Offshore wind power plant network health state assessment method considering dynamic weight adjustment

Abstract

The invention discloses an offshore wind plant network health state evaluation method considering dynamic weight adjustment in the technical field of new energy power generation. The method comprises the following steps: calculating a plurality of indexes for evaluation; respectively calculating the comprehensive weight of each index according to a subjective and objective weighting method; global sensitivity is calculated in real time, and the global sensitivity represents the overall tension degree of the system after the abnormal operation state of the offshore wind plant is diffused through the electrical coupling relation between the submarine cables under the current submarine cable topological structure; when the global sensitivity does not exceed the threshold value, calculating and evaluating the network health state of the offshore wind power plant according to the comprehensive weight of each index; when the global sensitivity exceeds a threshold value, the comprehensive weight of each index is adjusted according to the global sensitivity; and calculating and evaluating the health state of the offshore wind power plant network according to the new weight of each index. According to the method, the problem that the network health state of the offshore wind plant is difficult to accurately evaluate due to high dynamic nature and strong uncertainty of operation of the offshore wind plant can be solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of new energy power generation, and particularly relates to a method for evaluating the health state of an offshore wind farm network considering dynamic weight adjustment. BACKGROUND

[0002] The large-scale development of offshore wind farms makes efficient monitoring and accurate health assessment of their operating states crucial, which not only concerns system safety, stability and economic benefits, but also is a core requirement for sustainable development of the industry. However, the diverse types, large quantities, and frequently fluctuating operating conditions of offshore wind farm equipment, combined with the difficulties and high costs of manual maintenance, bring great difficulties to the assessment work.

[0003] More importantly, the operation has high dynamicity and strong uncertainty: abnormal operating conditions occur frequently; the topology of the internal aggregation network often needs to be dynamically adjusted for optimization or emergency. This uncertainty makes it difficult for traditional static assessment methods that rely on fixed models to be applicable. Meanwhile, the intermittency and volatility of wind power generation lead to highly non-stationary key monitoring indicators, and the multi-dimensional indicators such as machinery, electricity, and environment that need to be integrated for assessment have different dimensions, different sensitivities, and different importance levels that dynamically change with the operating state and topology structure. The great difference and dynamic correlation of such multi-source heterogeneous data make comprehensive analysis extremely complex. SUMMARY

[0004] The purpose of the present application is to overcome the deficiencies in the prior art and provide a method for evaluating the health state of an offshore wind farm network considering dynamic weight adjustment, which solves the problem of high dynamicity and strong uncertainty of offshore wind farm operation, making it difficult to accurately assess the health state of the offshore wind farm network.

[0005] To solve the above technical problems, the present application is implemented by using the following technical solutions:

[0006] The present application provides a method for evaluating the health state of an offshore wind farm network considering dynamic weight adjustment, comprising: obtaining electrical performance data of the offshore wind farm and calculating a plurality of indicators for evaluation; calculating the comprehensive weight of each indicator according to the subjective and objective weighting method; calculating the global sensitivity in real time according to the electrical coupling relationship and voltage distribution between the sea cables of the offshore wind farm, which represents the overall tension of the system after the abnormal operating state of the offshore wind farm spreads through the electrical coupling relationship between the sea cables under the current sea cable topology structure; comparing the global sensitivity with the global sensitivity threshold value: when the global sensitivity does not exceed the global sensitivity threshold value, evaluating the health state of the offshore wind farm network according to the comprehensive weight of each indicator; when the global sensitivity exceeds the global sensitivity threshold value, adjusting the comprehensive weight of each indicator according to the global sensitivity to obtain the new weight of each indicator; and evaluating the health state of the offshore wind farm network according to the new weight of each indicator.

[0007] The offshore wind farm network health state evaluation method considering dynamic weight adjustment, the offshore wind farm electrical performance data is obtained, and a plurality of indexes for evaluation are calculated, including: selecting the first index to the eighth index in turn: the cable transmission distance, the power collection system loss rate, the unit cable pressure drop, the booster station load rate, the grid connection point voltage qualified rate, the harmonic distortion rate, the wind power penetration rate and the wind power fluctuation; the original eight indexes are calculated; the original third index, the fifth index, the sixth index and the seventh index are respectively subjected to single transformation, so that the original eight indexes are transformed into single indexes representing overall characteristics; wherein, the unit cable pressure drop of the third index is transformed into a single index, which includes: after calculating the final attention coefficient of each section of the cable and its neighbor cable by using the pre-trained graph attention network, the final attention coefficient of each section of the cable is calculated based on the final attention coefficient of each section of the cable and its neighbor cable, and the unit pressure drop of each section of the cable is weighted; each single index is normalized to obtain eight indexes for evaluation.

[0008] The offshore wind farm network health state evaluation method considering dynamic weight adjustment, the original third index, the fifth index, the sixth index and the seventh index are respectively subjected to single transformation, including:

[0009] The third single index, the average value of the sum of the unit pressure drop weighted values of all cables is taken as the index Y3 of the unit cable pressure drop X3 after single transformation, and the calculation formula is:

[0010] ,

[0011] In the formula, is the unit pressure drop weighted value of the kth section of the cable; M represents the total number of cable sections;

[0012] ,

[0013] In the formula, X 3,k is the unit pressure drop of the kth section of the cable; α k is the final attention coefficient of the kth section of the cable;

[0014] ,

[0015] In the formula, V s,k , V e,k are the head and tail voltages of the kth section of the cable; L k is the length of the kth section of the cable;

[0016] The fifth single index, the grid connection point voltage qualified rate X5 is subjected to single transformation to obtain the index Y5, and the calculation formula is:

[0017] ,

[0018] In the formula, N b X represents the total number of grid connection points. 5,u It is the voltage qualification rate of the u-th grid connection point;

[0019] ,

[0020] ,

[0021] In the formula, for Voltage at the u-th grid connection point at time u, V N The voltage is the nominal voltage of the power grid, and T is the preset total time for calculating the voltage qualification rate at the grid connection point. It is a function that represents 0 or 1;

[0022] The sixth single-indicator, harmonic distortion rate X6, is calculated using the following formula after single-indicator transformation:

[0023] ,

[0024] In the formula, X 6,u Let w be the harmonic distortion rate at the u-th grid connection point. 6,u The weighting coefficient for the u-th grid connection point;

[0025] ,

[0026] In the formula, I 1,u I is the effective value of the fundamental current at the u-th grid connection point; h,u This represents the effective value of the h-th harmonic current at the u-th grid connection point. It is the sum of the squares of the effective values ​​of all harmonic currents at the u-th grid connection point;

[0027] ,

[0028] In the formula, L u,tr Let be the line distance from the u-th grid connection point to the step-up transformer in the substation;

[0029] The seventh single-indicator, wind power penetration rate X7, is transformed into index Y7, and the calculation formula is as follows:

[0030] ,

[0031] In the formula, N s X represents the total number of samples taken within the preset evaluation period. 7,t. Let X7 be the value of the wind power penetration rate in the t-th sampling.

[0032] ,

[0033] In the formula, P f,p (t.) represents the actual output of the p-th wind turbine at the t.th sampling time; N f P represents the total number of wind turbines in the wind farm. l,j (t.) represents the real-time power of the j-th load at the t.th sampling time; N l Total load;

[0034] The process of normalizing each individual indicator yields eight indicators for evaluation, including:

[0035] For the first to eighth singularity indicators: X1, X2, Y3, X4, Y5, Y6, Y7, and X8, normalization was performed respectively. The eight indicators after normalization are as follows:

[0036] , This represents the r-th index after normalization.

[0037] The aforementioned method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment, wherein the final attention coefficients of each submarine cable segment and its neighboring submarine cables calculated using a pre-trained graph attention network include: the pre-trained graph attention network is a two-layer graph attention network; the process of the two-layer graph attention network calculating the final attention coefficients of each submarine cable segment and its neighboring submarine cables is as follows: pre-construction of graph and feature initialization: each submarine cable segment is taken as a graph node, and the physical connection between submarine cables is taken as an edge, where physical connection refers to connection to the same electrical equipment; the terminal voltage difference of each submarine cable segment is calculated as a one-dimensional initial feature and normalized; the first layer uses multi-head attention, which expands and divides the normalized terminal voltage difference of each submarine cable segment into multiple sub-features, corresponding to the input of each attention head, calculates the attention coefficients with neighboring submarine cables in each attention head, aggregates the information, and outputs high-dimensional features to the second layer; the second layer uses single-head attention, which calculates the final attention coefficients of each submarine cable segment and its neighboring submarine cables based on the high-dimensional features output by the first layer;

[0038] The final attention coefficient of each submarine cable segment is calculated based on the final attention coefficients of each segment and its neighboring submarine cables. The final attention coefficient α of the k-th submarine cable segment, calculated by the second layer of the two-layer graph attention network, is then averaged with the final attention coefficients of its neighboring submarine cables. k The average calculation refers to summing the final attention coefficients of the k-th submarine cable segment and its neighboring submarine cables, and then dividing by the total number of neighboring submarine cables of the k-th submarine cable segment.

[0039] The aforementioned method for assessing the health status of offshore wind farm networks, which takes into account dynamic weight adjustments, involves calculating the comprehensive weight of each indicator based on a subjective and objective weighting method, including:

[0040] Based on the subjective weighting method, the subjective weights of each indicator are calculated, including:

[0041] Establish indicator order relationship: Based on the subjective experience of experts, the 8 indicators are ranked from high to low importance to form indicator order relationship B, and the correspondence between the 8 indicators and the importance level indicators after the importance ranking is established.

