A wind farm commissioning stage risk dynamic identification method and system based on digital twinning and related devices

Abstract

This invention provides a method, system, and related apparatus for dynamic risk identification during the commissioning phase of a wind farm based on digital twins, comprising the following steps: Step 1, preprocessing the collected commissioning data of the target wind farm at time t-1 to obtain a multi-source fusion feature vector; Step 2, using the multi-source fusion feature vector as input to a pre-constructed wind farm digital twin model to obtain the predicted value of the wind turbine status at time t; the wind farm digital twin model is constructed from a mechanism model and a data model; Step 3, calculating the deviation between the predicted value of the wind turbine status at time t and the actual detected value of the wind turbine status at time t; Step 4, identifying the risk of the wind turbine commissioning status of the target wind farm based on the multi-source fusion feature vector and the deviation; This invention achieves closed-loop intelligent control from risk perception, identification, assessment to decision-making and execution, significantly improving the safety, accuracy, and efficiency of the wind farm commissioning process.

Description

Technical Field

[0001] This invention belongs to the field of wind farm commissioning technology, specifically relating to a method, system, and related devices for dynamic risk identification during the commissioning phase of a wind farm based on digital twins. Background Technology

[0002] With the continuous growth of wind power installed capacity and the increase in unit capacity, the structure and operation mode of wind farm systems are becoming increasingly complex. The commissioning phase, as a critical stage before grid connection, involves frequent switching of operating conditions and continuous parameter tuning, requiring the coordinated operation of multiple subsystems such as the wind turbine control system, substation electrical equipment, and communication and protection systems. Common risks during this process include, but are not limited to: power or frequency oscillations caused by improper control parameter tuning; link instability due to communication delays or synchronization anomalies; malfunctions or failures to operate due to incorrect protection logic settings; and system instability caused by external meteorological disturbances (sudden wind speed changes, low voltage ride-through, lightning strikes, etc.).

[0003] Traditional debugging risk identification relies heavily on human experience and offline data analysis, which has the following shortcomings: 1) Delayed and lack of real-time detection: Traditional debugging risk identification relies on manual inspection and post-event data analysis. Risk events are often only discovered after they occur, making it difficult to capture the dynamic changes in the system's operating status in a timely manner, resulting in delayed response and risk spread. 2) Subjective and inaccurate judgment: Human experience judgment is affected by factors such as the skill level of personnel and the limited data sample, making it difficult to quantify the relationship between multiple coupled factors in complex systems. This can easily lead to misjudgment or omission, affecting the accuracy and consistency of risk identification. Summary of the Invention

[0004] The purpose of this invention is to provide a method, system, and related device for dynamic risk identification during the commissioning phase of a wind farm based on digital twins. By using digital twin models and intelligent algorithms, it achieves multi-source data fusion, dynamic risk identification, and strategy linkage, thereby improving the safety, accuracy, and efficiency of the commissioning phase and solving the aforementioned shortcomings in the prior art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins, comprising the following steps: Step 1: Preprocess the collected commissioning data of the target wind farm at time t-1 to obtain the multi-source fusion feature vector; Step 2: Use the multi-source fusion feature vector as input to the pre-constructed wind farm digital twin model to obtain the predicted value of the wind turbine state at time t of the target wind farm; the wind farm digital twin model is constructed from the mechanism model and the data model. Step 3: Calculate the deviation between the predicted value of the wind turbine status at time t and the actual detected value of the wind turbine status at time t in the target wind farm. Step 4: Based on the multi-source fusion feature vector and the deviation, identify the risk of wind turbine commissioning status in the target wind farm.

[0006] Preferably, the mechanism model is used to describe the physical operation mechanism of the wind turbine and electrical system, and is constructed based on the wind energy conversion model, the electrical grid connection model and the wind turbine control model.

[0007] Preferably, the expression of the mechanism model is:

[0008] in, for; for; Let be the mechanical power output from the fan side at time t; Let be the active power output from the grid-connected side at time t; Let t be the reactive power output from the grid-connected side. for; Let t be the angular velocity of the wind turbine shaft at time t; The mathematical model is expressed as follows:

[0009] in, It is the predicted value at time t; Represents the LSTM network function; n It is the length of the time window; This refers to the time series in the multi-source fusion feature vector.

[0010] Preferably, the expression for the pre-constructed digital twin model of the wind farm is:

[0011] in, The predicted value of the wind turbine state at time t for the target wind farm; For the mechanism model in t The physical response output vector at time t; It is the predicted value at time t.

