A cluster-type wind power generation terminal maintenance plan management system

Abstract

The present invention discloses a cluster-type micro-wind power generation terminal maintenance plan management system, which relates to the field of intelligent operation and maintenance technology. It includes: a data fusion module, which fuses SCADA data, laser radar wake scanning and power grid PMU data to obtain a health status vector; an impact assessment module, which obtains the wind turbine adjustment cost factor of the wind turbine according to the health status vector, and determines the composite impact value of each wind turbine through the LSTM network model; a coordinated maintenance module, which determines the credit score of each wind turbine according to the composite impact value and sorts the maintenance order of each wind turbine. The present invention preliminarily grades each wind turbine according to the credit score, and performs secondary division in each grade according to the maintenance score, so as to ensure that high-impact, low-credit wind turbines can be processed first, thereby improving the marginal benefit of maintenance resource investment.

Description

Technical Field

[0001] The present invention relates to the field of intelligent operation and maintenance technology, and in particular to a cluster-type micro-wind power generation terminal maintenance plan management system. Background Art

[0002] In recent years, micro-wind power generation technology has shown promising application prospects in distributed power supply systems in urban fringe areas, rural areas, and remote regions due to its adaptability to low wind speeds, flexible deployment, and small footprint. However, due to the low output power of individual micro-wind power generation terminals, clustered deployment to form power generation arrays is often required to improve overall energy supply capacity.

[0003] The large number of devices, widespread distribution, and complex operating environments in clustered wind power generation systems present significant challenges in operation and maintenance. Traditional O&M models, which rely primarily on manual inspections and scheduled maintenance, suffer from delayed response times, high costs, and low efficiency, making them difficult to meet the reliability and intelligent management requirements of modern energy systems. Furthermore, the lack of a unified O&M scheduling mechanism and condition monitoring methods can easily lead to wasted maintenance resources and untimely equipment failures, impacting the stability and economic viability of the entire power generation system.

[0004] However, due to the physical dependencies between wind turbines, such as wake effects and grid scheduling coupling, a single turbine failure can cause aerodynamic interference, reducing the power generation efficiency of adjacent turbines. Furthermore, current maintenance mechanisms for wind turbines lack a scientific basis for prioritizing maintenance, as traditional single-unit health monitoring cannot quantify the marginal impact of a failure on the cluster. Summary of the Invention

[0005] The object of the present invention is to provide a cluster-type wind power generation terminal maintenance plan management system to solve the problems raised in the above background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solutions: a cluster-type wind power generation terminal maintenance plan management system, comprising:

[0007] The data fusion module fuses SCADA data, lidar wake scans, and power grid PMU data to obtain the health status vector;

[0008] An impact assessment module obtains a fan adjustment cost factor of the fan according to the health state vector, and determines a composite impact value of each fan through an LSTM network model;

[0009] The coordinated maintenance module determines the credit score of each wind turbine based on the composite impact value and sorts the maintenance order of each wind turbine based on the credit score, including:

[0010] SC1: Determine the penalty result: Based on the composite impact value, obtain the credit score of each wind turbine through the blockchain smart contract;

[0011] SC2: Classification management: Wind turbines are classified into different levels according to the credit score;

[0012] SC3: Determine maintenance score: Obtain a maintenance score for each wind turbine based on the composite impact value and the neighbor impact weight coefficient, and rank all wind turbines in each level based on the maintenance score. The formula for obtaining the maintenance score is as follows:

[0013]

[0014] in: Score the maintenance priority, is the weight coefficient of the composite impact value, is the composite impact value of the j-th wind turbine, For the owner's credit score, is the neighbor influence weight coefficient, Efficiency loss for neighbors.

[0015] Furthermore, the health status vector is obtained, including:

[0016] SA1: Spatiotemporal alignment processing: Based on the SCADA data, LiDAR wake scan and power grid PMU data, a coordinate system conversion matrix is ​​established to obtain the aligned data coordinates, specifically:

[0017]

[0018] in: is the coordinate of the target point in the global coordinate system, is the rotation matrix, is the radial distance of the target point measured by the lidar, is the azimuth of the target point measured by the lidar, is the height of the target point measured by the lidar, is the translation vector;

[0019] SA2: Kalman filter processing: By improving the Kalman filter algorithm and aligning the data coordinates, the Mahalanobis distance is obtained, specifically:

[0020]

[0021] in: is the Mahalanobis distance, is the inverse of the residual vector, is the residual vector, is the residual covariance matrix, is the observation matrix, is the inversion of the observation matrix, is the state estimation covariance matrix, is the observation noise covariance matrix;