[0042] B={b1, b2, b3, b4, b5, b6, b7, b8},

[0043] In the formula, the elements b1, b2, b3, b4, b5, b6, b7, and b8 of B are importance level indicators after importance ranking, indicating that the importance of the corresponding indicators decreases in order, with b1 being the most important and b8 being the least important.

[0044] Determine the importance ratio of adjacent importance level indicators in the indicator order relationship: Based on the subjective experience of experts, set the importance ratio of adjacent importance level indicators in the indicator order relationship as: R2, R3, R4, R5, R6, R7, R8; representing the importance of b1 to b2, b2 to b3, b3 to b4, b4 to b5, b5 to b6, b6 to b7, and b7 to b8, respectively.

[0045] Calculate the subjective weights of each indicator:

[0046] Calculate the weight of the least important metric b8 :

[0047] ,

[0048] The weights of other importance level indicators are calculated recursively:

[0049] ,

[0050] ,

[0051] ,

[0052] ,

[0053] ,

[0054] ,

[0055] ,

[0056] to These represent the weight coefficients of the most important indicator b1 to the least important indicator b8, respectively.

[0057] Based on the correspondence between the eight indicators and the importance level indicators ranked by importance, the weights of the importance level indicators are assigned to the eight indicators accordingly, serving as the subjective weights of the corresponding indicators. , ;

[0058] Based on the objective weighting method using entropy weighting, the objective weights of each indicator are calculated, including:

[0059] Calculate the proportion of each normalized index of each wind farm relative to its category:

[0060] Get N w The eight indicators after normalization of wind farms, and the proportion P of the r-th indicator after normalization of wind farm a to the same category. r,a The calculation formula is:

[0061] ,

[0062] In the formula, ; This represents the r-th index after normalization for the a-th wind farm; N represents w Sum of the r-th index after normalization of each wind farm;

[0063] Determine the entropy weights of each normalized index:

[0064] The entropy value e of the r-th index after normalization r The calculation formula is:

[0065] ,

[0066] Calculate the objective weights of each indicator:

[0067] The objective weight of the r-th indicator The calculation formula is:

[0068] ;

[0069] Based on the subjective and objective weights of each indicator, the comprehensive weight of each indicator is calculated, including:

[0070] The comprehensive weight W of the r-th indicator r Calculation formula:

[0071] ,

[0072] In the formula, This represents the objective weight of the r-th indicator; This represents the subjective weight of the r-th indicator.

[0073] The aforementioned method for assessing the health status of offshore wind farm networks, which takes into account dynamic weight adjustment, includes the real-time calculation of global sensitivity based on the electrical coupling relationship and voltage distribution between submarine cables in the offshore wind farm, comprising:

[0074] The topology sensitivity characteristics of each cable segment are calculated by weighting and aggregating the voltage offset and voltage spatial gradient of each neighboring cable segment using the final attention coefficients of each segment and its neighboring cable; the topology sensitivity characteristic F of the k-th cable segment is also calculated. k The calculation formula is:

[0075] ,

[0076] In the formula, N(k) represents the set of neighboring submarine cables of the k-th submarine cable segment. The neighboring submarine cables of the k-th submarine cable segment refer to other submarine cables that are connected to the same electrical equipment as the k-th submarine cable segment. This represents the final attention coefficient between the k-th submarine cable segment and its neighboring submarine cable i, calculated by the pre-trained graph attention network. This represents the voltage offset of the neighboring submarine cable i of the k-th segment of the submarine cable; This represents the spatial voltage gradient of neighboring submarine cable i;

[0077] Based on the topological sensitivity characteristics of each segment of the submarine cable, the global sensitivity is calculated; the formula for calculating the global sensitivity F is:

[0078] ,

[0079] In the formula, M represents the total number of submarine cable segments.

[0080] The aforementioned method for assessing the health status of offshore wind farm networks, which takes into account dynamic weight adjustments, includes the voltage offset of the neighboring submarine cable i of the k-th segment of the submarine cable. The calculation formula is:

[0081] ,

[0082] In the formula, This represents the voltage offset at the beginning of neighboring submarine cable i. This indicates the voltage offset at the tail end of neighboring submarine cable i;

[0083] ,

[0084] In the formula, V s,i V represents the voltage at the beginning of neighboring submarine cable i; N Indicates the nominal voltage of the power grid;

[0085] ,

[0086] In the formula, V e,i This represents the voltage at the tail end of neighboring submarine cable i;

[0087] Voltage spatial gradient of the neighboring submarine cable i The calculation formula is:

[0088] ,

[0089] In the formula, This represents the spatial gradient of the voltage at the beginning of neighboring submarine cable i; This represents the spatial gradient of the tail voltage of neighboring submarine cable i;

[0090] ,

[0091] In the formula, N(i,s) * N(i,s) represents the total number of neighboring submarine cables at the beginning of neighboring submarine cable i; N(i,s) represents the set of neighboring submarine cables at the beginning of neighboring submarine cable i. It is a neighboring submarine cable at the head end of the neighboring submarine cable i, representing The submarine cable i-end is connected to the same electrical equipment as the neighboring cable; Indicates neighboring submarine cable average voltage, , Indicates neighboring submarine cable The voltage at the beginning of the circuit, Indicates neighboring submarine cable The voltage at the tail end;

[0092] ,

[0093] In the formula, N(i,e) * N(i,e) represents the total number of neighboring submarine cables at the end of neighboring submarine cable i; N(i,e) represents the set of neighboring submarine cables at the end of neighboring submarine cable i. It is a neighboring submarine cable at the end of the neighboring submarine cable i, representing The end of the neighboring submarine cable is connected to the same electrical equipment. Indicates neighboring submarine cable average voltage, , Indicates neighboring submarine cable The voltage at the beginning of the circuit, Indicates neighboring submarine cable The voltage at the tail end.

[0094] The aforementioned method for assessing the health status of offshore wind farm networks, which takes into account dynamic weight adjustments, includes the following steps before comparing the global sensitivity with the global sensitivity threshold:

[0095] The global sensitivity threshold ε is adjusted based on wind power penetration rate, and the formula for calculating the global sensitivity threshold ε is as follows:

[0096] ,

[0097] In the formula, X7 is the wind power penetration rate;

[0098] ,

[0099] In the formula, P f,p (t) represents the actual output of the p-th wind turbine at time t; N f P represents the total number of wind turbines in the wind farm. l,j (t) represents the real-time power of the j-th load at time t, N l This represents the total load.

[0100] The aforementioned method for assessing the health status of offshore wind farm networks, which takes into account dynamic weight adjustments, includes assessing the health status of offshore wind farm networks based on the comprehensive weights of various indicators, including:

[0101] The evaluation function value is calculated based on the health status evaluation function of the offshore wind farm;

[0102] The health level of the corresponding offshore wind farm is determined by matching and searching the evaluation function value in the preset health level assessment table.

[0103] The health status assessment function Y of the offshore wind farm * The calculation formula is:

[0104] ,

[0105] In the formula, W r Let r be the comprehensive weight of the r-th indicator; Let be the r-th index after normalization.

[0106] The aforementioned method for assessing the health status of offshore wind farm networks, which takes into account dynamic weight adjustment, involves adjusting the comprehensive weights of each indicator based on global sensitivity to obtain new weights for each indicator, including:

[0107] The weight increase of the first and second indicators is calculated based on global sensitivity.

[0108] The formula for calculating the weight increase is as follows:

[0109] ,

[0110] In the formula, F represents global sensitivity;

[0111] Based on the magnitude of the weight increase, the combined weight of the first and second indicators is increased, while the combined weight of the other indicators is decreased, resulting in the new weights for each indicator.

[0112] The formula for updating the overall weight of the first indicator is:

[0113] ,

[0114] In the formula, W1 represents the new weight of the first indicator; W1 represents the overall weight of the first indicator.

[0115] The formula for updating the overall weight of the second indicator is:

[0116] ,

[0117] In the formula, W1 represents the new weight of the second indicator; W2 represents the overall weight of the second indicator.

[0118] The formula for updating the combined weights of the third to eighth indicators is as follows:

[0119] ,

[0120] In the formula, For the first New weights for each indicator; For the first A comprehensive indicator of all indicators. ;

[0121] Assess the health status of offshore wind farm networks based on the new weights of each indicator;

[0122] The update formula for the health status assessment function of offshore wind farms is as follows:

[0123] ,

[0124] In the formula, For the new health status assessment function of offshore wind farms; As the new weight for the r-th indicator, Let r be the normalized index. .

[0125] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0126] The present invention provides a method for assessing the health status of offshore wind farm networks that incorporates dynamic weight adjustment. By selecting multiple indicators and calculating the comprehensive weight of each indicator using a subjective and objective weighting method, the method assesses offshore wind farms from multiple perspectives and calculates the global sensitivity, which characterizes the overall stress level of the system. By comparing the global sensitivity with a threshold, the method determines whether the offshore wind farm network health status is calculated and assessed based on the comprehensive weight of each indicator when the system has a large safety margin. When the system is operating close to the safety boundary, the method adjusts the comprehensive weight of each indicator according to the global sensitivity, so that the new weight of each indicator is more suitable for the assessment needs when the global sensitivity exceeds the threshold. This method solves the problem that the highly dynamic and uncertain nature of offshore wind farm operation makes it difficult to accurately assess the health status of offshore wind farm networks.