[0012] Preferably, risk identification is performed based on the multi-source fusion feature vector and the deviation, specifically by: The deviation is smoothed to obtain a smoothed deviation; Calculate Mahalanobis distance based on multi-source fused feature vectors; Risk dynamics are identified based on the obtained smoothing deviation and Mahalanobis distance.

[0013] Preferably, the identification method further includes classifying risk levels based on deviation, specifically: Risk score is calculated based on deviation. Risk levels are determined based on risk score.

[0014] Preferably, the method further includes generating an optimal control strategy based on risk level and multi-source fused data vectors, specifically through the following method: Optimization problem based on multi-source fused data vector construction; The optimal control vector is obtained by solving the constructed optimization problem. The obtained optimal control vector is transformed to obtain the optimal control strategy.

[0015] Secondly, the present invention provides a dynamic risk identification system for the commissioning phase of a wind farm based on digital twins, comprising: The data preprocessing unit is used to preprocess the collected commissioning data of the target wind farm at time t-1 to obtain a multi-source fusion feature vector; The wind turbine state prediction unit is used to take the multi-source fused feature vector as input to the pre-constructed wind farm digital twin model to obtain the predicted wind turbine state value of the target wind farm at time t; the wind farm digital twin model is constructed from the mechanism model and the data model; The state deviation calculation unit is used to calculate the deviation between the predicted value of the wind turbine state at time t of the target wind farm and the actual detected value of the wind turbine state at time t of the target wind farm. The commissioning risk identification unit is used to identify risks in the commissioning status of wind turbines in the target wind farm based on the multi-source fusion feature vector and the deviation amount.

[0016] Thirdly, the present invention provides an electronic device including a processor and a memory, wherein the memory stores computer instructions, and when the computer instructions are executed by the processor, the electronic device performs the method described thereon.

[0017] Fourthly, the present invention provides a computer program product, the computer program product including computer-executable instructions, which, when executed, implement any of the methods described herein.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This invention provides a method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins. By real-time collection and fusion of multi-source information such as SCADA, protection actions, communication, meteorological, and historical data, combined with synchronous comparison of the virtual and real digital twin model, it achieves continuous monitoring and dynamic early warning of the system status. This overcomes the identification lag problem caused by traditional manual offline analysis, enabling early detection of potential risks and real-time tracking of their evolution, significantly improving the real-time and forward-looking nature of risk perception. The invention employs a hybrid twin architecture that combines mechanistic and data models, maintaining the causal relationships and interpretability of the physical model while utilizing data-driven methods such as LSTM to learn complex nonlinear dynamics. This enhances the model's adaptability and prediction accuracy under complex and variable wind farm conditions, thereby significantly improving the accuracy and robustness of risk judgment and effectively reducing misjudgments and omissions.

[0019] Furthermore, through a quantitative risk scoring and grading system, the system can automatically construct multi-objective optimization problems, generate optimal control strategies such as adjusting pitch angle and power setpoints, and convert them into executable commands. This enables closed-loop automatic control from risk identification and decision-making to execution, greatly improving the efficiency and accuracy of risk management and reducing response delays and operational errors caused by human subjective factors. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the present invention.

[0021] Figure 2 This is the system architecture diagram of the present invention. Detailed Implementation

[0022] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0023] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0024] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0025] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0026] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0027] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0028] Example 1 This embodiment provides a method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins, which includes the following steps: Step 1: Preprocess the collected commissioning data of the target wind farm at time t-1 to obtain the multi-source fusion feature vector; Step 2: Use the multi-source fusion feature vector as input to the pre-constructed wind farm digital twin model to obtain the predicted value of the wind turbine state at time t of the target wind farm; the wind farm digital twin model is constructed from the mechanism model and the data model. Step 3: Calculate the deviation between the predicted value of the wind turbine status at time t and the actual detected value of the wind turbine status at time t in the target wind farm. Step 4: Based on the multi-source fusion feature vector and the deviation, identify the risk of wind turbine commissioning status in the target wind farm.

[0029] Example 2 like Figure 1 and Figure 2 As shown in the figure, this embodiment provides a method for dynamic risk identification and decision support during the commissioning phase of a wind farm based on digital twins, which includes the following steps: Step 1: Multi-type data collection and fusion of wind farms, realizing the collection, standardization and feature fusion of multi-type data during the wind farm commissioning phase.

