Offshore wind power cable monitoring and early warning system and method based on digital twinning

Abstract

The invention discloses an offshore wind power cable monitoring and early warning system and method based on digital twinning, and relates to the technical field of digital twinning and industrial Internet of Things monitoring, and the system comprises a data collection module which is used for obtaining the real-time multi-dimensional sensing data of a submarine cable, and determining the real-time heat production power of the cable; the physical rule model module is used for outputting a theoretical prediction state; the data driving model module is used for outputting a data fitting state; the bidirectional dynamic coupling module is used for generating a mismatch judgment result for indicating cognitive mismatch or cognitive consistency; the model correction module is used for outputting a corrected theoretical prediction state; the state evaluation and early warning module is used for generating the comprehensive evaluation result; and comparing the comprehensive evaluation result and the change trend thereof with a preset multi-level risk threshold, and issuing early warning information. According to the method, the limitation of traditional physical model solidification can be broken through, self-correction is carried out when encountering extremely unknown working conditions, and effective insight and prevention of deep risks are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital twinning and industrial internet of things monitoring, more specifically, the present application relates to a marine wind power cable monitoring and early warning system and method based on digital twinning. BACKGROUND

[0002] The submarine cable of the offshore wind power facility faces severe challenges in safe and stable operation due to long-term exposure to complex and variable marine environments. The existing technology uses digital twinning for monitoring, which is usually based on first principles to construct physical law models. The fatal flaw of such models is that their internal laws and assumptions are static and unchangeable once established. When encountering complex extreme marine events such as typhoons and submarine landslides, the actual physical boundary conditions of the cable, such as seabed coverage and surrounding medium thermal conductivity, will change dramatically, exceeding the original assumptions of the physical model. At this time, the physical model itself cannot reflect the physical reality, resulting in the loss of early warning capability of the entire twinning system for unknown risks. Therefore, how to construct a new monitoring and early warning paradigm that can dynamically adapt to environmental changes and has self-correction ability is a technical problem that needs to be solved in the field.

[0003] The above information disclosed in the background section is only used to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a marine wind power cable monitoring and early warning system and method based on digital twinning, the technical scheme is as follows, a marine wind power cable monitoring and early warning system based on digital twinning, comprising: A data acquisition module is configured to acquire real-time multi-dimensional sensing data of the submarine cable, wherein the real-time multi-dimensional sensing data includes real-time working condition data, and the real-time heat generation power of the cable is determined according to the real-time working condition data. A physical law model module is configured to receive the real-time multi-dimensional sensing data and output a theoretical prediction state. A data-driven model module is configured to receive the real-time multi-dimensional sensing data in parallel and output a data fitting state. A bidirectional dynamic coupling module is configured to acquire the theoretical prediction state and the data fitting state, calculate a model cognitive bias, compare the model cognitive bias with a preset cognitive mismatch threshold, and generate a mismatch discrimination result indicating cognitive mismatch or cognitive consistency. A model correction module is configured to, in response to the mismatch discrimination result being cognitive mismatch, calculate a physical parameter correction amount based on a preset adaptive gain, update the physical law model module using the physical parameter correction amount, and output a corrected theoretical prediction state. a state evaluation and early warning module, configured to: when the mismatch discrimination result is cognitive consistency, fuse the theoretical prediction state and the data fitting state to generate a comprehensive evaluation result; when the mismatch discrimination result is cognitive mismatch, fuse the corrected theoretical prediction state output by the model correction module and the data fitting state to generate the comprehensive evaluation result; compare the comprehensive evaluation result and a change trend of the comprehensive evaluation result with a plurality of preset risk threshold values, and issue early warning information.

[0005] Optionally, the physical law model is a long short-term memory network model based on a finite element method.

[0006] Optionally, the model cognitive bias is calculated by normalizing a difference between the theoretical prediction state and the data fitting state by an absolute value of the data fitting state.

[0007] Optionally, the cognitive mismatch threshold is determined by statistically analyzing model cognitive biases generated under historical normal working conditions and taking a percentile point of a distribution of the model cognitive biases.

[0008] Optionally, the model correction module calculates the physical parameter correction amount by normalizing a difference between the theoretical prediction state and the data fitting state by the real-time heat generation power of the cable and multiplying the difference by a preset adaptive gain.

[0009] Optionally, the real-time working condition data include real-time currents obtained from a power grid SCADA system, and the real-time heat generation power of the cable is determined according to the real-time currents and a known resistance of the cable.

[0010] Optionally, the plurality of risk threshold values include attention, alarm, and danger levels.