[0022] SA3: Determine parameter weights: Based on the Mahalanobis distance, determine the final comprehensive health of each wind turbine, specifically:

[0023]

[0024] in: is the comprehensive health of the j-th wind turbine, is the Mahalanobis distance, is the preset distance threshold, is the mechanical health of the j-th fan, is the electrical health of the j-th wind turbine, is the turbulence intensity of the j-th wind turbine, is the baseline turbulence intensity;

[0025] SA4: Determine the health state vector: Determine the final health state vector based on the final comprehensive health level, specifically:

[0026]

[0027] in: is the health status vector, is the final comprehensive health of the first wind turbine, is the final comprehensive health of the nth wind turbine.

[0028] Furthermore, the Mahalanobis distance is compared with a preset distance threshold, and the weight of the comprehensive health is adjusted according to the comparison result, specifically:

[0029] When the Mahalanobis distance is greater than a preset distance threshold, the weight of the comprehensive health is reduced; otherwise, the weight of the comprehensive health remains unchanged.

[0030] Furthermore, the mechanical health and electrical health of the wind turbine are obtained according to the adjusted weights in the comprehensive health, specifically:

[0031]

[0032] in: is the mechanical health of the fan, The electrical health of the wind turbine, is the actual temperature rise, is the maximum allowable temperature rise, is the effective value of vibration, is the vibration critical value, is 3 times the frequency energy, is the baseline band energy, is the current insulation resistance, is the insulation resistance of the new equipment, is the current harmonic distortion rate, is the maximum allowable current harmonic distortion rate.

[0033] Furthermore, the weight distribution is determined, including:

[0034] SB1: Determine the impact factor: Use the health status vector as the input of the constructed Jensen model and output the impact factor, specifically:

[0035]

[0036] in: is the wake influence factor of the j-th wind turbine, is the total number of fans, is the equivalent wake velocity of the jth affected wind turbine, is the upstream free stream wind speed, is the index of the fan;

[0037] SB2: Impact Factor Update: The turbulence correction coefficient of the Jensen model is modified through real-time CFD verification model, and the Jensen model is updated based on the modified turbulence correction coefficient. At the same time, an updated wake impact factor is obtained based on the updated Jensen model;

[0038] SB3: Determine the adjustment cost factor: According to the health state vector, obtain the fan adjustment cost factor, specifically:

[0039]

[0040] in: is the system adjustment cost factor of the j-th fan, is the grid frequency deviation, is the total number of fans, is the index of the fan, is the loss coefficient matrix element between the i-th fan and the j-th fan, is the output power of the jth wind turbine, is the linear loss coefficient of the i-th fan;

[0041] SB4: Determine the composite impact value: Based on the updated wake impact factor and wind turbine adjustment cost factor, the composite impact value is obtained through the LSTM network model, specifically:

[0042]

[0043] in: is the composite impact value of the j-th wind turbine, is the wake influence factor of the j-th wind turbine, is the system adjustment cost factor of the j-th fan, is the aerodynamic weight coefficient, is the grid regulation cost weight coefficient, is the neighbor coupling influence weight coefficient, is the neighbor coupling influence factor of the j-th wind turbine.

[0044] Furthermore, the output obtains impact factors, including:

[0045] SB1.1: Data processing: Based on SCADA data and lidar wake scan data, obtain turbulence intensity and upstream free stream wind speed, specifically:

[0046]

[0047] in: is the turbulence intensity of the fan, is the upstream free stream wind speed, is the statistical time window, is the instantaneous radial wind speed, is the average radial wind speed, is the fan output power, is the air density, is the impeller swept area, is the wind energy utilization coefficient;

[0048] SB1.2: Construct wake field: Based on the turbine wake radius and the constructed wind column coordinate system, obtain the wake velocity distribution at each grid point, specifically:

[0049]

[0050] in: is the wake velocity distribution, is the upstream free stream wind speed, is the initial wake radius, is the downstream axial distance, is the turbulence intensity of the fan, is the fan rotor swept diameter, is the vertical distance of the target position from the centerline of the wake;

[0051] SB1.3: Generate wake impact factor: Determine the wake impact factor based on the wake velocity distribution and the total number of wind turbines.

