Offshore wind power cable line fault positioning method and system

Abstract

The invention discloses an offshore wind power cable line fault positioning method and system, and relates to the technical field of line fault positioning, and the method comprises the steps: collecting cable distribution parameters and instantaneous data collected by a cable fault point and adjacent point sensor; the equivalent impedance of a fault cable is defined, a unified feature vector is formed, equal-interval division and logarithmic mapping are performed according to the total length of the cable, and the power loss of the discrete length of interval points is calculated by combining the module value of the equivalent impedance. According to the method, the discrete feature matrix is formed by combining the fault feature vector, the mapped power value and the logarithmic change discrete length, the effect of uniformly expressing the time-space characteristics of all fault points is achieved, and the time-space characteristics of all the fault points are uniformly expressed by respectively carrying out collaborative mapping on the logarithmic change discrete length and the power value of the candidate solution. A physical comprehension set matrix after comprehensive target value weighting is constructed, and the effect of unifying the attributes of the cable fault points and physical parameters is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of line fault location, in particular to a method and system for offshore wind power cable line fault location. BACKGROUND

[0002] Offshore wind farms are considered an important direction for future energy structure transformation due to stable wind resources and high power generation efficiency. As an important part of offshore wind power systems, cable lines bear the core task of transmitting power output from wind turbine units to the shore, and their operation reliability is crucial to the stability and economy of the entire wind farm. Since the submarine cable is usually laid in a complex seabed environment, it not only needs to withstand long-term electrical load pressure, but also is affected by water flow erosion, temperature changes, mechanical action and other factors, which increases the risk of cable failure. Although existing fault location methods based on electrical characteristics, sensor data analysis and signal reflection time calculation can solve the problem of fault detection to some extent, these technologies still have limited efficiency and insufficient positioning accuracy when applied in complex marine environments. Existing fault location technologies for offshore wind power cable lines mainly focus on electrical parameter measurement and reflection signal analysis. Some technologies use sensors to detect changes in voltage and current characteristics at the fault point and combine specific mathematical models to estimate the fault location. However, due to the complexity of the cable laying area environment, it is difficult to accurately determine faults in short distance areas. The signal resolution in remote areas also lacks sufficient detection sensitivity due to factors such as amplitude attenuation. In addition, methods based on signal propagation time delay calculation can be used for fault location within a certain range, but their single reliance on time difference estimation makes it difficult to handle the problem of multi-frequency signal aliasing between cable breakpoints, resulting in large deviations in fault point location. It is also difficult to fully consider the complex effects of power loss on fault region effects. SUMMARY

[0003] In view of the above existing problems, the present application is proposed.

[0004] Therefore, the present application provides a method and system for offshore wind power cable line fault location to solve the problem of limited environmental complexity of the cable laying area, difficulty in accurately determining faults in short distance areas, lack of sufficient detection sensitivity in signal resolution in remote areas due to factors such as amplitude attenuation, and difficulty in fully considering the complex effects of power loss on fault region effects.

[0005] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a method for offshore wind power cable line fault location, comprising: Collecting cable distribution parameters and instantaneous data collected by cable fault point and adjacent point sensors; Defining the equivalent impedance of the fault cable and forming a unified feature vector, dividing the total length of the cable at equal intervals and performing logarithmic mapping, calculating the power loss of the discrete length of the interval point in combination with the modulus value of the equivalent impedance, and performing logarithmic mapping to comprehensively determine the discrete feature matrix; Defining the discrete feature matrix as a candidate solution set, and according to the novelty evaluation value and matching degree of historical solutions, constructing a comprehensive objective function to dynamically weight the relative importance of the two objectives, introducing the Pareto optimization rule to check the dominance relationship of each candidate solution, and outputting a descending list according to the dominated solution set to determine the physical solution matrix; According to the initial distribution center and the initial covariance matrix in the physical solution matrix, and using the covariance adaptive evolution strategy CMA-ES algorithm, the distribution center and the covariance matrix are iteratively updated by random sampling, and the next generation of population solutions are output, and the candidate solution with the maximum objective value is selected as the optimal solution.

[0006] As a preferred scheme of the offshore wind power cable line fault location method of the present application, wherein: the unified feature vector is formed, the total length of the cable is divided at equal intervals and logarithmic mapping is performed, the power loss of the discrete length of the interval point is calculated in combination with the modulus value of the equivalent impedance, and logarithmic mapping is performed to comprehensively determine the discrete feature matrix, including, Defining the key characteristics of the fault cable as the equivalent impedance modeled by the damaged resistance, inductance and capacitance; Comprehensive cable time series data, temperature field data, unit length resistance value, unit length inductance value, unit length capacitance value, cable laying depth and time delay data of signal wave reflection on the propagation path to form a unified feature vector; According to the total length of the cable, the length of each interval point is defined, the logarithmic mapping is performed based on the length of each interval point, the corresponding logarithmic transformed distance is calculated, and the power loss of the discrete length of the interval point is calculated in combination with the modulus value of the equivalent impedance; Logarithmic mapping is performed on all segmented power loss values, interpolation is performed on each interval point according to the data of the unified feature vector, and logarithmic changes are performed on the discrete length, temperature field data, cable laying depth and time delay data, and the fault feature vector is formed in combination with the time series data, unit length resistance value, unit length inductance value and unit length capacitance value after the interval point difference value. According to different discrete lengths, a discrete feature matrix is formed in combination with the fault feature vector, the mapped power value and the logarithmic change discrete length.