[0011] A digital-twin-based offshore wind power cable monitoring and early warning method, comprising: obtaining real-time multi-dimensional sensing data of a submarine cable, the real-time multi-dimensional sensing data including real-time working condition data, and determining a real-time heat generation power of the cable according to the real-time working condition data; parallel processing the real-time multi-dimensional sensing data to respectively obtain a theoretical prediction state and a data fitting state; calculating a model cognitive bias between the theoretical prediction state and the data fitting state, comparing the model cognitive bias with a preset cognitive mismatch threshold, and generating a mismatch discrimination result indicating cognitive mismatch or cognitive consistency; judging the mismatch determination result: if it is cognitive mismatch, calculating a physical parameter correction amount based on a preset adaptive gain, updating a physical law model to obtain a corrected theoretical prediction state, and entering the next step; if it is cognitive consistency, directly entering the next step; performing state fusion: if the mismatch determination result is cognitive mismatch, fusing the corrected theoretical prediction state and the data fitting state; if the mismatch determination result is cognitive consistency, fusing the theoretical prediction state and the data fitting state to generate a comprehensive evaluation result; comparing the comprehensive evaluation result and a change trend of the comprehensive evaluation result with a plurality of preset risk threshold values, and issuing a warning information.

[0012] Compared with the prior art, the present application has the following beneficial effects: 1. The present application can break through the limitations of traditional physical model solidification, and perform self-correction when encountering extreme unknown working conditions such as typhoon and submarine landslide, instead of systematic failure, thereby realizing effective insight and prevention of deep-seated risks; 2. The present application sets a cognitive mismatch threshold by statistically analyzing model cognitive bias under historical normal working conditions, can scientifically distinguish between normal prediction fluctuations and abnormal deviations caused by real environment changes, and greatly improves the reliability of the system in complex marine environments; 3. When the physical model deviates from the real data, the system actively intervenes, according to the real situation revealed by the data-driven model, reversely corrects the key parameters in the physical law model, constitutes a close-loop feedback, and forces the physical model to evolve in a direction more consistent with the physical reality in real time and online; 4. By presetting a plurality of risk threshold values such as attention, warning and danger, the cable health state is finely evaluated and graded warning is made, so that the warning information is more forward-looking and operable, can help the operation and maintenance personnel to make more reasonable and effective resource allocation and decision response, and realizes the change from passive response to active prevention. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 The system flowchart of the present application.

[0014] Figure 2 The method flowchart of the present application. DETAILED DESCRIPTION

[0015] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0016] Embodiment 1 Please refer to Figure 1 The application discloses a digital-twin-based offshore wind power cable monitoring and early warning system, comprising: A data acquisition module is configured to acquire real-time multi-dimensional sensing data of the submarine cable, wherein the real-time multi-dimensional sensing data comprises real-time working condition data, and the real-time cable heat generation power is determined according to the real-time working condition data. A physical law model module is configured to receive the real-time multi-dimensional sensing data and output a theoretical prediction state. A data-driven model module is configured to receive the real-time multi-dimensional sensing data in parallel and output a data fitting state. A bidirectional dynamic coupling module is configured to acquire the theoretical prediction state and the data fitting state, calculate a model cognitive bias, compare the model cognitive bias with a preset cognitive mismatch threshold, and generate a mismatch discrimination result indicating cognitive mismatch or cognitive consistency. A model correction module is configured to, in response to the mismatch discrimination result being cognitive mismatch, calculate a physical parameter correction amount based on a preset adaptive gain, update the physical law model module using the physical parameter correction amount, and output a corrected theoretical prediction state. A state assessment and early warning module is configured to, when the mismatch discrimination result is cognitive consistency, fuse the theoretical prediction state and the data fitting state to generate a comprehensive assessment result; when the mismatch discrimination result is cognitive mismatch, fuse the corrected theoretical prediction state output by the model correction module and the data fitting state to generate a comprehensive assessment result; and compare the comprehensive assessment result and a change trend of the comprehensive assessment result with a preset multi-level risk threshold to issue early warning information. The application discloses a digital-twin-based offshore wind power cable monitoring and early warning system, which breaks through the limitations of existing physical models being fixed and unable to adapt to dramatic changes in marine environments. The system continuously captures actual submarine cable operating data through a data acquisition module, such as distributed fiber optic sensing technology. This data is simultaneously fed into a physical law model module and a data-driven model module, forming a dual-path parallel prediction architecture. The key to this parallel prediction architecture lies in the use of a bidirectional dynamic coupling module to quantify the cognitive bias between the two models in real time. Once the bias exceeds the statistically normal fluctuation range, the physical model is deemed distorted. At this point, the model correction module goes beyond passive alarms and proactively intervenes, reconstructing key parameters in the physical law model based on the actual conditions revealed by the data-driven model. Finally, the status assessment and early warning module integrates the dynamically verified or corrected optimal model output to assess the cable's health and issue multi-level early warnings. This architecture enables the system to self-correct rather than fail systematically when encountering extreme unknown conditions such as typhoons and submarine landslides, thereby achieving effective insight into and prevention of deep-seated risks.