[0052] Furthermore, the updated wake influence factors are obtained, including:

[0053] SB2.1: Simulation trigger: Compare the wake impact factor with the simulated impact factor of the real-time CFD verification model to obtain the impact factor difference. At the same time, compare the impact factor difference with the preset difference threshold. Based on the comparison result, simulate the real-time CFD verification model. Specifically:

[0054] When the difference in the impact factor is greater than a preset difference threshold, the simulation of the real-time CFD verification model is performed; otherwise, the simulation of the real-time CFD verification model is not performed;

[0055] SB2.2: Determine model error: Based on the simulation impact factor and the wake impact factor, determine the relative model error, specifically:

[0056]

[0057] in: is the relative error of the model, Simulation impact factors for real-time CFD validation models, is the wake influence factor of the Jensen model;

[0058] SB2.3: Determine the updated turbulence correction factor: Compare the model relative error with a preset error threshold and update the turbulence correction factor based on the comparison result, specifically:

[0059] When the model relative error is greater than a preset error threshold, the turbulence correction coefficient is updated; otherwise, the turbulence correction coefficient is not updated;

[0060] The updating formula of the turbulence correction coefficient is specifically:

[0061]

[0062] in: is the updated turbulence correction coefficient, is the initial turbulence correction coefficient, is the relative error of the model.

[0063] Furthermore, the credit score of each wind turbine is obtained, including:

[0064] SC1.1: Determine the pledge rate: Based on the composite impact value, adjust the pledge rate of the on-chain asset pledge mechanism, specifically:

[0065]

[0066] in: is the adjusted pledge rate, is the initial pledge rate, is the composite impact value of the j-th wind turbine, is the growth rate of the pledge rate;

[0067] SC1.2: Determine Penalty: Compare the composite impact value with the preset impact threshold, and compare the neighbor aggregation efficiency loss with the preset efficiency loss threshold. Based on the comparison results, determine the penalty. Based on the penalty and the current pledge amount, determine the credit score. Specifically:

[0068]

[0069] in: For the final credit score, is the weight coefficient of fine proportion, For fines, is the current total amount of pledge, is the weight coefficient of delayed maintenance times, The number of delayed maintenance.

[0070] Furthermore, the composite impact value is compared with a preset impact threshold, and the neighbor aggregation efficiency loss is compared with a preset efficiency loss threshold. Based on the comparison results, a penalty is determined, specifically:

[0071] When the composite impact value is greater than the preset impact threshold and the neighbor aggregation efficiency loss is greater than the preset efficiency loss threshold, a fine is deducted through the blockchain smart contract; otherwise, no fine is deducted.

[0072] Furthermore, the credit score is determined based on the fine and the current pledge amount, including:

[0073] SC1.2.1: Determine the penalty coefficient: Based on the composite impact value, obtain the penalty coefficient, specifically:

[0074]

[0075] in: is the penalty coefficient, is the weight coefficient of the composite impact value, is the composite impact value of the j-th wind turbine, is the progressive penalty base, The number of delayed maintenance;

[0076] SC1.2.2: Determine the penalty: The penalty is determined based on the penalty coefficient and the base fee of the pledged assets, specifically:

[0077]

[0078] in: For fines, As the basic fee, is the penalty coefficient;

[0079] SC1.2.3: Determine the credit score change: Based on the fine and pledge amount, determine the credit score change, specifically:

[0080]

[0081] in: is the credit score change, is the weight coefficient of fine proportion, For fines, is the current total amount of pledge, is the weight coefficient of delayed maintenance times, The number of delayed maintenance.

[0082] Compared with the prior art, the present invention has the following beneficial effects:

[0083] First, the present invention initially categorizes wind turbines by credit score, and then, within each tier, performs a secondary categorization based on maintenance scores. This ensures that high-impact, low-credit wind turbines are prioritized, thereby improving the marginal benefit of maintenance resource investment.

[0084] Second, this invention fuses lidar wake scans with grid PMU data through spatiotemporal alignment and Kalman filtering, and constructs a corresponding comprehensive health state vector. This solves the data silo problem and improves fault detection sensitivity. Simultaneously, through the Jensen model and real-time CFD verification, it generates a wake impact factor. Using an LSTM network, it dynamically adjusts the weights of aerodynamics, grid regulation costs, and neighbor coupling to obtain the corresponding composite impact value, thereby avoiding subjectivity in maintenance priority decisions.