[0127] This invention comprehensively measures the health status of offshore wind farms from multiple perspectives using eight indicators: submarine cable transmission distance, power collection system loss rate, unit submarine cable voltage drop, substation load rate, grid connection point voltage qualification rate, harmonic distortion rate, wind power penetration rate, and wind power fluctuation. Furthermore, based on the data characteristics of each of the eight indicators, a calculation method adapted to each indicator is designed, unifying the eight indicators into a single indicator that can characterize the overall features of the offshore wind farm. Each single indicator is then normalized, enabling the processed eight indicators to be used for offshore wind farm evaluation. This solves the problem of difficulty in comprehensive analysis caused by the significant differences and dynamic correlations of multi-source heterogeneous data.

[0128] Among them, when the third indicator, unit submarine cable voltage drop, is transformed into a single indicator, a pre-trained graph attention network is used to calculate the final attention coefficients of each submarine cable segment and its neighboring submarine cables, so that the unit submarine cable voltage drop of a single submarine cable segment can be transformed into a single indicator that characterizes the overall features.

[0129] This invention calculates the subjective weight of each indicator through a subjective weighting method, calculates the objective weight of each indicator through an objective weighting method, and then calculates the comprehensive weight of each indicator by combining the subjective and objective weights, so that the eight indicators of this invention can be weighted in a balanced and accurate manner.

[0130] In order to adapt to the changes in system electrical characteristics under different wind power output levels and avoid false triggering or missed triggering of global sensitivity guidance weight adjustment, this invention designs a global sensitivity threshold that is adjusted according to wind power penetration rate.

[0131] To make the new weights of each indicator more suitable for the evaluation needs when the global sensitivity exceeds the threshold, this invention increases the weights of the two indicators that are highly correlated with the global sensitivity, namely the submarine cable transmission distance and the power collection system loss rate, while adaptively reducing the weights of other indicators. This allows the evaluation method to better adapt to the highly dynamic and uncertain nature of offshore wind farm operation. Attached Figure Description

[0132] Figure 1 This is a schematic diagram of a method for assessing the health status of offshore wind farm networks that takes into account dynamic weight adjustment, according to Embodiment 1 of the present invention.

[0133] Figure 2 This is a schematic diagram of the graph construction process of a graph attention network in an offshore wind farm network health status assessment method that takes into account dynamic weight adjustment, according to Embodiment 1 of the present invention. Detailed Implementation

[0134] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations thereof. In the absence of conflict, the embodiments and technical features in the embodiments can be combined with each other.

[0135] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0136] Example 1:

[0137] This embodiment introduces a method for assessing the health status of offshore wind farm networks that takes into account dynamic weight adjustment, such as... Figure 1 As shown, it includes:

[0138] S1: Acquire electrical performance data of offshore wind farms and calculate multiple metrics for evaluation;

[0139] S2: Calculate the comprehensive weight of each indicator according to the subjective and objective weighting method;

[0140] S3: Based on the electrical coupling relationship and voltage distribution between submarine cables in the offshore wind farm, calculate the global sensitivity in real time. The global sensitivity characterizes the overall severity of the abnormal operation of the offshore wind farm under the current submarine cable topology after it spreads through the electrical coupling relationship between the submarine cables.

[0141] S4: Compare the global sensitivity with the global sensitivity threshold:

[0142] When the global sensitivity does not exceed (i.e., is not greater than) the global sensitivity threshold, the health status of the offshore wind farm network is assessed based on the comprehensive weight of each indicator.

[0143] When the global sensitivity exceeds (i.e., is greater than) the global sensitivity threshold, the comprehensive weight of each indicator is adjusted according to the global sensitivity to obtain the new weight of each indicator; the health status of the offshore wind farm network is assessed according to the new weight of each indicator.

[0144] Step S1 includes:

[0145] S11: Selection of Multi-Dimensional Indicators

[0146] When designing the health status assessment index system for offshore wind farms, the first to eighth indicators selected are, in order: submarine cable transmission distance, collector system loss rate, unit submarine cable voltage drop, substation load rate, grid connection point voltage qualification rate, harmonic distortion rate, wind power penetration rate, and wind power fluctuation. Submarine cable transmission distance, collector system loss rate, and unit submarine cable voltage drop can be used as assessment parameters for line operation status. These three indicators quantify line performance from three dimensions: spatial distribution, energy loss, and voltage variation. Regarding the operational characteristics of offshore wind farms, the substation load rate reflects equipment carrying capacity, the grid connection point voltage qualification rate reflects power supply stability, and the harmonic distortion rate measures the purity of power quality. Wind power penetration rate and wind power fluctuation are two key indicators specifically used to describe the characteristics of wind power integration. The explanations and calculation formulas for each indicator are as follows:

[0147] Submarine cable transmission distance: Specifically refers to the maximum straight-line distance from the farthest wind turbine to the substation, calculated using the following formula:

[0148] ,

[0149] In the formula, X1 is the transmission distance of the submarine cable; N f This represents the total number of wind turbines in the wind farm. This represents the maximum value calculated for a single wind turbine among all the turbines in the wind farm; (x p y p Let (x0, y0) be the coordinates of the p-th wind turbine, and (x0, y0) be the coordinates of the booster station. In this embodiment, a Cartesian coordinate system is constructed based on the geographical location of the wind farm.

[0150] Current collector system loss rate: In actual operation analysis, researchers mainly select the theoretical part of the comprehensive line loss rate for study. The formula for calculating the current collector system loss rate X2 is:

[0151] ,

[0152] In the formula, X2 is the loss rate of the collector system; I k R represents the effective current value of the k-th segment of the submarine cable; k T represents the resistance of the k-th segment of the submarine cable. run T represents the wind farm's operating time; M represents the total number of submarine cable segments; E represents the wind farm's operating time. runTotal electricity generation within the region.

[0153] Unit submarine cable voltage drop X3: This quantifies the voltage drop per unit length of the submarine cable, calculated using the following formula:

[0154] ,

[0155] In the formula, V s,k V e,k These are the voltages at the beginning and end of the k-th segment of the submarine cable, respectively; L k Let k be the length of the submarine cable segment.

[0156] Substation load rate: This refers to the ratio of the actual operating power of the substation to its rated capacity. The formula for calculating the substation load rate x 4 is as follows:

[0157] ,

[0158] In the formula, S real S represents the actual operating power of the step-up transformer in the substation; rated This indicates the rated capacity of the step-up transformer. The indicators calculated using these two parameters can intuitively reflect the load level of the entire offshore wind farm, i.e., the intensity of the load.

[0159] Grid connection point voltage qualification rate: refers to the percentage of time during which the grid connection point voltage of a wind farm remains within the allowable range of national standards. The calculation formula for grid connection point voltage qualification rate X5 is as follows:

[0160] ,

[0161] ,

[0162] In the formula, for Voltage at grid connection point, V N The voltage is the nominal voltage of the power grid, and T is the preset total time for calculating the voltage qualification rate at the grid connection point. It is a function that represents 0 or 1.

[0163] Harmonic distortion rate: defined as the ratio of the sum of the squares of the effective values ​​of all harmonic currents at the grid connection point to the effective value of the fundamental current. The result is presented as a percentage. The formula for calculating harmonic distortion rate X6 is:

[0164] ,

[0165] In the formula, I1 is the effective value of the fundamental current at the grid connection point; I h The effective value of the h-th harmonic current at the grid connection point; This is the sum of the squares of the effective values ​​of all harmonic currents at the grid connection point. This indicator quantifies the degree of distortion in the current waveform at the grid connection point and directly reflects the impact of power electronic equipment on power quality.

[0166] Wind power penetration rate: This indicator is used to assess the intensity of new energy integration, representing the proportion of wind power generation in the total load of offshore power plants. The formula for calculating wind power penetration rate X7 is:

[0167] ,

[0168] In the formula, P f,p (t) represents the actual output of the p-th wind turbine at time t; P l,j (t) represents the real-time power of the j-th load at time t, N l Total load;

[0169] Wind power fluctuation: The intensity of fluctuation in wind power generation is quantified and defined as the ratio of the standard deviation of total output to the mean. The formula for calculating wind power fluctuation x8 is:

[0170] ,

[0171] In the formula, N s This represents the total number of samples taken within the preset evaluation period. The sum of the total power of the wind farm sampled within the preset evaluation period is divided by the total number of samplings. In this embodiment, the preset evaluation period is 24 hours. Let t be the total power of the wind farm in the t-th sampling.

[0172] S12: Obtain the raw data for the eight indicators and transform each indicator into a unique indicator representing the overall characteristics; Heterogeneous indicator conversion:

[0173] In the health status assessment system of offshore wind farms, the scientific classification and multi-dimensional integration of the indicator system are crucial for comprehensively reflecting the system's operational status. Based on the characteristics of the indicators, a classification system is established, dividing the evaluation parameters into three categories. The first category consists of single-indicators: submarine cable transmission distance (X1), power collection system loss rate (X2), and substation load rate (X4). The second category consists of distributed indicators, including unit submarine cable voltage drop (X3), grid connection point voltage qualification rate (X5), and harmonic distortion rate (X6). The third category consists of dynamic characteristic indicators, involving wind power penetration rate (X7) and wind power fluctuation (X8). This classification method, based on the different mechanisms by which each indicator functions in grid operation, can systematically reflect the power supply quality, network losses, and renewable energy access characteristics of offshore wind farms.

