbine array flow field sensing and intelligent control system, characterized in that, It includes a flow field sensing module, an array status monitoring module, an intelligent control module, a data storage and analysis module, and an emergency response module; The flow field sensing module consists of an underwater acoustic Doppler current meter, a three-dimensional flow direction sensor, a pressure sensor, and a data preprocessing unit. The array status monitoring module equips each turbine with a speed sensor, blade angle of attack sensor, torque sensor, vibration acceleration sensor, and generator electrical parameter acquisition device; The intelligent control module includes a main controller and a distributed turbine controller. The main controller uses an industrial-grade PLC, and the distributed controller is connected one-to-one with each turbine. Based on flow field perception data and status monitoring data, the blade angle of attack and turbine speed are adjusted through the maximum power point tracking algorithm, and PID control is used to achieve real-time correction of yaw angle. The data storage and analysis module consists of a local storage server and a cloud database. The local server uses a solid-state drive array. The cloud database synchronizes data via a 4G / 5G module and has a built-in BP neural network model. The emergency response module integrates an anomaly detection unit and a handling execution unit. The anomaly detection unit can set thresholds for parameters such as flow rate, vibration acceleration, and generator overvoltage.

2. The tidal current turbine array flow field sensing and intelligent control system according to claim 1, characterized in that, It also includes a short-term flow field prediction submodule, which is based on real-time data from the flow field sensing module and a historical database, and calculates the flow field prediction value by V. t+Δt =V t ×(1+aΔt)+bV t―1 Calculate the future short-term flow velocity; where V t+Δt V represents the flow velocity at a future time Δt. t V represents the current flow velocity. t―1 Δt represents the flow velocity at the previous moment, a is the flow velocity attenuation coefficient, b is the historical data weighting coefficient, and Δt is the prediction time interval. The submodule has a built-in data cleaning unit. Flow velocity prediction can reserve adjustment time for the intelligent control module.

3. The tidal current turbine array flow field sensing and intelligent control system according to claim 1, characterized in that, It also includes a power optimization and conditioning submodule, which is configured according to... Calculate the target power, where P t Let k be the overall efficiency coefficient of the turbine, ρ be the density of seawater, A be the swept area of ​​the turbine, and V be the target output power. t Given the current flow rate; the submodule constructs a power deviation graded adjustment mechanism.

4. The tidal current turbine array flow field sensing and intelligent control system according to claim 1, characterized in that, The flow field sensing module also includes a dynamic calibration submodule, which employs a three-level calibration mechanism: standard signal comparison, deviation compensation, and effect verification. It periodically calls preset standard flow velocity and pressure signals, calculates the deviation rate using δ = (measured value - standard value) / standard value, and generates a linear calibration coefficient k. c al=1 / (1+δ); The calibration cycle is adaptively adjusted according to the flow field stability: calibrate once every 24 hours when the flow velocity fluctuation is ≤0.3m / s; calibrate once every 12 hours when the fluctuation is 0.3-1m / s; calibrate once every 6 hours when the fluctuation is >1m / s; The calibration process adopts online compensation and offline verification mode, and cross-verification is performed through adjacent sensor data after calibration; The flow field sensing module also includes a wave field reconstruction and power correction submodule: Wave field reconstruction: based on the number of pressure sensors, using the formula ΔP is the pressure fluctuation, ρ is the seawater density, and g is the gravitational acceleration. Calculate the wave height H and wave period T. Power correction: Establish a wave-power correlation model—for example, when H>1.5m, power fluctuations increase by 15-20%, and the target power is dynamically reduced by 10% to avoid sudden changes; Collaborative control: Synchronize wave parameters to the intelligent control module. When the waves are violent, reduce the adjustment step size by 50% and extend the interval to 10 seconds.

5. The tidal current turbine array flow field sensing and intelligent control system according to claim 1, characterized in that, The array status monitoring module also includes a fault classification and diagnosis submodule. This submodule is based on an improved decision tree algorithm and combined with random forest-assisted verification to classify faults into three levels: minor, moderate and severe, and refine the judgment logic.

6. The tidal current turbine array flow field sensing and intelligent control system according to claim 1, characterized in that, The intelligent control module also includes a wake avoidance coordination submodule; first, it combines the turbine diameter D and the current flow velocity V. t Calculation of turbulence intensity I and wake influence radius R w =0.5D + 0.15V t / I, and then determine the wake diffusion direction based on the three-dimensional flow direction sensor data, and construct an array wake distribution heat map.

7. The tidal current turbine array flow field sensing and intelligent control system according to claim 1, characterized in that, The data storage and analysis module also includes an offline modeling submodule, which uses a combination of operating condition classification, hybrid modeling, and incremental update techniques to build an optimized model. First, typical flow field operating conditions are quantitatively classified according to the frequency, amplitude, and intensity of flow velocity fluctuations. For each type of operating condition, an LSTM neural network and a BP neural network are integrated to build a correlation model between the flow field operating conditions, control parameters, and power generation efficiency. The model uses a sliding window incremental learning approach.