[0030] (1) Multi-type data acquisition: Collect various types of data during the wind farm commissioning phase through multi-channel sensors and data interfaces. : (1) in, It is the monitoring data of the supervisory control and data acquisition system (SCADA) of the unit, including active / reactive power, voltage, current, speed, wind speed and blade pitch angle, etc. It is a record of actions of the protection and control system, including action signals, alarm codes and trigger logic identifiers, etc. It is communication data, including link delay, packet loss rate, and clock synchronization deviation; It includes meteorological and environmental data, such as wind direction, temperature, air pressure, and humidity. It is a historical commissioning sample, including past commissioning risks, equipment failures and recovery records.

[0031] (2) Preprocessing and fusion of multiple data types To standardize the units of measurement and improve data comparability and numerical stability, the Z-score method was used to evaluate the original data. Standardization process: (2) in, It is the average value of the data; The standard deviation is denoted as .

[0032] Principal Component Analysis (PCA) was used to analyze the standardized... Multi-source fusion feature vectors are fused into a unified dimension. : (3) in, This represents a PCA-based fusion function; the output... Used to describe wind farms t The operational status at any given moment.

[0033] Step 2: Construction of the digital twin model of the wind farm. The digital twin model of the wind farm is composed of a mechanism model and a data model.

[0034] (1) Mechanism Model: This model describes the physical operation mechanism of the wind turbine and its electrical system, and is based on three main models: wind energy conversion, grid connection, and turbine control. The input to the mechanism model is the multi-source fusion feature vector obtained in step 1. The characteristics related to physical quantities (such as wind speed, voltage, and blade pitch angle) are output as the physical response variables of the system.

[0035] 1) Wind energy conversion model: (4) in, It is the mechanical power output from the fan side; It is air density; It is the impeller swept area; It's about the tip speed ratio. With pitch angle The power coefficient function; It's wind speed.

[0036] 2) Electrical grid connection model: (5) in, , These are the active power and reactive power output from the grid-connected side, respectively. V It is the grid-connected voltage; I am Grid-connected current; φ It is the power factor angle.

[0037] 3) Wind turbine control model, including pitch control and speed control, wherein: Pitch control: (6) in, P ref This is the reference power; K p and K i These are the proportional and integral coefficients of the proportional-integral controller.

[0038] Speed ​​control: (7) in, T e It is electromagnetic torque; K t It is the current torque adjustment coefficient; ω It is the rotational angular velocity of the wind turbine shaft; ω ref It is the reference rotational angular velocity.

[0039] Therefore, combining formulas (4)-(7), the overall output of the mechanistic model is: (8) in, For the mechanism model at time t The physical response output vector; This is a physical mechanism mapping function established based on wind energy conversion, grid connection, and wind turbine control; For the wind turbine at all times t The pitch angle; For the wind turbine shaft at time t angular velocity of rotation.

[0040] (2) Data Model: To compensate for the shortcomings of the mechanistic model, a Long Short-Term Memory (LSTM) network is used to capture complex nonlinear dynamic characteristics. The input to the data model is the multi-source fusion feature vector obtained from step 1. Time series The output is the predicted system state value: (9) in, It is the predicted value at time t; Represents the LSTM network function; n It is the length of the time window.

[0041] (3) Integration of mechanism model and data model The output of the mechanism model (8) is superimposed with the output of the data model (9) to obtain the prediction output of the wind farm digital twin model: (10) in, Let t be the predicted value of the wind turbine state at time t in the target wind farm.

[0042] This hybrid twin structure, which integrates "mechanism + data," retains the physical interpretability of the model while introducing data learning capabilities, which can significantly improve the prediction accuracy and robustness under complex dynamic conditions.

[0043] (4) Virtual-real synchronization and deviation calculation Compare the predicted wind turbine states at time t of the target wind farm Actual measured value of wind turbine status at time t of the target wind farm : (11) in, The deviation is a quantity that characterizes the health status of the digital twin model of a wind farm. ( When the threshold is set to a preset value, the system determines that the virtual and real states deviate significantly, triggering step three: dynamic risk identification and grading assessment.

[0044] Step 3: Dynamic Risk Identification and Classification Assessment Based on the multi-source fusion data vector obtained in step 1 Deviation from step 2 This step involves identifying and classifying debugging risks.

[0045] (1) Risk Identification To mitigate the impact of instantaneous fluctuations, the deviation is smoothed (i.e., the instantaneous deviation is time-filtered). The smoothed deviation is defined as: (12) in, This is the smoothing coefficient. If... Preliminary assessment indicates that the system poses a risk.

[0046] In addition, calculate the Mahalanobis distance. To assess the degree of deviation in the state distribution: (13) in, It is the covariance matrix; if If so, the system is considered to be in an abnormal operating state.

[0047] If the conditions in formula (12) are met simultaneously With formula (13) The system was confirmed to have experienced a risk event.