[0017] Example 2 The physical law model is a long short-term memory network model based on the finite element method; The cognitive bias of the model is calculated by normalizing the difference between the theoretically predicted state and the data-fitted state by the absolute value of the data-fitted state; To further illustrate, in the specific implementation of the present invention, the physical law model module is a thermal-mechanical coupling model constructed based on the finite element method. This model is established based on prior knowledge such as the cable design blueprint and marine geological survey reports to describe its theoretical thermodynamic behavior. In parallel, the data-driven model module uses a long-short-term memory network to capture complex nonlinear correlations by learning massive historical operation data. The core function of the bidirectional dynamic coupling module is implemented through a specific quantitative formula, which is designed to generate model cognitive bias. Its inherent logic is that in order to objectively measure the degree of deviation between the physical model and the actual data, the influence of physical dimensions and numerical amplitudes must be eliminated. Therefore, the idea of ​​standardized relative error in statistics is borrowed to construct a dimensionless indicator. In the scenario of cognitive mismatch, the data-driven model is regarded as a reference system that is closer to physical reality, so its output is normalized. The calculation formula of model cognitive bias is: ; in, for Model cognitive bias at every moment, The theoretical prediction state output by the physical law model module, such as the predicted temperature, The data fitting state output by the data-driven model module corresponds to the same physical quantity. is a positive real constant set to prevent the denominator from being zero, whose value is much smaller than the normal measurement value, and its dimension is the same as Same; the bidirectional dynamic coupling module uses this formula to continuously calculate ,The calculation result is the key judgment basis for triggering the adaptive correction of the model; By comparing a dynamically changing, standardized deviation value with a fixed threshold, the system can accurately identify the moment of failure of the physical model, thereby initiating the subsequent correction process and ensuring the entire twin system's ability to perceive unknown risks.

[0018] Example 3 The cognitive mismatch threshold is determined by statistically analyzing the model cognitive bias generated under historical normal operating conditions and taking the percentile of its distribution; The cognitive mismatch threshold setting logic abandons the experience-based, human-based setting method and is driven by statistical principles. The system must be able to distinguish between normal fluctuations in model predictions and abnormal deviations caused by real changes in the physical environment. A scientific threshold should maximize sensitivity to real anomalies while minimizing the false alarm rate caused by data noise. Specifically, the threshold is determined by first collecting and storing a large amount of historical model cognitive deviations under normal working conditions. , forming a large statistical data set; then the probability distribution analysis of the data set is performed, and a high percentile of the distribution, such as the 99th percentile, is selected as the cognitive mismatch threshold Based on this method, the system establishes a dynamic normal behavior boundary; any deviation beyond this boundary, i.e. , will be determined as a cognitive mismatch event with high confidence, thus reliably triggering the model correction module; This mechanism greatly improves the reliability of the system's operation in complex marine environments and avoids frequent false triggering or missed reporting of key risks due to improper threshold settings.

[0019] Example 4 The model correction module calculates the physical parameter correction in the following way: the difference between the theoretical prediction state and the data fitting state is normalized by the real-time heat generation power of the cable and multiplied by the preset adaptive gain; Real-time operating data includes real-time current obtained from the grid SCADA system. The real-time heating power of the cable is determined by calculation based on the real-time current and the known resistance of the cable; When a cognitive mismatch is confirmed, the model correction module is immediately activated. Its core task is to calculate the correction amount of physical parameters, which is the key execution link to achieve system cognitive flexibility. The calculation formula of this correction amount is derived from the basic relationship of steady-state thermodynamics. , which reveals the temperature difference deviation and the real-time heat generation power of the cable Strong correlation; in order to keep the correction process consistent and stable under different load conditions, the temperature difference must be normalized with the real-time heat generation power; The calculation formula for the physical parameter correction is:

[0020] in, for The physical parameter correction calculated at each moment, here taking equivalent thermal resistance as an example; is a dimensionless adaptive gain hyperparameter; is the difference between the data fitting state and the theoretically predicted state; The real-time heating power of the cable is obtained by obtaining the real-time current of the power grid SCADA system. and the known cable resistance per unit length , according to Joule's law Calculated; It is a minimum positive constant power value that represents non-Joule heating effects such as dielectric loss and is set to avoid calculation divergence near zero load; Adaptive gain The method to determine the value is to calibrate it through offline simulation experiments, simulate a known physical parameter mutation event in the simulation environment, and then For example, in the range of 0.01 to 1.0, find an optimal value that can make the correction process converge fastest without oscillation; Based on the above results, the model correction module calculates After that, it is immediately used to update the corresponding parameters in the physical law model, such as ,in, is the old parameter, is the new parameter after update; This process forms a tight closed-loop feedback loop, forcing the physical model to evolve online and in real time in a direction that is more consistent with physical reality.