[0174] Single-value indicators characterize the overall features of a certain dimension with a single numerical value; distributed indicators describe the discrete characteristics of physical quantities in spatial or temporal dimensions; dynamic indicator indicators quantify the volatility and nonlinear changes of a system over time. To construct a unified health assessment model, distributed and dynamic indicators need to be transformed into single-value indicators. This transformation brings the indicators to the same scale, enhancing the interpretability and cross-sectional comparability of the assessment results and providing theoretical support for refined modeling of power system health status assessment.

[0175] S121: Transforming a distributed indicator into a single indicator:

[0176] Distributed metrics typically describe the operational characteristics of a localized area within a transformer substation. These metrics may correspond to the operational status of a single line or node. When a comprehensive assessment of the overall operational status of a transformer substation is required, these distributed metrics must be converted into integrated metrics to directly reflect the overall characteristics of the substation.

[0177] For processing the unit submarine cable voltage drop (X3), a GAT (Graph Attention Network) is used for transformation, including: calculating the final attention coefficients of each submarine cable segment and its neighboring submarine cables using a pre-trained graph attention network; calculating the final attention coefficients of each submarine cable segment based on these final attention coefficients; and weighting the unit voltage drop of each submarine cable segment; constructing a graph attention network with submarine cables as nodes and the connections between submarine cables as edges. Based on this, the single-head attention coefficients calculated by the second layer of GAT are used to achieve a simplified transformation of the unit voltage drop of the submarine cable. Transformation steps ZH1 to ZH4 are as follows:

[0178] ZH1: Calculate the final attention coefficient between any two connected submarine cable segments:

[0179] The pre-trained graph attention network used in this embodiment is a two-layer graph attention network. The process of calculating the final attention coefficient between any two connected submarine cable segments using the pre-trained two-layer GAT includes the following steps:

[0180] XS1: Graph Construction and Feature Initialization

[0181] Each segment of the submarine cable is treated as a node in the graph, and the physical connections between the cables are treated as edges. Here, a physical connection refers to a submarine cable being connected to the same electrical equipment; for example... Figure 2The diagram shows the process of constructing a 5-segment submarine cable. The top of the first segment of the submarine cable is defined as the end closest to the substation. Taking the second segment as an example, the neighboring submarine cables of the second segment are the first, third, and fourth segments, with a total of 3 neighboring submarine cables. Among them, the neighboring submarine cables at the beginning of the second segment are the first and third segments, with a total of 2 neighboring submarine cables at the beginning of the second segment. The neighboring submarine cable at the end of the second segment is the fourth segment, with a total of 1 neighboring submarine cable at the end of the second segment.

[0182] The voltage difference at the ends of each submarine cable segment is calculated as an initial feature. The initial feature is one-dimensional, and the initial feature of each submarine cable segment is normalized.

[0183] XS2: First Layer: Multi-head Attention Feature Mechanism

[0184] XS21: Linear mapping: The one-dimensional voltage difference of each segment of the submarine cable is extended into a high-dimensional feature through a preset mapping weight matrix. In this embodiment, the high-dimensional feature is 64-dimensional.

[0185] Multi-head attention parallel computation; in this embodiment, it involves 4 heads of attention:

[0186] The high-dimensional features of each submarine cable segment are evenly divided into four sub-features according to their dimensions, which are used as the input features of the four heads respectively. In this embodiment, each sub-feature is 16-dimensional, and each head is input with a 16-dimensional sub-feature.

[0187] Taking the calculation of the k-th segment of the submarine cable as an example, the calculation process for other submarine cables is the same as that for the k-th segment:

[0188] The sub-features of the four heads of the k-th segment of the submarine cable input are as follows: ;

[0189] XS22: For each head:

[0190] A sub-feature of the k-th submarine cable segment is concatenated with the sub-feature corresponding to one of its neighboring submarine cables i to obtain a 32-dimensional feature. The concatenated 32-dimensional feature is weighted using the first-layer weighting vector preset for each head, so that the concatenated 32-dimensional feature is mapped to a single real value. This real value is output as the attention coefficient of the first-layer head of the k-th submarine cable segment and its neighboring submarine cable i through a linear rectified function with leakage.

[0191] The attention coefficient of the first layer and nth head of the k-th segment of the submarine cable and its neighboring submarine cable i. The calculation formula is:

[0192] ,

[0193] In the formula, N(k) represents the set of neighboring submarine cables of the k-th submarine cable segment. The neighboring submarine cables of the k-th submarine cable segment refer to other submarine cables that are connected to the same electrical equipment as the k-th submarine cable segment. This represents the sub-feature of the nth header of the k-th segment of the submarine cable input. The dimensions are 16-dimensional. ; This represents the sub-feature of the nth header of the neighboring submarine cable i. The size is 16 dimensions; || represents a concatenation operation. The dimensions are 32-dimensional; This represents the pre-defined weighted vector for the nth head of the first layer. The size is 32-dimensional; T represents the transpose operation; LeakyReLU is a non-linear activation function, i.e., a leaky linear rectified function that provides a small gradient for negative inputs;

[0194] XS23: Normalization coefficient:

[0195] The attention coefficients of the first layer of the k-th submarine cable and its neighboring cable i are normalized to between 0 and 1 using the softmax function:

[0196] In the formula, is the normalized attention coefficient of the first layer of the k-th segment of the submarine cable and its neighboring submarine cable i;

[0197] XS24: Weighted Aggregation

[0198] The attention coefficients of the first layer of each neighboring submarine cable and the normalized k-th segment of the submarine cable are weighted and aggregated with the sub-features of each neighboring submarine cable, and then the summation result is processed by the Sigmoid function to obtain the weighted aggregation result of the first layer of the k-th segment of the submarine cable. In this embodiment, the weighted aggregation result of the first layer is 16-dimensional.

[0199] The weighted aggregation result of the nth head of the k-th segment of the submarine cable The calculation formula is:

[0200] ;

[0201] In the formula, σ represents the Sigmoid function;

[0202] XS25: Four head features stitched together:

[0203] Output characteristics of the first layer of the k-th segment of the submarine cable The calculation formula is:

[0204] ,

[0205] In the formula, The above formula represents the concatenation of the features of the four heads to obtain a 64-dimensional output feature. The above formula represents the concatenation of the weighted aggregation results of the four heads of the k-th segment of the submarine cable to restore the high dimension before uniform division by dimension, and to obtain the output feature of the first layer of the k-th segment of the submarine cable. In this embodiment, the four 16-dimensional weighted aggregation results are concatenated by dimension to restore the 64-dimensional output feature of the first layer of the k-th segment of the submarine cable.

[0206] XS3: Second layer: Single-head attention feature:

[0207] The calculation of the attention coefficient for the second layer is similar to the calculation of the single-head attention coefficient for the first layer:

[0208] Input features: 64-dimensional features of each segment of the submarine cable output from the first layer.

[0209] Calculate the attention coefficient:

[0210] The output characteristics of the first layer of the k-th segment of the submarine cable Output characteristics of its neighboring submarine cable i in the first layer After splicing, the pre-set second-layer weighted vector α2 is multiplied with the spliced ​​features of the second layer, so that the spliced ​​features of the second layer are mapped to a single real value. This real value is then output as the attention coefficient of the second layer between the k-th submarine cable segment and its neighboring submarine cable i through a leaky linear rectified function.

[0211] The attention coefficient of the second layer between the k-th segment of the submarine cable and its neighboring submarine cable i The calculation formula is:

[0212] ,

[0213] In the formula, This represents the output characteristics of the first layer of the k-th segment of the submarine cable; α1 represents the output feature of the first layer of the neighboring submarine cable i; α2 represents the preset weighted vector of the second layer; || represents the concatenation operation; T represents the transpose operation;

[0214] Normalization coefficient:

[0215] The softmax function is used to normalize the second-layer attention coefficients of the k-th submarine cable segment and its neighboring cable i to a range of 0 to 1, which are then used as the final attention coefficients between the k-th submarine cable segment and its neighboring cable i. :

[0216] ;

[0217] ZH2: Calculates the final attention coefficient of each submarine cable segment based on the final attention coefficients of each segment and its neighboring submarine cables, including:

[0218] The final attention coefficient α of the k-th submarine cable is obtained by averaging the final attention coefficients of the k-th submarine cable calculated by the second layer of the pre-trained two-layer GAT with its neighboring submarine cables. k The average is calculated by summing the final attention coefficients of the k-th submarine cable segment and its neighboring submarine cables, and then dividing the sum by the total number of neighboring submarine cables of the k-th submarine cable segment.

[0219] ZH3: The unit voltage drop of each section of the submarine cable is weighted by the final attention coefficient of each section of the submarine cable to obtain the weighted value of the unit voltage drop of each section of the submarine cable.

[0220] The final attention coefficient α of the k-th segment of the submarine cable k The unit voltage drop of the k-th segment of the submarine cable is weighted to obtain the weighted value of the unit voltage drop of the k-th segment; the weighted value of the unit voltage drop of the k-th segment of the submarine cable. The calculation formula is:

[0221] ,

[0222] In the formula, X 3,k Let k be the unit voltage drop of the k-th segment of the submarine cable. ;

[0223] ZH4: Finally, a global aggregation is performed, and the average of the weighted sums of the unit voltage drop values ​​of all submarine cables is taken as the index Y3 after the unit submarine cable voltage drop X3 is transformed by the singleness factor. The calculation formula is as follows:

[0224] .