[0048] (2) Risk classification To reflect the intensity of risk events, a risk score is defined as follows: (14) in, This is the largest deviation in historical data; It refers to the duration of the risk; This is the normal operating time; This refers to the weighting of equipment importance; These are the weighting coefficients for deviation intensity, duration, and equipment importance, respectively.

[0049] Based on risk score Risk level classification L : 1) If The risk level is L 1 (Slight deviation), continue monitoring and recording; 2) If The risk level is L2 (Medium deviation), triggering an early warning; 3) If The risk level is L 3 (Severe deviation), implement protection.

[0050] Step 4: Precise Decision Making and Strategy Generation Based on the risk score obtained in step 3 Risk level L This step combines multi-source fused data vectors This generates the optimal control strategy, enabling rapid risk suppression and operational recovery.

[0051] When the risk level is determined to be L 2 or L At time 3, the system initiates an active control mechanism and constructs the following optimization problem: (15) in, J It is an optimization objective function; These are the control variables to be solved (such as pitch angle, active / reactive power setpoints, voltage reference values, etc.). These are weighting coefficients; To control the upper and lower limits of variables; This is the maximum allowable adjustment rate.

[0052] By solving the optimization problem (15), the optimal control vector is obtained: (16) Then, based on the communication protocol and interface specifications between the wind farm SCADA system and the underlying actuators (such as pitch systems and converters), the transformation function is used. Will Transformed into an executable set of control commands: (17) Ultimately, the optimal control command set is generated. Real-time control is achieved by sending control signals to the corresponding actuators (such as pitch systems, converters, etc.) through the control interface or SCADA system.

[0053] Step 5, Closed-loop feedback and model self-learning This step utilizes the optimal control command set generated and executed in step 4. The resulting system feedback enables closed-loop verification and dynamic optimization of the digital twin model and decision-making strategy.

[0054] (1) Evaluation of the effectiveness of the optimal control command set By comparing the execution of digital twin models The predicted state and the new actual monitoring data are then used to calculate the execution. The subsequent discrepancy between real and virtual reality: (18) Will Deviation from before execution Comparison, for quantitative assessment Validity: If If the control is deemed effective, the model's prediction accuracy is improved; if This triggers the self-learning mechanism of the wind farm digital twin model.

[0055] (2) Self-learning of digital twin models of wind farms To ensure that the digital twin model of the wind farm continuously approximates the characteristics of the actual system, gradient descent is used for self-learning. (19) in, It is a loss function; It is the learning rate; It is the first j The twin model parameters at the next iteration It will be continuously updated during the iteration process and fed back into the wind farm digital twin model in step 2 to predict the system state at the next moment. This enables the model to make more accurate predictions and decisions in similar risk scenarios in the future.

[0056] This embodiment establishes a dynamic risk identification and precise decision support system for the wind farm commissioning phase by integrating multiple types of data and digital twin technology. It realizes intelligent control of the entire process from risk identification and hierarchical assessment to intelligent decision-making and closed-loop optimization, effectively improving the safety, accuracy and work efficiency of the wind farm commissioning phase. It has high engineering application value and promotion significance.

[0057] Example 3 This embodiment provides a dynamic risk identification system for the commissioning phase of a wind farm based on digital twins, including: The data preprocessing unit is used to preprocess the collected commissioning data of the target wind farm at time t-1 to obtain a multi-source fusion feature vector; The wind turbine state prediction unit is used to take the multi-source fused feature vector as input to the pre-constructed wind farm digital twin model to obtain the predicted wind turbine state value of the target wind farm at time t; the wind farm digital twin model is constructed from the mechanism model and the data model; The state deviation calculation unit is used to calculate the deviation between the predicted value of the wind turbine state at time t of the target wind farm and the actual detected value of the wind turbine state at time t of the target wind farm. The commissioning risk identification unit is used to identify risks in the commissioning status of wind turbines in the target wind farm based on the multi-source fusion feature vector and the deviation amount.

[0058] Example 4 This embodiment also provides a computing device. The computing device includes a bus, a processor, a memory, and a communication interface. The processor, memory, and communication interface communicate with each other via the bus. The computing device can be a server or a terminal device. It should be understood that this application does not limit the number of processors and memory in the computing device.

[0059] A bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, a bus can include a path for transmitting information between various components of a computing device (e.g., memory, processor, communication interfaces).

[0060] The processor may include any one or more of the following: central processing unit (CPU), graphics processing unit (GPU), tensor processing unit (TPU), application specific integrated circuit (ASIC), field-programmable gate array (FPGA), microprocessor (MP), or digital signal processor (DSP).