[0007] As a preferred scheme of the offshore wind power cable line fault positioning method, the dynamic weighting of the two objectives is adjusted by constructing a comprehensive objective function, the Pareto optimization rule is introduced, the domination relationship of each candidate solution is checked, and a descending list is output according to the dominated solution set, According to the discrete characteristics of different discrete lengths in the discrete characteristic matrix as the candidate solution set, and based on the discrete characteristic matrix of the historical data as the historical solution, all the accessed solutions in the search process are included; A metric function is constructed, and the novelty metric is defined according to the Euclidean distance between the candidate solution and the historical solution; The electrical performance target value of each candidate solution is defined as the matching degree of the discrete characteristics of the candidate solution and the actual collected fault signal; The novelty evaluation value and the matching value are normalized respectively, the comprehensive objective function is constructed by searching the candidate solution, and the dynamic weighting mechanism is used to adjust the relative importance of the two objectives in the comprehensive objective function; The Pareto optimization rule is introduced, the domination relationship of each candidate solution is checked, and if the novelty evaluation value and the matching value of the jth candidate solution are greater than or equal to the novelty evaluation value and the matching value of the ith candidate solution, the jth candidate solution is defined as dominating the ith candidate solution, and each candidate solution is traversed, the dominated solution set is retained, the candidate solutions are sorted according to the comprehensive target value of the dominated solution set, and a descending list is output; In combination with the descending list, the logarithmic change discrete length and the power value of the candidate solution are respectively mapped in coordination to construct a physical solution set matrix weighted by the comprehensive target value.

[0008] As a preferred scheme of the offshore wind power cable line fault positioning method, the random sampling iteration is performed to update the distribution center and the covariance matrix by using the covariance adaptive evolution strategy CMA-ES algorithm, including, The weighted mean value is calculated according to the comprehensive target value in the physical solution set matrix as the initial distribution center; The initialization covariance matrix is determined according to the dominated solution set candidate solution of the physical solution set matrix; The initial population is generated by sampling using the initial mean value and the initial covariance matrix, the population number is q, the covariance adaptive evolution strategy CMA-ES algorithm is used to randomly sample the dominated solution set candidate solution to generate new candidate solutions, and the corresponding comprehensive target value is recalculated as the fitness value; Based on the current randomly sampled candidate solution, the mean value and the covariance matrix are updated according to the fitness value, and the updated distribution center; The mutation range of the population solution is gradually contracted according to the updated covariance matrix, and the covariance is dynamically adjusted.

[0009] As a preferred embodiment of the offshore wind power cable line fault location method of the present invention, the step of outputting a new generation population solution includes: Based on the updated distribution center and covariance, a new generation of population solutions is generated by sampling from a multivariate normal distribution.

[0010] In a preferred embodiment of the offshore wind power cable line fault location method of the present invention, the step of stopping iteration and selecting the candidate solution with the largest objective value as the optimal solution includes: When the change in the overall objective value of the population is no longer significant, stop iterating and output the final population solution. Based on the final population solution, select the candidate solution with the largest objective value as the optimal solution for fault location prediction.

[0011] As a preferred embodiment of the offshore wind power cable line fault location method of the present invention, the method for collecting cable distribution parameters and instantaneous data collected by sensors includes: Collect and statistically analyze the physical characteristics of the cable as distributed parameters, including the total length of the cable, resistance per unit length, inductance per unit length, and capacitance per unit length. Based on the fault signals from the cable sensors, instantaneous data collected by the sensors at the cable fault point and adjacent points are acquired, including instantaneous voltage and instantaneous current. The electrical signal waveforms are recorded using high-frequency sensors and stored as time series data. Furthermore, the temperature field distribution along the cable length is obtained using a distributed temperature measurement device.

[0012] Secondly, the present invention provides a fault location system for offshore wind power cable lines, comprising, The data acquisition module is responsible for collecting cable distribution parameters and instantaneous data of historical fault points; The feature processing module defines the equivalent impedance of the cable and generates a unified feature vector. The candidate solution evaluation module generates candidate solutions and evaluates them through novelty and matching analysis. The filtering module checks the dominance relationship using novelty evaluation values ​​and matching values, and retains the dominated solutions. The dynamic optimization module utilizes the physical unset matrix to dynamically optimize the mean and covariance matrices. The optimal solution extraction module selects the optimal solution for fault area location.