[0225] When dealing with the parameter of grid connection point voltage qualification rate (X5), it is assumed that all grid connection points in offshore wind farms have the same weight, and the difference in importance between grid connection points is not considered. The calculation formula for the index Y5 after the single-integrity transformation of the grid connection point voltage qualification rate X5 is as follows:

[0226] ,

[0227] In the formula, N b X represents the total number of grid connection points. 5,u It is the voltage qualification rate of the u-th grid connection point;

[0228] ,

[0229] ,

[0230] In the formula, for Voltage at the u-th grid connection point at time u, V N The voltage is the nominal voltage of the power grid, and T is the preset total time for calculating the voltage qualification rate at the grid connection point. It is a function that represents 0 or 1.

[0231] For harmonic distortion rate (X6), the grid connection point near the transformer has a greater impact on harmonic propagation and can be assigned a higher weight; the farther the grid connection point is from the transformer, the smaller its impact on subsequent lines, and the lower its weight. The formula for calculating the local weight of this index is defined as follows:

[0232] ,

[0233] In the formula, L u,tr Let be the line distance from the u-th grid connection point to the step-up transformer in the substation.

[0234] The formula for converting X6 to Y6 is as follows:

[0235] ,

[0236] In the formula, X 6,u Let w be the harmonic distortion rate at the u-th grid connection point. 6,u The weighting coefficient for the u-th grid connection point;

[0237] ,

[0238] In the formula, I 1,u I is the effective value of the fundamental current at the u-th grid connection point; h,u This represents the effective value of the h-th harmonic current at the u-th grid connection point. It is the sum of the squares of the effective values ​​of all harmonic currents at the u-th grid connection point. The index of X6 after the single-integer transformation is represented by Y6.

[0239] S122: Dynamic characteristic indicators are transformed into single-function indicators:

[0240] The average value of X7 within the preset assessment period is taken to reflect the average wind power penetration level. The average wind power penetration level within the preset assessment period is used as the index Y7 after the wind power penetration rate X7 has undergone a single-value transformation. The calculation formula is as follows:

[0241] ,

[0242] In the formula, N s X represents the total number of samples taken within the preset evaluation period. 7,t. Let X7 be the value of the wind power penetration rate in the t-th sampling.

[0243] ,

[0244] In the formula, P f,p (t.) represents the actual output of the p-th wind turbine at the t.th sampling time; N f P represents the total number of wind turbines in the wind farm. l,j(t.) represents the real-time power of the j-th load at the t.th sampling time; N l Total load;

[0245] Since the wind power fluctuation X8 already meets the requirement of uniformity, no further conversion is needed.

[0246] The eight unique indicators X1, X2, Y3, X4, Y5, Y6, Y7, and X8 mentioned above are all data obtained and calculated in real time from the power grid system.

[0247] S13: Normalize each individual indicator to obtain eight indicators for evaluation; Normalize multi-source indicators:

[0248] When assessing the health status of offshore wind farms, relevant indicators can be divided into three categories: positive indicators, negative indicators, and range indicators. Positive indicators generally represent benefits, with higher values ​​indicating a closer approximation to the ideal state. Negative indicators typically reflect costs, with lower values ​​indicating a more ideal state. Range indicators have specific ranges, with values ​​closer to the midpoint of that range indicating a closer approximation to the ideal state. In constructing an offshore wind farm health status assessment indicator system, negative indicators include: submarine cable transmission distance (X1), collector system loss rate (X2), unit submarine cable voltage drop (X3), harmonic distortion rate (X6), and wind power fluctuation (X8). Positive indicators include grid connection point voltage qualification rate (X5). Range indicators include: substation load rate (X4) and wind power penetration rate (X7). A low X4 indicates insufficient equipment utilization, a high X4 indicates overload risk, a low X7 indicates low clean energy utilization, and a high X7 may jeopardize grid stability. Among them, the submarine cable transmission distance (X1) and the unit submarine cable voltage drop (X3) are dimensional indicators, while the other 6 indicators are dimensionless indicators.

[0249] Because the evaluation indicators lack a unified dimension, direct comparative analysis would lack comparability. To address this issue, this study employs an extreme value processing method to standardize all indicators, transforming the original data to the [0,1] range. Through this data normalization preprocessing, different types of indicators obtain a unified measurement standard.

[0250] For the first to eighth individuality indicators: X1, X2, Y3, X4, Y5, Y6, Y7, and X8, normalization was performed respectively.

[0251] The reverse and forward indices are normalized using the following formula:

[0252] ,

[0253] In the formula, and The indicators X are respectively derived from historical data of wind farms.r1 The minimum and maximum values.

[0254] The interval indicator is normalized using the following formula:

[0255] ,

[0256] In the formula, It is a preset indicator. The optimal operating range should be the midpoint value of the optimal operating range. The optimal operating range of each indicator is determined comprehensively based on power grid security constraints, equipment operating limits, energy consumption policies, and economic objectives. Indicator X in historical data of wind farms r2 Corresponding calculation The maximum value.

[0257] These two formulas eliminate dimensional differences between different indicators. Through standardization, all indicators are uniformly converted to a numerical range of [0,1], making them comparable. When applying the formulas, it is necessary to first obtain the extreme value data for each indicator. For interval-type indicators, the midpoint value of their optimal operating interval needs to be determined.

[0258] The 8 indicators after normalization:

[0259] and , , ,Will and The comprehensive representation is as follows:

[0260] , This represents the r-th index after normalization.

[0261] In step S2, the preliminary weights of multiple indicators are calculated according to the subjective and objective weighting method. The specific steps are as follows:

[0262] S21: Subjective empowerment method;

[0263] The subjective empowerment method effectively integrates subjective judgment and is suitable for the health status assessment system of offshore wind farms. The specific implementation process consists of three stages:

[0264] S211: Establish indicator order relationship: Based on the subjective experience of experts, rank the 8 indicators from high to low importance to form indicator order relationship B, and establish the correspondence between the 8 indicators and the importance level indicators after the importance ranking.

[0265] B={b1, b2, b3, b4, b5, b6, b7, b8},

[0266] In the formula, the elements b1, b2, b3, b4, b5, b6, b7, b8 of B are importance level indicators after importance ranking, indicating that the importance of the corresponding indicators decreases in order, with b1 being the most important and b8 being the least important.

[0267] In one specific embodiment, the correspondence between the eight indicators and the importance level indicators is as follows: The importance ranking of the corresponding indicators is as follows: Of these, the fifth indicator, X5, is the most important, while the first indicator, X1, is the least important.

[0268] S212: Determine the importance ratio of adjacent indicators in the indicator order relationship:

[0269] Based on the subjective experience of experts, the importance ratios of adjacent indicators in the indicator order relationship are set as follows: R2, R3, R4, R5, R6, R7, R8; representing the importance of b1 to b2, b2 to b3, b3 to b4, b4 to b5, b5 to b6, b6 to b7, and b7 to b8, respectively.

[0270] S213: Calculate the subjective weights of each indicator:

[0271] Calculate the weight of the least important metric b8 :

[0272] ,

[0273] The weights of other importance level indicators are calculated recursively:

[0274] ,

[0275] ,

[0276] ,

[0277] ,

[0278] ,

[0279] ,

[0280] ,

[0281] to These represent the weight coefficients of the most important indicator b1 to the least important indicator b8, respectively.

[0282] Based on the correspondence between the eight indicators and the importance level indicators ranked by importance, the weights of the importance level indicators are assigned to the eight indicators accordingly, serving as the subjective weights of the corresponding indicators. ,Right now In one specific embodiment, the correspondence between the eight indicators and the importance level indicators is as follows: The weighting coefficient of the least important indicator b8 is The subjective weight corresponding to the first indicator yes ;

[0283] S22: Objective weighting method based on entropy weighting;

[0284] In the implementation of the entropy weight method, the variability of indicator data directly affects the weight allocation. When the variability of a certain indicator increases, the corresponding entropy weight will decrease, reflecting that the indicator contains more effective information. In the evaluation system, such indicators will play a more significant role, and therefore their weight allocation will be correspondingly increased. The specific implementation process consists of three steps:

[0285] S221: Calculate the proportion of each normalized index of each wind farm relative to the same category:

[0286] Get N w The eight indicators after normalization of wind farms, and the proportion P of the r-th indicator after normalization of wind farm a to the same category. r,a The calculation formula is:

[0287] ,

[0288] In the formula, ; This represents the r-th index after normalization for the a-th wind farm; N represents w Sum of the r-th index after normalization of each wind farm.

[0289] S222: Determine the entropy weights of each normalized index:

[0290] The entropy value e of the r-th index after normalization r The calculation formula is:

[0291] ;

[0292] S223: Calculate the objective weights of each indicator:

[0293] The objective weight of the r-th indicator The calculation formula is:

[0294] ;

[0295] S23: A combined subjective and objective optimization strategy;

[0296] Based on the findings of the two methods described above, combining both approaches can take into account both subjective and objective information, thus improving the rationality of weight allocation. The specific implementation steps are as follows:

[0297] Based on the subjective weights of each indicator calculated by the subjective weighting method and the objective weights of each indicator calculated by the objective weighting method, the comprehensive weights of each indicator are calculated accordingly.