8. The tidal current turbine array flow field sensing and intelligent control system according to claim 1, characterized in that, The intelligent control module also includes a yaw angle adaptive adjustment submodule, which adjusts the yaw angle according to θ = arccos(V x / V t Calculate the optimal yaw angle, where θ is the yaw angle and V x The velocity component along the turbine's rotation axis is represented; the submodule introduces a multi-factor correction mechanism: if the turbine's vibration acceleration > 2 m / s² 2 The optimal yaw angle is corrected by ±1° in the direction of vibration reduction; if it is in the wake influence zone of the preceding turbine, it is corrected by ±2° to ±3° in combination with the wake direction; yaw adjustment adopts speed grading and torque closed-loop control: when the deviation is 5°-10°, the yaw motor runs at a low speed of 0.5° / s; when the deviation is 10°-15°, it runs at a medium speed of 1° / s; when the deviation is >15°, it runs at a high speed of 1.5° / s; after the adjustment is in place, the yaw locking torque is detected by the torque sensor; at the same time, energy consumption optimization and error avoidance logic are introduced: when the single adjustment angle is <5°, the yaw adjustment action is not started temporarily; when the adjustment angle is <3° for 3 consecutive times, the adjustment interval is extended to 180s; The intelligent control module also includes an array global power optimization submodule, which employs a dual-drive mechanism of numerical simulation pre-training and long-term operational data iteration to achieve the maximum overall power of the array. Numerical simulation pre-training: A flow field-power coupling model was established based on STAR-CCM+ to simulate the impact of single-unit regulation on the global flow under different turbulence intensities; Long-term data iteration: Extract nearly 90 days of running data every 15 days and update the strategy using an LSTM-BP hybrid model; Real-time adjustment: Compare the current total power with the model's optimal value every 30 seconds. If the deviation is greater than 2%, issue instructions based on the wake effect priority. After adjustment, verify the wake change through flow field sensing.

9. The tidal current turbine array flow field sensing and intelligent control system according to claim 1, characterized in that, The emergency response module also includes a tiered handling submodule, which pre-sets full-process solutions for detection, judgment, handling, and verification for six typical emergency scenarios. Scenario 1: First, adjust the blade angle of attack to the feathering position to reduce turbine resistance, while monitoring flow velocity changes. Scenario 2: First, reduce the speed by 20%, and simultaneously determine the fault type through vibration spectrum analysis. If mechanical resonance occurs, adjust the speed to avoid the resonance range; if the flow field is turbulent, adjust the yaw angle by ±5°. Scenario 3: First, reduce the output power by 30% and start the cooling system's underwater circulation pump in high-flow mode. Scenario 4: Switch to redundant sensor data and enable short-term prediction values ​​from the flow field prediction submodule to assist control. Scenario 5: Attempt to unlock by small-amplitude reciprocating adjustment of the angle of attack; if unlocking fails twice, shut down the turbine. Scenario 6: Immediately reduce the excitation current; if the voltage does not recover within 100ms, cut off the output. All handling processes push progress to the operation and maintenance platform in real time. After handling is completed, the equipment status is checked within 10 minutes, and if the restart conditions are met, the turbine is gradually restarted. The emergency response module also includes a local emergency control submodule, which is used to deal with overall control failures. Local data dependency: Based on real-time data from the turbine's built-in sensors, combined with a locally stored 30-day historical optimal parameter database; Control logic: A simplified PID algorithm is adopted, prioritizing safety and then outputting power at 80% of the optimal power at the same historical flow rate. Recovery mechanism: After communication is restored, the main controller status is verified by synchronizing data for nearly 10 seconds, and then gradually switched back to global control within 30 seconds; The graded handling submodule also includes scenario 7 during extreme waves: Triggering conditions: The wave field reconstruction and power correction submodule detects H>3m and T<8s; Handling procedure: 0-100ms: Adjust blade angle of attack to -15° to reduce the frontal area; 100-500ms: Cut off generator output; 500ms-1s: Yaw to be parallel to the wave propagation direction; Restart conditions: H≤1.5m and continue for 5 minutes, then restart at a power gradient of 20% / min.

10. The tidal current turbine array flow field sensing and intelligent control system according to claim 1, characterized in that, It also includes a remote operation and maintenance and energy efficiency assessment sub-module. This sub-module is based on a cloud platform to build an integrated system of visual monitoring, intelligent assessment, and precise operation and maintenance. Remote visualization uses WebGL technology to generate a 3D interface, displaying the array layout, flow field heat map, turbine status dashboard, and fault location map in real time. Energy efficiency assessment generates multi-dimensional reports on a daily, weekly, and monthly basis: core indicators include total array power generation, power generation per unit flow velocity, equipment utilization rate, and wake influence coefficient; comparison indicators include the deviation rate from historical best values ​​and design values; operation and maintenance suggestions are generated using rule reasoning and case matching: based on equipment runtime, fault frequency, and energy efficiency decline trend, combined with a historical operation and maintenance case library, the optimal solution is output; remote control adopts a dual authentication, command rehearsal, and feedback confirmation security mechanism: operation and maintenance personnel must be authenticated by fingerprint and password, and the effect is rehearsed in a virtual simulation environment before the command is issued. The command is executed only after the rehearsal is passed.

Images