[0013] The beneficial effects of this invention are as follows: By combining the fault feature vector, the mapped power value, and the logarithmic variation discrete length to form a discrete feature matrix, the spatiotemporal characteristics of all fault points are uniformly expressed. By co-mapping the logarithmic variation discrete length and power value of the candidate solutions, a physical solution set matrix after weighting the comprehensive target value is constructed, which unifies the cable fault point attributes with physical parameters. By using multiple iterations of the covariance matrix and combining the dynamic weight feedback of the distribution center and fitness target value, the efficiency of candidate solution screening is improved. By integrating the discrete length, power value, and fault feature vector multi-dimensional attributes in the population search process through the optimal solution, a transparent analysis method with data-driven core is directly provided for the final fault location. This allows the fault location to not only accurately indicate the fault point but also present the fault characteristics, supporting subsequent maintenance and problem optimization processes. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Fig. 1 This is a flowchart illustrating the fault location method for offshore wind power cable lines in Example 1.

[0016] Fig. 2 This is a schematic diagram of the offshore wind power cable line fault location system in Example 1. Detailed Implementation

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0018] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0019] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0020] Example 1, referring toFigs. 1-2 This is the first embodiment of the present invention, which provides a method for locating faults in offshore wind power cable lines, including the following steps: S1 collects cable distribution parameters and instantaneous data collected by sensors at cable fault points and adjacent points; Preferably, the data includes cable distribution parameters and instantaneous data collected by sensors, including: Collect and statistically analyze the physical characteristics of the cable as distributed parameters, including the total length of the cable, resistance per unit length, inductance per unit length, and capacitance per unit length. Based on the fault signals from the cable sensors, instantaneous data collected by the sensors at the cable fault point and adjacent points are acquired, including instantaneous voltage and instantaneous current. The electrical signal waveforms are recorded using high-frequency sensors and stored as time series data. Furthermore, the temperature field distribution along the cable length is obtained using a distributed temperature measurement device.

[0021] By collecting and statistically analyzing the physical characteristics of the cable, including total cable length, resistance per unit length, inductance per unit length, and capacitance per unit length, the cable material properties and structural parameters were fully quantified. This provides an accurate distributed parameter basis for fault location, enabling the electrical properties of the fault point to be combined with material properties to establish a dynamic response model that truly reflects the physical characteristics of the fault area. By collecting high-frequency data of instantaneous voltage and current from historical fault points and adjacent sensor points, as well as stored time-series signal waveforms, and combining this with temperature field data along the cable length obtained by a distributed temperature measurement device, a multi-dimensional data representation of the fault point was formed. This improved the ability to capture the changing trends of the fault's electrical characteristics. At the same time, the impact of environmental conditions on the electrical performance of the fault point was explained using real-time temperature field data.

[0022] S2, define the equivalent impedance of the faulty cable and form a unified characteristic vector. Divide the cable into equal intervals according to the total length and perform logarithmic mapping. Combine the magnitude of the equivalent impedance to calculate the power loss of the discrete length of the interval point and perform logarithmic mapping. Comprehensively determine the discrete characteristic matrix. Preferably, a unified feature vector is formed. The cable is divided into equal intervals based on its total length and logarithmically mapped. The power loss at discrete intervals is calculated using the magnitude of the equivalent impedance, and logarithmically mapped again. The discrete feature matrix is ​​then comprehensively determined, including... The key characteristics of a faulty cable are defined by modeling its damaged resistance, inductance, and capacitance as equivalent impedance, expressed as:

[0023] in Represents the equivalent impedance. Represents the resistance value per unit length. Representing complex units, used to describe the influence of inductance and capacitance on the phase of a signal in alternating current. Represents angular frequency, derived from system frequency. The conversion is represented as: , C represents the inductance value per unit length, and C represents the capacitance value per unit length. By integrating cable time series data, temperature field data, resistance value per unit length, inductance value per unit length, capacitance value per unit length, cable laying depth, and time delay data of signal wave reflection on the propagation path, a unified feature vector is formed. The signal wave reflection time delay data is represented as follows:

[0024]

[0025] in d represents the signal wave reflection time delay, d represents the transmission distance, and v represents the wave propagation speed. The cable is divided into equal intervals based on its total length, and the length of each interval is defined as follows:

[0026] in This represents the discrete length of the i-th interval point. The total length of the cable is represented by n, and the discrete points are represented by n. Based on the length of each interval point, a logarithmic mapping is performed, and the corresponding logarithmically transformed distance is calculated, expressed as:

[0027] in The logarithmic distance space represents the discrete length of the i-th interval point. The transformed logarithmic distance space effectively compresses the long-range segment (far-field region) while improving the resolution of the short-range segment (near-field region), thereby enhancing the ability to accurately locate fault points in the near range. The power loss at the discrete length of the interval point is calculated by combining the magnitude of the equivalent impedance, and is expressed as:

[0028]

[0029] in This represents the equivalent impedance magnitude from the starting point to the i-th interval point, representing the discrete length. This represents the power loss at the discrete length of the i-th interval point. Indicates the current value. This represents the interval inductance with a discrete length from the starting point to the i-th interval point. Let represent the interval capacitance with a discrete length from the starting point to the i-th interval point. This represents the interval resistance with a discrete length from the starting point to the i-th interval point; Logarithmically mapping all segmented power loss values ​​reduces the range of power variation and makes the far-field region more uniform and analyzable, as shown below:

[0030] in This represents the mapped power value; For each interval point, interpolation is performed on the data of the unified feature vector, and logarithmic transformation is performed on the discrete length, temperature field data, cable laying depth and time delay data. The fault feature vector is composed of the time series data after the interval point difference, the resistance value per unit length, the inductance value per unit length and the capacitance value per unit length. Combining the fault feature vector, the mapped power value, and the logarithmic variation discrete length, a discrete feature matrix is ​​constructed based on different discrete lengths, as follows:

[0031] in Represents the discrete characteristic matrix. Let represent the fault feature vector of the i-th discrete length.

[0032] By defining the key characteristics of the faulty cable, the damaged resistance, inductance, and capacitance are modeled as equivalent impedance, achieving the effect of comprehensively normalizing the changes in micro-parameters into a single index. This facilitates the quantitative analysis of electrical changes in cross-regional faults. At the same time, by directly associating the power loss characteristics with the equivalent impedance, the transmission characteristics of the fault point can be unified with the overall cable model, thereby avoiding the problem of the disconnect between single-point characteristics and the transmission network relationship. By defining the signal wave reflection time delay, the transmission distance is combined with the wave propagation speed, establishing a physical and geometric mapping relationship of time delay, achieving the effect of providing time-domain error control for fault location. At the same time, it is directly associated with the depth of the cable laying path, enhancing the spatial accuracy of fault range identification. By dividing the total cable length into equal intervals and defining the discrete length of each interval point, the fault location space is unified into equal areas, achieving the effect of standardizing the fault distance in complex cable laying scenarios. By combining the modulus of equivalent impedance to calculate the power loss of the discrete length of the interval point, the effect of unified analysis of local faults and overall cable loss characteristics is achieved. At the same time, it directly reflects the current operating status and power change range of the fault point, improving the ability to analyze the changing trend of the operating medium of the fault point. By performing a logarithmic mapping on the power loss values ​​of all segments, the range of power variation is further compressed, and the uniformity of power value distribution in time and space is enhanced, which makes it easier to conduct global analysis of power variation trends in the far field. By combining the dynamic fusion of electrical, physical and environmental data of discrete points, the effect of simultaneously capturing the horizontal (spatial distribution) and vertical (temporal variation) of cable faults is achieved, thus enabling the fault location analysis to be finely quantified under multi-dimensional data conditions. By combining fault feature vectors, mapped power values, and logarithmically varying discrete lengths to form a discrete feature matrix, the spatiotemporal characteristics of all fault points are uniformly expressed. This enables the discrete region fault characteristic matrix to cover the global statistical results of long-distance cable fault intervals, solving the problem of local isolation in fault analysis and providing complete technical support for comprehensive fault location and fault point nature determination.

[0033] S3 defines the discrete feature matrix as the candidate solution set, and constructs a comprehensive objective function to dynamically weight and adjust the relative importance of the two objectives based on the novelty evaluation value and matching degree of the historical solutions. Pareto optimization rule is introduced to check the dominance relationship of each candidate solution. Based on the dominated solution set, a descending sequence list is output to determine the physical solution set matrix. Preferably, a comprehensive objective function is constructed to dynamically weight and adjust the relative importance of the two objectives. The Pareto optimization rule is introduced, and a dominance relationship check is performed on each candidate solution. A descending sequence list is output based on the dominated solution set, including... The candidate solution set is based on discrete features of different discrete lengths in the discrete feature matrix, and the historical solution is based on the discrete feature matrix of historical data, which includes all visited solutions in the search process. Construct a metric function, defining a novelty measure based on the Euclidean distance between candidate solutions and historical solutions, expressed as:

[0034]