[0298] S231: To make the overall weights as close as possible to the subjective and objective weights, based on the principle of minimizing entropy, the objective function minE is determined as follows:

[0299] The constraints are: ;

[0300] In the formula, W r This represents the overall weight of the r-th indicator; This represents the objective weight of the r-th indicator; This represents the subjective weight of the r-th indicator;

[0301] S232: After optimization using the Lagrange multiplier method, the formula for calculating the comprehensive weight is derived as follows:

[0302] The Lagrange multiplier method transforms a constrained optimization problem into an unconstrained one by introducing the Lagrange multiplier λ. Constructing the Lagrange function... :

[0303] ,

[0304] By analyzing W r By taking the partial derivatives of λ and λ respectively, we can finally obtain W. r :

[0305] Lagrange function with respect to a fixed W r Find the partial derivative:

[0306] ,

[0307] Let the partial derivative If the value is 0, then: ,

[0308] Partial derivative of the Lagrange function with respect to λ: ,

[0309] Let the partial derivative If the value is 0, then: ,

[0310] Will Substitution Solving the equation yields:

[0311] ,

[0312] Finally, the comprehensive weight W of the r-th indicator is obtained. r Calculation formula:

[0313] ,

[0314] In the formula, This represents the objective weight of the r-th indicator; This represents the subjective weight of the r-th indicator;

[0315] Example 2

[0316] Based on the same inventive concept as Embodiment 1, this embodiment introduces a method for assessing the health status of offshore wind farm networks that takes into account dynamic weight adjustment.

[0317] Step S3 calculates the global sensitivity in real time based on the electrical coupling relationship and voltage distribution between submarine cables in the offshore wind farm, including:

[0318] S31: The topology sensitivity characteristics of each submarine cable segment are calculated by weighting and aggregating the voltage offset and voltage spatial gradient of each neighboring submarine cable segment by the final attention coefficient of each submarine cable segment and its neighboring submarine cables.

[0319] Based on the two-layer GAT model, the second layer calculates the final attention coefficient between the k-th submarine cable segment and its neighboring submarine cable i. This implies the strength of the electrical coupling relationship between the k-th submarine cable segment and its neighboring cable i. Dynamically updating the final attention coefficients between each cable segment using an attention mechanism can characterize the differences in importance of different cables within the topology. Specifically, the final attention coefficients between the k-th submarine cable segment and its neighboring cable i... The larger the value, the stronger the electrical correlation between the k-th submarine cable segment and its neighboring cable i, and the more significant the mutual influence of their voltage, power, etc. Therefore, the topology sensitivity characteristic F of the k-th submarine cable segment is defined. k The calculation formula is:

[0320] ,

[0321] In the formula, N(k) represents the set of neighboring submarine cables of the k-th submarine cable segment. The neighboring submarine cables of the k-th submarine cable segment refer to other submarine cables that are connected to the same electrical equipment as the k-th submarine cable segment. This represents the final attention coefficient between the k-th submarine cable segment and its neighboring submarine cable i, calculated in the second layer of the double-layer GAT. Let be the voltage offset of the neighboring submarine cable i of the k-th segment of the submarine cable. It is the voltage spatial gradient of the neighboring submarine cable i, reflecting the steepness of the local voltage distribution;

[0322] Voltage offset of neighboring submarine cable i of segment k of submarine cable The calculation formula is:

[0323] ,

[0324] In the formula, This represents the voltage offset at the beginning of neighboring submarine cable i. This indicates the voltage offset at the tail end of neighboring submarine cable i; in this embodiment, the end of each submarine cable closest to the booster station is the head end, and the other end is the tail end.

[0325] ,

[0326] In the formula, V s,i V represents the voltage at the beginning of neighboring submarine cable i; N Indicates the nominal voltage of the power grid;

[0327] ,

[0328] In the formula, V e,i This represents the voltage at the tail end of neighboring submarine cable i;

[0329] Voltage spatial gradient of neighboring submarine cable i The calculation formula is:

[0330] ,

[0331] In the formula, This represents the spatial gradient of the voltage at the beginning of neighboring submarine cable i; This represents the spatial gradient of the tail voltage of neighboring submarine cable i;

[0332] ,

[0333] In the formula, N(i,s) * N(i,s) represents the total number of neighboring submarine cables at the beginning of neighboring submarine cable i; N(i,s) represents the set of neighboring submarine cables at the beginning of neighboring submarine cable i. It is a neighboring submarine cable at the head end of the neighboring submarine cable i, representing The submarine cable i-end is connected to the same electrical equipment as the neighboring cable; Indicates neighboring submarine cable average voltage, , Indicates neighboring submarine cable The voltage at the beginning of the circuit, Indicates neighboring submarine cable The voltage at the tail end;

[0334] ,

[0335] In the formula, N(i,e) * N(i,e) represents the total number of neighboring submarine cables at the end of neighboring submarine cable i; N(i,e) represents the set of neighboring submarine cables at the end of neighboring submarine cable i. It is a neighboring submarine cable at the end of the neighboring submarine cable i, representing The end of the neighboring submarine cable is connected to the same electrical equipment. Indicates neighboring submarine cable average voltage, , Indicates neighboring submarine cable The voltage at the beginning of the circuit, Indicates neighboring submarine cable The voltage at the tail end;

[0336] S32: Based on the topology sensitivity characteristics of each segment of the submarine cable, calculate the global sensitivity, which characterizes the overall topology sensitivity of the submarine cable. The formula for calculating the global sensitivity F is:

[0337] ,

[0338] In the formula, M represents the total number of submarine cable segments;

[0339] The voltage offset of each submarine cable segment reflects the deviation of its voltage from the reference value, and is a direct result of the system's operating state. The voltage spatial gradient reflects the combined effect of the current cable connection topology and operating state. Both voltage offset and voltage spatial gradient characterize abnormal system operating states. The electrical coupling relationship between submarine cables is reflected by attention coefficients. The topology sensitivity characteristics of each cable segment are dynamically weighted by an attention mechanism based on voltage offset and voltage spatial gradient, achieving joint modeling of topology sensitivity and time-varying operating states. This process transforms the abstract cable connection relationships into quantifiable and calculable attention weights, solving the problem of accurately describing equipment state associations.

[0340] This provides key input for subsequent dynamic weight adjustment, enabling the evaluation method to adaptively adjust the importance of indicators based on changes in topological connectivity and operating status, thereby enhancing the ability to identify dynamic operating states.

[0341] Step S4: Compare the global sensitivity with the threshold to calculate and assess the health status of the offshore wind farm network.

[0342] Global sensitivity characterizes the overall system stress level after an operational anomaly propagates through the electrical coupling between submarine cables under the current submarine cable topology.

[0343] By comparing global sensitivity with a threshold, it is possible to determine whether an offshore wind farm system is approaching the boundary for safe operation.

[0344] If the global sensitivity does not exceed the threshold, it indicates that the offshore wind farm system is operating within the normal margin range. The comprehensive weight of each indicator is calculated and the health status of the offshore wind farm network is evaluated.

[0345] If the global sensitivity exceeds the threshold, it indicates that the system is approaching the safe operating boundary. At this point, the overall weight of each indicator should be adjusted:

[0346] Increase the weight of the first indicator, submarine cable transmission distance, and the second indicator, power collection system loss rate, while adaptively reducing the weight of other indicators to make the evaluation results more sensitive to the deterioration of these two types of indicators.

[0347] This weighting adjustment strategy adjusts the focus of the assessment according to the system's stress level, balancing the comprehensiveness of routine monitoring with the efficiency of response in emergency situations.

[0348] When the global sensitivity exceeds the threshold, the weights of the first and second indicators are increased because:

[0349] A global sensitivity exceeding the threshold indicates that voltage deviation and voltage spatial gradient anomalies have spread to a relatively serious extent in the network. The most likely causes are twofold: one is related to the submarine cable transmission distance: long-distance submarine cables experience significant impedance voltage drops under heavy loads, leading to excessive voltage drops along the cable; the other is related to the collector system loss rate: high losses cause heat generation, voltage drops, and reduced efficiency, further exacerbating voltage distribution anomalies. Therefore, in the offshore wind farm network health status assessment method that considers dynamic weight adjustments, increasing the weights of the submarine cable transmission distance and collector system loss rate can make the assessment results more sensitive to the deterioration trends of these two indicators, thereby guiding staff to more quickly identify and locate the root causes of electrical anomalies.

[0350] The specific process of step S4 includes:

[0351] S41: Determine the global sensitivity threshold:

[0352] ε is the global sensitivity threshold. In this embodiment, the global sensitivity threshold is dynamically adjusted.

[0353] Adjust the global sensitivity threshold based on wind power penetration rate:

[0354] ,

[0355] In the formula, X7 is the wind power penetration rate;

[0356] ,

[0357] In the formula, P f,p (t) represents the actual output of the p-th wind turbine at time t; N fP represents the total number of wind turbines in the wind farm. l,j (t) represents the real-time power of the j-th load at time t, N l The total load is used as the basis; the global sensitivity threshold is adjusted according to the wind power penetration rate, so that the triggering conditions for dynamic weight adjustment can adapt to changes in system electrical characteristics under different wind power output levels, avoiding false triggering or missed triggering, and improving the accuracy and effectiveness of the offshore wind farm network health status assessment method. Specifically, this is because:

[0358] When wind power penetration is high, the system is operating at high wind power output. Increasing the threshold can prevent the increase in voltage gradient caused by normal heavy load from being misjudged as an abnormal system operation. When wind power penetration is low, the system is operating at low wind power output and is generally quiet. Small anomalies can have a significant impact, so lowering the threshold can help detect problems earlier. The threshold is dynamically adjusted according to the wind power penetration rate, ensuring that the offshore wind farm network health status assessment method always matches the current operating mode. During changes in wind power penetration, the method distinguishes between normal electrical quantity changes and actual anomalies, improving the accuracy of the assessment method.