[0021] Example 5 Multi-level risk thresholds include concern, warning, and danger levels; The status assessment and early warning module is the final output of the system, and the multi-level risk threshold it relies on is a preset rule system.

[0022] For high-value assets such as offshore wind power, a single binary safety assessment is too crude to meet the needs of refined operation and maintenance and risk management. Providing graded early warnings can help operation and maintenance personnel make more reasonable and effective resource allocation and decision-making responses. The preset multi-level risk thresholds are specifically defined as three levels: attention, warning, and danger. These thresholds are the comprehensive assessment results of the final output of the system. The comprehensive evaluation results It is generated by fusing the theoretical prediction state with the data fitting state. After cognitive mismatch occurs, the revised theoretical prediction state and data fitting state are fused. The application effect of this multi-level risk threshold is that the system can The current value of and its changing trend over time are compared with these three thresholds to issue warning information of different levels; Level of concern: The lowest level of warning; when the comprehensive assessment results When the value or its changing trend reaches the preset attention threshold for the first time, it is triggered. It indicates that there is a slight but noteworthy deviation in the health status of the cable. Although it does not pose a direct threat, the system will mark it and prompt the operation and maintenance personnel to continuously observe and track the data of the cable section. Alarm level: medium-level warning; when comprehensive assessment results If the value or trend deteriorates further, exceeding the concern level and reaching the alarm threshold, a warning message is generated, signaling a more significant abnormality in the cable's health and the emergence of a potential risk. The system issues this warning to operators, prompting them to include the issue in their work plan and prepare for a more detailed analysis or schedule an inspection soon.

[0023] Danger level: the highest level of warning; when the comprehensive assessment results It is triggered when the value or its changing trend reaches or exceeds the most serious danger threshold, which means that the cable may be on the verge of structural or functional damage and the probability of failure is significantly increased. The danger warning issued by the system is the highest priority instruction, requiring operation and maintenance personnel to take immediate intervention measures, such as adjusting the cable load or performing emergency shutdown and maintenance, to avoid catastrophic failure.

[0024] This differentiated early warning mechanism makes early warning information more forward-looking and operational, and realizes the transition from passive response to active prevention.

[0025] Example 6 See also Figure 2 , a digital twin-based offshore wind power cable monitoring and early warning method, including: Acquire real-time multi-dimensional sensor data of submarine cables, including real-time operating condition data, and determine the real-time heat generation power of the cables based on the real-time operating condition data; parallel processing real-time multidimensional sensing data, respectively obtaining a theoretical prediction state and a data fitting state; calculating a model cognitive bias between the theoretical prediction state and the data fitting state, comparing the model cognitive bias with a preset cognitive mismatch threshold, and generating a mismatch discrimination result indicating cognitive mismatch or cognitive consistency; judging the mismatch discrimination result: if it is cognitive mismatch, calculating a physical parameter correction amount based on a preset adaptive gain, updating the physical law model to obtain a corrected theoretical prediction state, and entering the next step; if it is cognitive consistency, directly entering the next step; state fusion: if the mismatch discrimination result is cognitive mismatch, fusing the corrected theoretical prediction state and the data fitting state; if the mismatch discrimination result is cognitive consistency, fusing the theoretical prediction state and the data fitting state to generate a comprehensive evaluation result; comparing the comprehensive evaluation result and the change trend of the comprehensive evaluation result with a preset multi-level risk threshold, and issuing a warning information; The monitoring and warning method provided by the application strictly links the data sensing, double-mode comparison, dynamic correction and risk assessment links in the process; The initial step of the method is to continuously acquire real-time multidimensional sensing data of the submarine cable, and to calculate the real-time heat power of the cable according to the working condition data in the data; The collected data is sent to a double-track parallel processing process: the physical law model outputs a theoretical prediction state according to the physical law, and the data-driven model outputs a data fitting state based on the mode of historical data learning; The key judgment mechanism of the process is to calculate the model cognitive bias between the two states, and compare it with the preset cognitive mismatch threshold to determine whether the physical model is still effective; If it is determined that the cognitive mismatch, the method will seamlessly link to a closed-loop correction step: calculating the physical parameter correction amount and updating the physical law model to obtain a corrected theoretical prediction state closer to reality; After the judgment step, whether the cognition is consistent or not, the method will enter the state fusion step, combining the optimal theoretical prediction and data fitting state to generate a comprehensive evaluation result; The method refers to the preset multi-level risk threshold according to the comprehensive evaluation result and its change trend, and issues accurate warning information to complete a complete monitoring and warning cycle; This method builds a self-correcting closed loop that can dynamically adapt to environmental changes, fundamentally improves the robustness and accuracy of the monitoring and warning system, and ensures the long-term safe and stable operation of offshore wind power facilities in complex and variable marine environments.