[0035] in Let represent the novelty evaluation value of the discrete feature candidate solution of the i-th discrete length, and k represent the nearest historical solutions of the candidate solution, determined based on historical experience. Indicates candidate solutions Interpreting History The Euclidean distance, where m represents the total number of discrete features. Let p represent the p-th eigenvalue of the discrete feature of the i-th discrete length in the candidate solution. The p-th eigenvalue represents the discrete feature of the j-th discrete length in the historical solution. For each candidate solution, the electrical performance target value is defined as the degree of matching between the discrete characteristics of the candidate solution and the actual acquired fault signal, expressed as:

[0036]

[0037] in This represents the voltage prediction of the candidate solution, where I represents the instantaneous current at the fault point. The voltage collected at historical fault points. Indicates candidate solutions The matching value; The calculated novelty evaluation value and matching value are normalized respectively. A comprehensive objective function is constructed by searching for candidate solutions, and a dynamic weighting mechanism is used to adjust the relative importance of the two objectives in the comprehensive objective function. This ensures that the emphasis on novelty and performance at different stages of the solution adapts to the actual optimization requirements, as expressed in:

[0038]

[0039] in Represents dynamic weighting factors. This represents the normalized novelty score. This represents the normalized matching value. This represents the overall target value. and These represent the standard deviations of the matching value and the novelty evaluation value, respectively. The Pareto optimization rule is introduced to check the dominance relationship for each candidate solution. It is defined that if the novelty evaluation value and matching value of the j-th candidate solution are both greater than or equal to the novelty evaluation value and matching value of the ith candidate solution, then the j-th candidate solution is defined as dominating the ith candidate solution. Each candidate solution is traversed, the dominated solution set is retained, and the candidate solutions are sorted according to the comprehensive objective value of the dominated solution set, and a descending sequence list is output. By combining the descending sequence list, the logarithmic variation discrete length and power value of the candidate solutions are co-mapped to construct a physical solution set matrix weighted by the comprehensive objective value, as follows:

[0040]

[0041]

[0042] in The cooperative mapping value represents the discrete length of the logarithmic change of the i-th candidate solution. The cooperative mapping value represents the power value of the i-th candidate solution. Represents the set matrix of the physical distribution.

[0043] By using discrete features of different discrete lengths in the discrete feature matrix as candidate solution sets, the effect of uniformly modeling the cable fault point attributes distributed in a discrete region is achieved. At the same time, the discretization of the candidate solution enhances the hierarchical clarity of the search data, enabling the algorithm to efficiently process the discrete parameters of the fault location problem. By using a discrete feature matrix based on historical data as a historical solution, and including all visited solutions in the search process, the effect of associating real-time fault candidate solutions with historical data is achieved. This enables the search process to improve the accuracy of evaluating candidate solutions by leveraging historical information, while forming a dynamically updated search database. This allows for the accumulation of effective experience for fault location over long-term operation, and further optimizes the efficiency of candidate solution search through a feedback mechanism. By constructing a metric function and defining a novelty metric based on the Euclidean distance between candidate solutions and historical solutions, the novelty of candidate solutions is automatically identified, improving the coverage of the global search. By defining electrical performance target values ​​for each candidate solution and based on the degree of matching between the discrete features of the candidate solutions and the actual collected fault signals, the consistency between the extracted electrical characteristics of the fault point and the collected measured data is achieved. By normalizing the calculated novelty evaluation value and matching value, and constructing a dynamically weighted comprehensive objective function by searching for candidate solutions, the goal of balancing the importance of novelty and performance objectives is achieved. By introducing the Pareto optimization rule and checking the dominance relationship, unimportant candidate solutions are further reduced, and the search range is effectively narrowed to the high-precision candidate solution region. By outputting a descending sequence list, the goal of establishing the priority of candidate solutions and forming a global ranking is achieved, avoiding the efficiency loss caused by scattered search. By co-mapping the discrete length and power value of the logarithmic change of the candidate solutions, a physical solution set matrix after weighting the comprehensive target value is constructed. This achieves the effect of unifying the cable fault point attributes with physical parameters, so that the spatial information, electrical information and power attributes of the fault point can be reflected in a unified relation matrix at the same time, avoiding the complexity of the fault location process caused by the dispersion of attributes.

[0044] S4. Calculate the initial distribution center and initial covariance matrix based on the physical set matrix, and use the covariance adaptive evolution strategy CMA-ES algorithm to perform random sampling and iterative updates of the distribution center and covariance matrix, output the next generation population solution, stop the iteration and select the candidate solution with the largest objective value as the optimal solution. Preferably, the CMA-ES algorithm, a covariance adaptive evolution strategy, is used to perform random sampling and iterative updates of the distribution center and covariance matrix, including: The weighted mean of the comprehensive objective value in the physical settling matrix is ​​used as the initial distribution center, expressed as:

[0045] in represents the initial distribution center, and q represents the number of candidate solutions in the dominated solution set; Based on the candidate solutions of the dominant solution set of the physical solution set matrix, the initial covariance matrix is ​​determined, expressed as:

[0046] in Denotes the initial covariance matrix. Indicates transpose calculation; Using the initial mean and initial covariance matrix, an initial population is generated by sampling, with a population size of q. The covariance adaptive evolution strategy CMA-ES algorithm is used to randomly sample the candidate solutions of the dominated solution set to generate new candidate solutions, and the corresponding comprehensive objective value is recalculated as the fitness value. Based on the candidate solutions sampled randomly, the mean and covariance matrices are updated according to the fitness values. The updated distribution centers are represented as follows:

[0047]

[0048] in This indicates an update to the target value ranking weight. This represents the combined objective value of the candidate solution updated in the r-th iteration. This represents the number of candidate solutions in the updated dominated solution set. and Let represent the distribution centers at the (t+1)th and tth updates, respectively; The range of variation of the population solution gradually shrinks according to the update of the covariance matrix, and the covariance is dynamically adjusted, which is expressed as:

[0049] in and as well as Let r+1, r-1 (the previous update), and r (the current update) represent the covariance matrices, respectively, and c represent the learning rate. This can be expressed as: .

[0050] By using the weighted mean calculated from the comprehensive objective value in the physical solution set matrix as the initial distribution center, the optimization process is ensured to start the search from high-quality regions. The initial covariance matrix is ​​determined by the candidate solutions of the dominant solution set of the physical solution set matrix, which improves the rationality of the search range of the optimization algorithm and its high adaptability to the characteristics of the problem. At the same time, the calculation method of the covariance matrix reduces the loss of information in weakly correlated regions in the early search process. By using initial mean and initial covariance matrix sampling to generate an initial population, the optimization process can avoid focusing on a single region and losing global perspective in the early stages. By updating the mean and covariance matrix based on the current randomly sampled candidate solutions and fitness values, it can not only reflect the true importance of the solution corresponding to the ranking weight of the target value, but also gradually evolve the mean from the centrality of candidate solutions to the center of the possible solution region of the problem through a step-by-step replacement method, thereby reducing the global deviation of the target search. By utilizing multiple iterative adjustments to the covariance matrix, combined with dynamic weight feedback of the distribution center and fitness target value, the efficiency of candidate solution screening is improved. This allows the search solution to quickly converge to high-value candidate regions in each generation of optimization. At the same time, the weight relationship gradually ignores low-fitness regions, reducing the time and space consumption of invalid regions and promoting efficient optimization of the solution space and accurate acquisition of solution content for fault location problems.

[0051] Furthermore, output a new generation of population solutions, including, A new generation of population solutions is generated by sampling from a multivariate normal distribution based on the updated distribution center and covariance. As the population iterates, the distribution center (mean) and range (covariance matrix) are adjusted, and the discrete length gradually converges in the search space. Similarly, after updates, the power value is gradually optimized by combining the mean and fitness weights. Fault feature updates are based on dynamic optimization results and are adjusted in conjunction with the discrete length and power value, as shown below:

[0052] in This represents the next generation of population solutions, i.e., the updated physical solution set matrix, where rows represent candidate solutions and columns represent the parameter values ​​of the solutions. and as well as Let represent the fault feature vector, discrete length, and power value of the updated i-th candidate solution, respectively.

[0053] By using multivariate normal distribution sampling based on the updated distribution center and covariance matrix to generate a new generation of population solutions, the new generation of candidate solutions can inherit the distribution trend of the previous generation population, while also having the ability to explore local randomness. This not only improves the coverage of the solution space, but also avoids getting caught in the local optimum problem in the global convergence process, achieving the effect of steadily fitting the power characteristics of the fault region. This allows the fault features to maintain the flexibility of local convergence in each generation, and to reflect the coupling of the physical and electrical characteristics of the fault point in the real scenario through the co-optimization of other solution parameters, thus forming complete analytical support for the multidimensional characteristics of complex faults. By updating the new generation of population solutions into a new physical solution set matrix, the effect of directly reflecting the design changes of the optimization process in the solution space is achieved. The rows of the new physical solution set matrix represent candidate solutions, and the columns represent the multidimensional parameters of the solutions. This structure integrates the logic of each generation of optimization results, ensuring that the evolution trend of candidate solutions can intuitively reflect the global adjustment law of the target value.

[0054] Furthermore, the iteration stops, and the candidate solution with the largest objective value is selected as the optimal solution, including: When the change in the overall objective value of the population is no longer significant, stop iterating and output the final population solution. Based on the final population solution, select the candidate solution with the largest objective value as the optimal solution for fault location prediction.