[0359] S42: If the global sensitivity does not exceed the threshold:

[0360] The health status assessment function value of an offshore wind farm is calculated based on the health status assessment function of the offshore wind farm; the formula for calculating the health status assessment function of an offshore wind farm is as follows:

[0361] ,

[0362] In the formula, Y * It is a health status assessment function for offshore wind farms; W r Let r be the comprehensive weight of the r-th indicator; Let be the r-th index after normalization.

[0363] The specific assessment function value of the offshore wind farm is calculated by using the health status assessment function of the offshore wind farm. The health level of the corresponding offshore wind farm is determined by matching and searching the assessment function value in the preset health level assessment table.

[0364] In one specific embodiment, a preset health level assessment table is shown in Table 1. The assessment function value is matched and searched against the grading standards in Table 1 to finally determine the health level of the offshore wind farm corresponding to the assessment function value.

[0365] Table 1 Health Level Assessment Table

[0366]

[0367] S43: If the global sensitivity exceeds the threshold:

[0368] When the global sensitivity F > ε, a weight adjustment is triggered, increasing the weight of submarine cable transmission distance and power collection system loss rate indicators.

[0369] A global sensitivity exceeding the threshold signifies a large-scale voltage shift and voltage gradient anomaly in the system, indicating that the system's operational abnormality has spread to a relatively serious level through submarine cable topology coupling. This suggests that the root cause is likely a factor affecting the overall electrical characteristics of the network. Submarine cable transmission distance and collector system loss rate are core parameters directly determining the voltage distribution and losses across the entire network; these two parameters directly reflect the operational status of offshore wind farms. Excessive submarine cable transmission distance can lead to voltage drops, especially with a large number of wind turbines connected, making monitoring of submarine cable transmission distance particularly important. Anomalies in the collector system loss rate may indicate problems with the lines. When the global sensitivity exceeds the threshold, increasing the weight of submarine cable transmission distance and collector system loss rate indicators gives these two indicators a greater weight in assessing the health of the offshore wind farm network. The assessment results are more sensitive to the deterioration trends of these two indicators, thus identifying risks caused by submarine cable voltage drops or abnormal collector system loss rates earlier and more accurately. This guides staff to quickly locate the root cause of the problem and reduce power generation losses and equipment failure risks.

[0370] When the global sensitivity is greater than the global sensitivity threshold, the weighting increase of the first indicator (submarine cable transmission distance) and the second indicator (loss rate) is calculated based on the global sensitivity. The formula for calculating the weighting increase is as follows: ,

[0371] Based on the magnitude of the weight increase, the combined weight of the first and second indicators is increased, while the combined weight of the other indicators is decreased, resulting in the new weights for each indicator.

[0372] The formula for updating the overall weight of the first indicator is:

[0373] ,

[0374] In the formula, W1 represents the new weight of the first indicator; W1 represents the overall weight of the first indicator.

[0375] The formula for updating the overall weight of the second indicator is:

[0376] ,

[0377] In the formula, W1 represents the new weight of the second indicator; W2 represents the overall weight of the second indicator.

[0378] The formula for updating the combined weights of the third to eighth indicators is as follows:

[0379] ,

[0380] In the formula, For the first New weights for each indicator; For the first A comprehensive indicator of all indicators. ,Right now .

[0381] The health status of the offshore wind farm network is calculated and assessed based on the new weights of each indicator.

[0382] The update formula for the health status assessment function of offshore wind farms is as follows:

[0383] ,

[0384] In the formula, For the new health status assessment function of offshore wind farms; As the new weight for the r-th indicator, Let r be the normalized index. .

[0385] Based on the evaluation function value of the new offshore wind farm health status assessment function, a search is performed in a preset health level assessment table to determine the corresponding offshore wind farm health level. In one specific embodiment, based on the evaluation function value of the new offshore wind farm health status assessment function, Table 1 is consulted to determine the corresponding offshore wind farm health level.

[0386] When F=0.05, ;when hour, The dynamic weight adjustment strategy precisely controls the weights of submarine cable transmission distance and power collection system loss rate by real-time monitoring of the global sensitivity F and dynamically setting the global sensitivity threshold ε that triggers weight adjustments based on wind power penetration rate. By strengthening the weights of key indicators, this strategy effectively improves the response speed and adaptability of the assessment method to changes in the operating conditions of offshore wind farms, enabling rapid capture of dynamic signals. This provides strong support for ensuring the stable and efficient operation of offshore wind farms with a high proportion of distributed wind power integration, enhances the reliability and practicality of the assessment system under complex operating conditions, makes the assessment method more aligned with actual operational needs, and contributes to the refined management and optimized operation of offshore wind farms.

[0387] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0388] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0389] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0390] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0391] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment, characterized in that, include: Acquire electrical performance data of offshore wind farms and calculate multiple metrics for evaluation; Based on the subjective and objective weighting method, the comprehensive weight of each indicator is calculated separately; Based on the electrical coupling relationship and voltage distribution between submarine cables in offshore wind farms, global sensitivity is calculated in real time. The global sensitivity characterizes the overall tension of the system after an abnormal operating state of the offshore wind farm under the current submarine cable topology is spread through the electrical coupling relationship between the submarine cables. Compare global sensitivity with the global sensitivity threshold: When the global sensitivity does not exceed the global sensitivity threshold, the health status of the offshore wind farm network is assessed based on the comprehensive weight of each indicator. When the global sensitivity exceeds the global sensitivity threshold, the comprehensive weight of each indicator is adjusted according to the global sensitivity to obtain the new weight of each indicator; the health status of the offshore wind farm network is assessed based on the new weight of each indicator.

2. The method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment according to claim 1, characterized in that, The process involves acquiring electrical performance data from offshore wind farms and calculating multiple indicators for evaluation, including: The first to eighth indicators are selected as follows: submarine cable transmission distance, power collection system loss rate, unit submarine cable voltage drop, substation load rate, grid connection point voltage qualification rate, harmonic distortion rate, wind power penetration rate, and wind power fluctuation. Obtain electrical performance data of offshore wind farms and calculate the original eight indicators; The original third, fifth, sixth, and seventh indicators were transformed into single indicators to represent the overall characteristics. The transformation of the third indicator, unit cable voltage drop, into a single indicator involves: calculating the final attention coefficients of each cable segment with its neighboring cables using a pre-trained graph attention network; calculating the final attention coefficients of each cable segment based on these final attention coefficients; and weighting the unit voltage drop of each cable segment. After normalizing each individual indicator, eight indicators are obtained for evaluation.

3. The method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment according to claim 2, characterized in that, The process of performing a single-value transformation on the original third, fifth, sixth, and seventh indicators includes: The third singularity index is the average of the weighted sums of the unit voltage drop values ​​of all submarine cables, which is taken as the singularity index Y3 after the unit submarine cable voltage drop X3 is transformed. The calculation formula is as follows: ， In the formula, is the weighted value of the unit voltage drop for the k-th segment of the submarine cable; M represents the total number of segments of the submarine cable; ， In the formula, X 3,k α represents the unit voltage drop of the k-th segment of the submarine cable; k Let be the final attention coefficient of the k-th segment of the submarine cable; ， In the formula, V s,k V e,k These are the voltages at the beginning and end of the k-th segment of the submarine cable, respectively; L k Let k be the length of the submarine cable segment. The fifth single-indicator, the grid connection point voltage qualification rate X5, is calculated using the following formula after single-indicator conversion: ， In the formula, N b X represents the total number of grid connection points. 5,u It is the voltage qualification rate of the u-th grid connection point; ， ， In the formula, for Voltage at the u-th grid connection point at time u, V N The voltage is the nominal voltage of the power grid, and T is the preset total time for calculating the voltage qualification rate at the grid connection point. It is a function that represents 0 or 1; The sixth single-indicator, harmonic distortion rate X6, is calculated using the following formula after single-indicator transformation: ， In the formula, X 6,u Let w be the harmonic distortion rate at the u-th grid connection point. 6,u The weighting coefficient for the u-th grid connection point; ， In the formula, I 1,u I is the effective value of the fundamental current at the u-th grid connection point; h,u This represents the effective value of the h-th harmonic current at the u-th grid connection point. It is the sum of the squares of the effective values ​​of all harmonic currents at the u-th grid connection point; ， In the formula, L u,tr Let be the line distance from the u-th grid connection point to the step-up transformer in the substation; The seventh single-indicator, wind power penetration rate X7, is transformed into index Y7, and the calculation formula is as follows: ， In the formula, N s X represents the total number of samples taken within the preset evaluation period. 7,t. Let X7 be the value of the wind power penetration rate in the t-th sampling. ， In the formula, P f,p (t.) represents the actual output of the p-th wind turbine at the t.th sampling time; N f P represents the total number of wind turbines in the wind farm. l,j (t.) represents the real-time power of the j-th load at the t.th sampling time; N l Total load; The process of normalizing each individual indicator yields eight indicators for evaluation, including: For the first to eighth singularity indicators: X1, X2, Y3, X4, Y5, Y6, Y7, and X8, normalization was performed respectively. The eight indicators after normalization are as follows: , This represents the r-th index after normalization.