[0026] The above merely provides the specific implementation of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An offshore wind power cable monitoring and early warning system based on digital twins, characterized by: include: A data acquisition module is used to obtain real-time multi-dimensional sensor data of the submarine cable, wherein the real-time multi-dimensional sensor data includes real-time operating condition data, and determine the real-time heat generation power of the cable based on the real-time operating condition data; A physical law model module, configured to receive the real-time multi-dimensional sensor data and output a theoretically predicted state; A data driven model module, configured to receive the real-time multi-dimensional sensor data in parallel and output a data fitting state; a bidirectional dynamic coupling module, configured to obtain the theoretical prediction state and the data fitting state, calculate a model cognitive bias, compare the model cognitive bias with a preset cognitive mismatch threshold, and generate a mismatch discrimination result indicating cognitive mismatch or cognitive consistency; a model correction module, configured to, in response to the mismatch determination result being a cognitive mismatch, calculate a physical parameter correction amount based on a preset adaptive gain, update the physical law model module using the physical parameter correction amount, and output a corrected theoretical prediction state; A state evaluation and early warning module is used to fuse the theoretical prediction state and the data fitting state to generate a comprehensive evaluation result when the mismatch discrimination result is cognitive consistency; When the mismatch determination result is cognitive mismatch, fusing the corrected theoretical prediction state output by the model correction module with the data fitting state to generate the comprehensive evaluation result; The comprehensive assessment results and the changing trends of the comprehensive assessment results are compared with the preset multi-level risk thresholds to issue early warning information.

2. The system according to claim 1, wherein: The physical law model is a long short-term memory network model based on the finite element method.

3. The system according to claim 1, wherein: The model cognitive bias is calculated by normalizing the difference between the theoretical prediction state and the data fitting state by the absolute value of the data fitting state.

4. The system according to claim 1, wherein: The cognitive mismatch threshold is determined by statistically analyzing the model cognitive bias generated under historical normal operating conditions and taking the percentile of its distribution.

5. The system according to claim 1, wherein: The model correction module calculates the physical parameter correction amount in the following manner: normalizes the difference between the theoretical prediction state and the data fitting state by the real-time heat generation power of the cable, and multiplies the difference by the preset adaptive gain.

6. The system according to claim 5, characterized in that The real-time operating condition data includes real-time current obtained from the power grid SCADA system, and the real-time heating power of the cable is determined by calculation based on the real-time current and the known resistance of the cable.

7. The system according to claim 1, wherein: The multi-level risk thresholds include concern, warning and danger levels.

8. A digital twin-based offshore wind power cable monitoring and early warning method, characterized in that: include: Acquiring real-time multi-dimensional sensing data of the submarine cable, the real-time multi-dimensional sensing data including real-time operating condition data, and determining the real-time heat generation power of the cable based on the real-time operating condition data; Processing the real-time multi-dimensional sensor data in parallel to obtain a theoretical prediction state and a data fitting state respectively; Calculating a model cognitive deviation between the theoretically predicted state and the data-fitted state, comparing the model cognitive deviation with a preset cognitive mismatch threshold, and generating a mismatch discrimination result indicating cognitive mismatch or cognitive consistency; Determine the mismatch determination result: if it is a cognitive mismatch, calculate the physical parameter correction amount based on the preset adaptive gain, update the physical law model to obtain the corrected theoretical prediction state, and proceed to the next step; If the cognition is consistent, go directly to the next step; Performing state fusion: if the mismatch discrimination result is cognitive mismatch, fusing the revised theoretical prediction state with the data fitting state; If the mismatch determination result is cognitive consistency, then fusing the theoretical prediction state with the data fitting state to generate a comprehensive evaluation result; The early warning information is issued based on the comparison of the comprehensive assessment results and the changing trends of the comprehensive assessment results with the preset multi-level risk thresholds.

Images