[0061] Memory can include volatile memory, such as random access memory (RAM). Processors can also include non-volatile memory. volatile memory, such as read-only memory (ROM). ROM (memory only), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0062] The memory stores executable program code, which the processor executes to implement the functions of the aforementioned units, thereby achieving, for example, the method described in Embodiment 1. That is, the memory may store instructions for the methods and functions relating to the computing device in any of the above embodiments.

[0063] The communication interface uses transceiver modules such as, but not limited to, network interface cards and transceivers to enable communication between computing devices and other devices or communication networks.

[0064] Example 5 This embodiment also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, cause the processor to perform the methods and functions of the computing device involved in any of the above embodiments.

[0065] Generally, the various embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software, which can be executed by a controller, microprocessor, or other computing device. Although various aspects of the embodiments of this disclosure are shown and described as block diagrams, flowcharts, or represented using some other illustration, it should be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as, as non-limiting examples, in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0066] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins, characterized in that, Includes the following steps: Step 1: Preprocess the collected commissioning data of the target wind farm at time t-1 to obtain the multi-source fusion feature vector; Step 2: Use the multi-source fusion feature vector as input to the pre-constructed wind farm digital twin model to obtain the predicted value of the wind turbine state at time t of the target wind farm; the wind farm digital twin model is constructed from the mechanism model and the data model. Step 3: Calculate the deviation between the predicted value of the wind turbine status at time t and the actual detected value of the wind turbine status at time t in the target wind farm. Step 4: Based on the multi-source fusion feature vector and the deviation, identify the risk of wind turbine commissioning status in the target wind farm.

2. The method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins according to claim 1, characterized in that, The mechanism model is used to describe the physical operation mechanism of wind turbines and electrical systems, and is constructed based on wind energy conversion model, electrical grid connection model and wind turbine control model.

3. The method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins according to claim 2, characterized in that, The expression for the mechanism model is: in, for; for; Let be the mechanical power output from the fan side at time t; Let be the active power output from the grid-connected side at time t; Let t be the reactive power output from the grid-connected side. for; Let t be the angular velocity of the wind turbine shaft at time t; The mathematical model is expressed as follows: in, It is the predicted value at time t; Represents the LSTM network function; n It is the length of the time window; This refers to the time series in the multi-source fusion feature vector.

4. The method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins according to claim 1, characterized in that, The expression for the pre-constructed digital twin model of a wind farm is: in, The predicted value of the wind turbine state at time t for the target wind farm; For the mechanism model in t The physical response output vector at time t; It is the predicted value at time t.

5. The method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins according to claim 1, characterized in that, Risk identification is performed based on the multi-source fusion feature vector and the deviation amount. The specific method is as follows: The deviation is smoothed to obtain a smoothed deviation; Calculate Mahalanobis distance based on multi-source fused feature vectors; Risk dynamics are identified based on the obtained smoothing deviation and Mahalanobis distance.

6. The method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins according to claim 1, characterized in that, The identification method also includes classifying risk levels based on deviation, specifically: Risk score is calculated based on deviation. Risk levels are determined based on risk score.

7. The method for dynamic risk identification during the commissioning phase of a wind farm based on digital twins according to claim 1, characterized in that, The method also includes generating an optimal control strategy based on risk level and multi-source fused data vectors, specifically: Optimization problem based on multi-source fused data vector construction; The optimal control vector is obtained by solving the constructed optimization problem. The obtained optimal control vector is transformed to obtain the optimal control strategy.

8. A dynamic risk identification system for the commissioning phase of a wind farm based on digital twins, characterized in that, include: The data preprocessing unit is used to preprocess the collected commissioning data of the target wind farm at time t-1 to obtain a multi-source fusion feature vector; The wind turbine state prediction unit is used to take the multi-source fused feature vector as input to the pre-constructed wind farm digital twin model to obtain the predicted wind turbine state value of the target wind farm at time t; the wind farm digital twin model is constructed from the mechanism model and the data model; The state deviation calculation unit is used to calculate the deviation between the predicted value of the wind turbine state at time t of the target wind farm and the actual detected value of the wind turbine state at time t of the target wind farm. The commissioning risk identification unit is used to identify risks in the commissioning status of wind turbines in the target wind farm based on the multi-source fusion feature vector and the deviation amount.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing computer instructions that, when executed by the processor, cause the electronic device to perform the method of any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes computer-executable instructions that, when executed, implement the method of any one of claims 1 to 7.

Images