[0055] Spatial location of the fault point is achieved by using the discrete length in the optimal solution, which directly maps the distance information of the fault point to the physical distribution of the cable. The discrete length of the optimal solution reflects the local accuracy of the spatial location of the fault point during the search process. The power value in the optimal solution describes the electrical influence range of the fault point, which quantitatively displays the energy loss and severity of the fault point. It can also further indicate the processing priority required for the fault area. The fault feature vector in the optimal solution fully expresses the electrical attributes of the fault point, which combines the instantaneous voltage, current and delay time of the fault point with physical parameters. By expanding the fault properties from spatial location to a multi-dimensional description of the regional electrical state, a comprehensive diagnostic result of the fault point in a dynamic state is formed. By integrating the discrete length, power value, and fault feature vector multidimensional attributes during the population search process through the optimal solution, a transparent analysis method with data-driven core is directly provided for the final fault location. This enables fault location to not only accurately indicate the fault point, but also to present the fault characteristics, supporting subsequent maintenance and problem optimization processes.

[0056] This embodiment also provides a fault location system for offshore wind power cable lines, including, The data acquisition module is responsible for collecting cable distribution parameters and instantaneous data of historical fault points; The feature processing module defines the equivalent impedance of the cable and generates a unified feature vector. The candidate solution evaluation module generates candidate solutions and evaluates them through novelty and matching analysis. The filtering module checks the dominance relationship using novelty evaluation values ​​and matching values, and retains the dominated solutions. The dynamic optimization module utilizes the physical unset matrix to dynamically optimize the mean and covariance matrices. The optimal solution extraction module selects the optimal solution for fault area location.

[0057] This embodiment also provides a computer device applicable to the offshore wind power cable line fault location method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the offshore wind power cable line fault location method proposed in the above embodiment.

[0058] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0059] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for locating faults in offshore wind power cable lines as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0060] In summary, this invention achieves a unified expression of the spatiotemporal characteristics of all fault points by combining fault feature vectors, mapped power values, and logarithmic variation discrete lengths to form a discrete feature matrix. By co-mapping the logarithmic variation discrete lengths and power values ​​of candidate solutions, a physical solution set matrix weighted by the comprehensive target value is constructed, achieving the unification of cable fault point attributes with physical parameters. By utilizing multiple iterations of the covariance matrix and combining dynamic weight feedback of distribution center and fitness target values, the efficiency of candidate solution screening is improved. By integrating the multi-dimensional attributes of discrete length, power value, and fault feature vectors during the population search process through the optimal solution, a transparent analysis method with data-driven core is directly provided for the final fault location. This allows fault location to not only accurately indicate the fault point but also present the fault characteristics, supporting subsequent maintenance and problem optimization processes.

[0061] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for locating faults in offshore wind power cable lines, characterized in that, include: Collect cable distribution parameters and instantaneous data collected by sensors at cable fault points and adjacent points; Define the equivalent impedance of the faulty cable and form a unified characteristic vector. Divide the cable into equal intervals according to the total length and perform logarithmic mapping. Calculate the power loss of the discrete length of the interval point by combining the magnitude of the equivalent impedance and perform logarithmic mapping. Determine the discrete characteristic matrix in a comprehensive manner. The discrete feature matrix is ​​defined as the candidate solution set. Based on the analysis of historical solutions, the novelty evaluation value and matching degree are analyzed. A comprehensive objective function is constructed to dynamically adjust the relative importance of the two objectives. The Pareto optimization rule is introduced to check the dominance relationship of each candidate solution. Based on the dominated solution set, a descending sequence list is output to determine the physical solution set matrix. The initial distribution center and initial covariance matrix are calculated based on the physical set matrix. Then, the distribution center and covariance matrix are updated by random sampling using the Covariance Adaptive Evolution Strategy (CMA-ES) algorithm. The next generation of population solutions is output, and the iteration stops. The candidate solution with the largest objective value is selected as the optimal solution.

2. The method for locating faults in offshore wind power cable lines as described in claim 1, characterized in that: The unified feature vector is composed of equal intervals based on the total cable length, logarithmically mapped, and the power loss at discrete intervals is calculated using the magnitude of the equivalent impedance, followed by logarithmic mapping. This comprehensive approach determines the discrete feature matrix, including... The key characteristics of a faulty cable are defined by modeling its damaged resistance, inductance, and capacitance as equivalent impedance. By integrating cable time series data, temperature field data, resistance value per unit length, inductance value per unit length, capacitance value per unit length, cable laying depth, and time delay data of signal wave reflection on the propagation path, a unified feature vector is formed. The cable is divided into equal intervals based on its total length. The length of each interval is defined. A logarithmic mapping is performed based on the length of each interval to calculate the corresponding logarithmically transformed distance. The power loss of the discrete length of the interval is calculated by combining the magnitude of the equivalent impedance. Logarithmically map the power loss values ​​of all segments, interpolate the data of each interval point based on the unified feature vector, and logarithmically transform the discrete length, temperature field data, cable laying depth, and time delay data. Combine the time series data after the interval point difference, the resistance value per unit length, the inductance value per unit length, and the capacitance value per unit length to form a fault feature vector. By combining the fault feature vector, the mapped power value, and the logarithmic variation discrete length, a discrete feature matrix is ​​formed according to different discrete lengths.