[0085] Third: The present invention binds individual wind turbines to wind turbine clusters by dynamically adjusting the pledge rate and delayed maintenance times of the smart contract, and realizes automated cost sharing through economic leverage. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1This is a flowchart of the implementation of the cluster-type wind power generation terminal maintenance plan management system of the present invention;

[0087] Figure 2 It is the spatial distribution diagram of the spatiotemporal alignment feature in the present invention;

[0088] Figure 3 This is a trend analysis diagram of the equipment health status in the present invention;

[0089] Figure 4 A diagram showing the relationship between credit points and penalties in the present invention;

[0090] Figure 5 This is the final maintenance sorting graph in the present invention. DETAILED DESCRIPTION

[0091] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0092] Due to the physical correlations between wind turbines, such as wake effects and grid scheduling coupling, when a single wind turbine fails, it will cause the power generation efficiency of adjacent wind turbines to decrease due to aerodynamic interference. At the same time, in the process of maintaining wind turbines by the current maintenance mechanism, the traditional single-machine health monitoring cannot quantify the marginal impact of the failure on the cluster, which will lead to a lack of scientific basis for maintenance priority decisions. The technical solution of this application obtains the health state vector of the wind turbine through the fusion of multi-source data, and determines the composite impact value through the wake effect and the grid regulation cost. At the same time, based on the blockchain smart contract, it dynamically adjusts the credit point mechanism, and through the LSTM network, quantifies the marginal impact of the failure on the cluster, and through hierarchical sorting maintenance priorities, realizes intelligent maintenance scheduling of the individual health of the wind turbine and the overall efficiency of the cluster.

[0093] Example 1

[0094] refer to Figures 1-4 This embodiment provides a maintenance plan management system for clustered micro-wind power generation terminals. The system includes a data fusion module, an impact assessment module, and a coordinated maintenance module. In this embodiment, the data fusion module is used to fuse SCADA data, LiDAR wake scans, and power grid PMU data using an improved Kalman filter algorithm and a spatiotemporal alignment engine to obtain the corresponding health status vector. The details are as follows:

[0095] Step SA1: Spatiotemporal alignment. This involves deploying a data acquisition terminal inside each wind turbine nacelle, including but not limited to an ADC module, an IEPE vibration sensor interface, and an Ethernet interface, to obtain the corresponding SCADA data. Simultaneously, a scanning Doppler lidar is used to scan the target wind turbine cluster and obtain the corresponding lidar wake scan data. A PMU concentrator is also installed at the wind farm's booster station to obtain the corresponding grid PMU data.

[0096] Furthermore, a global coordinate system is established based on the acquired SCADA data, lidar wake scanning data, and power grid PMU data. A coordinate system transformation matrix is ​​established with the wind farm control center as the origin to obtain the corresponding aligned data coordinates, specifically:

[0097]

[0098] in: is the coordinate of the target point in the global coordinate system, is the rotation matrix, is the radial distance of the target point measured by the lidar, is the azimuth of the target point measured by the lidar, is the height of the target point measured by the lidar, is the translation vector.

[0099] Step SA2: Kalman filter processing. That is, based on the alignment data obtained in step SA1, the corresponding Mahalanobis distance is obtained by improving the Kalman filter algorithm, specifically:

[0100]

[0101] in: is the Mahalanobis distance, is the inverse of the residual vector, is the residual vector, is the residual covariance matrix, is the observation matrix, is the inversion of the observation matrix, is the state estimation covariance matrix, is the observation noise covariance matrix.

[0102] Step SA3: Determine parameter weights. The Mahalanobis distance obtained in step SA2 is compared with a preset distance threshold. Based on the comparison result, the weight of the comprehensive health of each wind turbine is adjusted to determine the final comprehensive health of each wind turbine, specifically:

[0103] When the obtained Mahalanobis distance is greater than the preset distance threshold, the parameter corresponding to the Mahalanobis distance is considered an abnormal parameter, and the weight coefficient corresponding to the abnormal parameter is reduced. Conversely, when the obtained Mahalanobis distance is not greater than the preset distance threshold, the parameter corresponding to the Mahalanobis distance is considered a normal parameter, and the weight coefficient corresponding to the normal parameter is not changed.

[0104] In this embodiment, the formula for obtaining the final comprehensive health of each wind turbine is specifically:

[0105]

[0106] in: is the comprehensive health of the j-th wind turbine, is the Mahalanobis distance, is the preset distance threshold, is the mechanical health of the j-th fan, is the electrical health of the j-th wind turbine, is the turbulence intensity of the j-th wind turbine, is the baseline turbulence intensity.

[0107] Step SA4: Determine the health state vector. This is based on the acquired SCADA data, LiDAR wake scan data, grid PMU data, and the adjusted weights of the comprehensive health of each wind turbine determined in step SA3. The corresponding mechanical health and electrical health of each wind turbine are obtained, specifically:

[0108]

[0109] in: is the mechanical health of the fan, The electrical health of the wind turbine, is the actual temperature rise, is the maximum allowable temperature rise, is the effective value of vibration, is the vibration critical value, is 3 times the frequency energy, is the baseline band energy, is the current insulation resistance, is the insulation resistance of the new equipment, is the current harmonic distortion rate, is the maximum allowable current harmonic distortion rate.