4. The method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment according to claim 2 or 3, characterized in that, The calculation of the final attention coefficients between each submarine cable segment and its neighboring submarine cables using a pre-trained graph attention network includes: The pre-trained graph attention network is a two-layer graph attention network; The process of calculating the final attention coefficient between each submarine cable segment and its neighboring submarine cables using a two-layer graph attention network is as follows: Pre-construct the graph and initialize the features: treat each submarine cable segment as a graph node, and the physical connection between submarine cables as an edge. A physical connection refers to a connection to the same electrical equipment; calculate the terminal voltage difference of each submarine cable segment as a one-dimensional initial feature and perform normalization processing. The first layer uses multi-head attention, which expands and divides the normalized terminal voltage difference of each submarine cable segment into multiple sub-features, corresponding to the input attention heads. Attention coefficients with neighboring submarine cables are calculated for each attention head, and the information is aggregated to output high-dimensional features to the second layer. The second layer uses single-head attention, which calculates the final attention coefficients of each submarine cable segment with its neighboring submarine cables based on the high-dimensional features output by the first layer. The calculation of the final attention coefficient of each submarine cable segment based on the final attention coefficient of each segment and its neighboring submarine cables includes: The final attention coefficient α of the k-th submarine cable is obtained by averaging the final attention coefficients of the k-th submarine cable calculated by the second layer of the two-layer graph attention network with its neighboring submarine cables. k The average calculation refers to summing the final attention coefficients of the k-th submarine cable segment and its neighboring submarine cables, and then dividing by the total number of neighboring submarine cables of the k-th submarine cable segment.

5. The method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment according to claim 2, characterized in that, The method of subjective and objective weighting is used to calculate the comprehensive weight of each indicator, including: Based on the subjective weighting method, the subjective weights of each indicator are calculated, including: Establish indicator order relationship: Based on the subjective experience of experts, the 8 indicators are ranked from high to low importance to form indicator order relationship B, and the correspondence between the 8 indicators and the importance level indicators after the importance ranking is established. B={b1, b2, b3, b4, b5, b6, b7, b8}, In the formula, the elements b1, b2, b3, b4, b5, b6, b7, and b8 of B are importance level indicators after importance ranking, indicating that the importance of the corresponding indicators decreases in order, with b1 being the most important and b8 being the least important. Determine the importance ratio of adjacent importance level indicators in the indicator order relationship: Based on the subjective experience of experts, set the importance ratio of adjacent importance level indicators in the indicator order relationship as: R2, R3, R4, R5, R6, R7, R8; representing the importance of b1 to b2, b2 to b3, b3 to b4, b4 to b5, b5 to b6, b6 to b7, and b7 to b8, respectively. Calculate the subjective weights of each indicator: Calculate the weight of the least important metric b8 : ， The weights of other importance level indicators are calculated recursively: ， ， ， ， ， ， ， to These represent the weight coefficients of the most important indicator b1 to the least important indicator b8, respectively. Based on the correspondence between the eight indicators and the importance level indicators ranked by importance, the weights of the importance level indicators are assigned to the eight indicators accordingly, serving as the subjective weights of the corresponding indicators. , ; Based on the objective weighting method using entropy weighting, the objective weights of each indicator are calculated, including: Calculate the proportion of each normalized index of each wind farm relative to its category: Get N w The eight indicators after normalization of wind farms, and the proportion P of the r-th indicator after normalization of wind farm a to the same category. r,a The calculation formula is: ， In the formula, ; This represents the r-th index after normalization for the a-th wind farm; N represents w Sum of the r-th index after normalization of each wind farm; Determine the entropy weights of each normalized index: The entropy value e of the r-th index after normalization r The calculation formula is: ， Calculate the objective weights of each indicator: The objective weight of the r-th indicator The calculation formula is: ； Based on the subjective and objective weights of each indicator, the comprehensive weight of each indicator is calculated, including: The comprehensive weight W of the r-th indicator r Calculation formula: ， In the formula, This represents the objective weight of the r-th indicator; This represents the subjective weight of the r-th indicator.

6. The method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment according to claim 2, characterized in that, The method of calculating global sensitivity in real time based on the electrical coupling relationship and voltage distribution between submarine cables in offshore wind farms includes: The topology sensitivity characteristics of each cable segment are calculated by weighting and aggregating the voltage offset and voltage spatial gradient of each neighboring cable segment using the final attention coefficients of each segment and its neighboring cable; the topology sensitivity characteristic F of the k-th cable segment is also calculated. k The calculation formula is: ， In the formula, N(k) represents the set of neighboring submarine cables of the k-th submarine cable segment. The neighboring submarine cables of the k-th submarine cable segment refer to other submarine cables that are connected to the same electrical equipment as the k-th submarine cable segment. This represents the final attention coefficient between the k-th submarine cable segment and its neighboring submarine cable i, calculated by the pre-trained graph attention network. This represents the voltage offset of the neighboring submarine cable i of the k-th segment of the submarine cable; This represents the spatial voltage gradient of neighboring submarine cable i; Based on the topological sensitivity characteristics of each segment of the submarine cable, the global sensitivity is calculated; the formula for calculating the global sensitivity F is: ， In the formula, M represents the total number of submarine cable segments.

7. The method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment according to claim 6, characterized in that, Voltage offset of the neighboring submarine cable i of the k-th segment of the submarine cable The calculation formula is: ， In the formula, This represents the voltage offset at the beginning of neighboring submarine cable i. This indicates the voltage offset at the tail end of neighboring submarine cable i; ， In the formula, V s,i V represents the voltage at the beginning of neighboring submarine cable i; N Indicates the nominal voltage of the power grid; ， In the formula, V e,i This represents the voltage at the tail end of neighboring submarine cable i; Voltage spatial gradient of the neighboring submarine cable i The calculation formula is: ， In the formula, This represents the spatial gradient of the voltage at the beginning of neighboring submarine cable i; This represents the spatial gradient of the tail voltage of neighboring submarine cable i; ， In the formula, N(i,s) * N(i,s) represents the total number of neighboring submarine cables at the beginning of neighboring submarine cable i; N(i,s) represents the set of neighboring submarine cables at the beginning of neighboring submarine cable i. It is a neighboring submarine cable at the head end of the neighboring submarine cable i, representing The submarine cable i-end is connected to the same electrical equipment as the neighboring cable; Indicates neighboring submarine cable average voltage, , Indicates neighboring submarine cable The voltage at the beginning of the circuit, Indicates neighboring submarine cable The voltage at the tail end; ， In the formula, N(i,e) * N(i,e) represents the total number of neighboring submarine cables at the end of neighboring submarine cable i; N(i,e) represents the set of neighboring submarine cables at the end of neighboring submarine cable i. It is a neighboring submarine cable at the end of the neighboring submarine cable i, representing The end of the neighboring submarine cable is connected to the same electrical equipment. Indicates neighboring submarine cable average voltage, , Indicates neighboring submarine cable The voltage at the beginning of the circuit, Indicates neighboring submarine cable The voltage at the tail end.

8. The method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment according to claim 2, characterized in that, Before comparing the global sensitivity with the global sensitivity threshold, the following steps are included: The global sensitivity threshold ε is adjusted based on wind power penetration rate, and the formula for calculating the global sensitivity threshold ε is as follows: ， In the formula, X7 is the wind power penetration rate; ， In the formula, P f,p (t) represents the actual output of the p-th wind turbine at time t; N f P represents the total number of wind turbines in the wind farm. l,j (t) represents the real-time power of the j-th load at time t, N l This represents the total load.

9. The method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment according to claim 8, characterized in that, The assessment of the health status of the offshore wind farm network based on the comprehensive weight of each indicator includes: The evaluation function value is calculated based on the health status evaluation function of the offshore wind farm; The health level of the corresponding offshore wind farm is determined by matching and searching the evaluation function value in the preset health level assessment table. The health status assessment function Y of the offshore wind farm * The calculation formula is: ， In the formula, W r Let r be the comprehensive weight of the r-th indicator; Let be the r-th index after normalization.

10. The method for assessing the health status of offshore wind farm networks considering dynamic weight adjustment according to claim 8, characterized in that, The process of adjusting the overall weight of each indicator based on global sensitivity to obtain the new weight of each indicator includes: The weight increase of the first and second indicators is calculated based on global sensitivity. The formula for calculating the weight increase is as follows: ， In the formula, F represents global sensitivity; Based on the magnitude of the weight increase, the combined weight of the first and second indicators is increased, while the combined weight of the other indicators is decreased, resulting in the new weights for each indicator. The formula for updating the overall weight of the first indicator is: ， In the formula, W1 represents the new weight of the first indicator; W1 represents the overall weight of the first indicator. The formula for updating the overall weight of the second indicator is: ， In the formula, W1 represents the new weight of the second indicator; W2 represents the overall weight of the second indicator. The formula for updating the combined weights of the third to eighth indicators is as follows: ， In the formula, For the first New weights for each indicator; For the first A comprehensive indicator of all indicators. ; Assess the health status of offshore wind farm networks based on the new weights of each indicator; The update formula for the health status assessment function of offshore wind farms is as follows: ， In the formula, For the new health status assessment function of offshore wind farms; As the new weight for the r-th indicator, Let r be the normalized index. .

Images