3. The method for locating faults in offshore wind power cable lines as described in claim 2, characterized in that: The construction of a comprehensive objective function dynamically weights and adjusts the relative importance of the two objectives. A Pareto optimization rule is introduced, and a dominance relationship check is performed on each candidate solution. Based on the dominated solution set, a descending sequence list is output, including... The candidate solution set is based on discrete features of different discrete lengths in the discrete feature matrix, and the historical solution is based on the discrete feature matrix of historical data, which includes all visited solutions in the search process. Construct a metric function and define a novelty metric based on the Euclidean distance between candidate solutions and historical solutions; For each candidate solution, the electrical performance target value is defined as the degree of matching between the discrete characteristics of the candidate solution and the actual acquired fault signal; The novelty evaluation value and matching value are normalized respectively. A comprehensive objective function is constructed by searching for candidate solutions, and a dynamic weighting mechanism is used to adjust the relative importance of the two objectives in the comprehensive objective function. The Pareto optimization rule is introduced to check the dominance relationship for each candidate solution. It is defined that if the novelty evaluation value and matching value of the j-th candidate solution are both greater than or equal to the novelty evaluation value and matching value of the ith candidate solution, then the j-th candidate solution is defined as dominating the ith candidate solution. Each candidate solution is traversed, the dominated solution set is retained, and the candidate solutions are sorted according to the comprehensive objective value of the dominated solution set, and a descending sequence list is output. By combining the descending sequence list, the logarithmic variation discrete length and power value of the candidate solutions are co-mapped to construct a physical solution set matrix weighted by the comprehensive objective value.

4. The method for locating faults in offshore wind power cable lines as described in claim 3, characterized in that: The CMA-ES algorithm, which utilizes a covariance adaptive evolution strategy, performs random sampling and iterative updates to the distribution center and covariance matrix, including: The weighted mean is calculated based on the comprehensive objective value in the physical settling matrix and used as the initial distribution center. The initial covariance matrix is ​​determined based on the candidate solutions of the dominant solution set of the physical solution set matrix; Using the initial mean and initial covariance matrix, an initial population is generated by sampling, with a population size of q. The covariance adaptive evolution strategy CMA-ES algorithm is used to randomly sample candidate solutions in the dominated solution set to generate new candidate solutions, and the corresponding comprehensive objective value is recalculated as the fitness value. Based on the candidate solutions of the current random sampling, update the mean and covariance matrix according to the fitness value, and update the distribution center; The range of variation of the population solution gradually shrinks according to the update of the covariance matrix, and the covariance is dynamically adjusted.

5. The method for locating faults in offshore wind power cable lines as described in claim 4, characterized in that: The output is a new generation of population solution. include, Based on the updated distribution center and covariance, a new generation of population solutions is generated by sampling from a multivariate normal distribution.

6. The method for locating faults in offshore wind power cable lines as described in claim 5, characterized in that: The step of stopping iteration and selecting the candidate solution with the largest objective value as the optimal solution includes: When the change in the overall objective value of the population is no longer significant, stop iterating and output the final population solution. Based on the final population solution, select the candidate solution with the largest objective value as the optimal solution for fault location prediction.

7. The method for locating faults in offshore wind power cable lines as described in claim 1, characterized in that: The data acquisition cable distribution parameters and the instantaneous data acquired by the sensors are described. include, Collect and statistically analyze the physical characteristics of the cable as distributed parameters, including the total length of the cable, resistance per unit length, inductance per unit length, and capacitance per unit length. Based on the fault signals from the cable sensors, instantaneous data collected by the sensors at the cable fault point and adjacent points are acquired, including instantaneous voltage and instantaneous current. The electrical signal waveforms are recorded using high-frequency sensors and stored as time series data. Furthermore, the temperature field distribution along the cable length is obtained using a distributed temperature measurement device.

8. A fault location system for offshore wind power cable lines, based on the fault location method for offshore wind power cable lines according to any one of claims 1 to 7, characterized in that: include, The data acquisition module is responsible for collecting cable distribution parameters and instantaneous data of historical fault points; The feature processing module defines the equivalent impedance of the cable and generates a unified feature vector. The candidate solution evaluation module generates candidate solutions and evaluates them through novelty and matching analysis. The filtering module checks the dominance relationship using novelty evaluation values ​​and matching values, and retains the dominated solutions. The dynamic optimization module utilizes the physical unset matrix to dynamically optimize the mean and covariance matrices. The optimal solution extraction module selects the optimal solution for fault area location.

Images