[0110] Furthermore, based on the final comprehensive health of each wind turbine, the final health state vector is determined, specifically:

[0111]

[0112] in: is the health status vector, is the final comprehensive health of the first wind turbine, is the final comprehensive health of the nth wind turbine.

[0113] refer to Figure 2 , Figure 2 is the spatial distribution diagram of the spatiotemporal alignment feature in this embodiment, Figure 2 It can be seen that fault data forms a continuous trajectory in the feature space. Furthermore, under fault conditions, the changes in SCADA data, LiDAR wake scan data, and power grid PMU data exhibit a nonlinear coupling relationship. Specifically, SCADA data exhibits an upward trend, LiDAR wake scan data exhibits a downward trend, and power grid PMU data exhibits a doubling trend. Normal data, on the other hand, clusters in a compact area. Therefore, data fusion in this embodiment can effectively reduce systematic biases between sensors.

[0114] refer to Figure 3 , Figure 3 This is the equipment health status trend analysis diagram in this embodiment, Figure 3 The comprehensive health index showed a slow downward trend, then plummeted during the fault period, exceeding the fault threshold. The fault was triggered by a sudden increase in the Mahalanobis distance and a drop in the health index below the threshold, with the fault duration remaining for 4.8 hours. Furthermore, within the fault interval, moving average filtering smoothed the data, avoiding false alarms caused by transient jitter. Furthermore, before the fault occurred, the comprehensive health index was close to the warning line, indicating that the set fault threshold accurately captured the fault event and formed a dual verification mechanism with the dynamic Mahalanobis distance threshold.

[0115] In this embodiment, the impact assessment module is used to obtain the corresponding wake impact factor and wind turbine adjustment cost factor based on the health status vector determined in step SA4, and determine the composite impact value of each wind turbine through the LSTM network model. The details are as follows:

[0116] Step SB1: Determine the impact factor. The health status vector determined in step SA4 is used as the input of the constructed Jensen model, and the corresponding impact factor is obtained as output, specifically:

[0117]

[0118] in: is the wake influence factor of the j-th wind turbine, is the total number of fans, is the equivalent wake velocity of the jth affected wind turbine, is the upstream free stream wind speed, The index of the fan.

[0119] During the specific implementation process, the speeds of the three wind turbines were attenuated by 20%, 30% and 40% respectively, and the corresponding wake impact factor was 0.0967.

[0120] Step SB2: Update the impact factor. This involves modifying the turbulence correction coefficient of the Jensen model in step SB1 using the real-time CFD validation model. The Jensen model in step SB1 is then updated based on the modified turbulence correction coefficient. The updated impact factor is then obtained using the updated Jensen model. The details are as follows:

[0121] Step SB2.1: Simulation triggering. This involves comparing the impact factor obtained in step SB1 with the simulated impact factor obtained by the real-time CFD verification model, obtaining the impact factor difference between the two, and comparing the obtained impact factor difference with a preset difference threshold. Based on the comparison result, it is determined whether to simulate the real-time CFD verification model. Specifically:

[0122] When the obtained difference in the impact factor is greater than the preset difference threshold, the real-time CFD verification model simulation is performed. Conversely, when the obtained difference in the impact factor is not greater than the preset difference threshold, the real-time CFD verification model simulation is not performed.

[0123] Step SB2.2: Determine the model error. That is, based on the simulation impact factor corresponding to the real-time CFD verification model and the impact factor obtained in step SB1, determine the relative model error between the two. Specifically,

[0124]

[0125] in: is the relative error of the model, Simulation impact factors for real-time CFD validation models, is the wake influence factor of the Jensen model.

[0126] Step SB2.3: Determine the updated turbulence correction coefficient. Compare the model relative error determined in step SB2.2 with the preset error threshold, and update the turbulence correction coefficient based on the comparison result. Specifically:

[0127] When the obtained model relative error is greater than a preset error threshold, the turbulence correction coefficient is updated. Conversely, when the obtained model relative error is not greater than the preset error threshold, the turbulence correction coefficient is not updated.

[0128] In this embodiment, the updating formula of the turbulence correction coefficient is specifically:

[0129]

[0130] in: is the updated turbulence correction coefficient, is the initial turbulence correction coefficient, is the relative error of the model.

[0131] In the specific implementation process, the preset error threshold is set to 0.2, and the model relative error is 0.25, so the initial turbulence correction coefficient is updated. At the same time, the initial turbulence correction coefficient is 1.08, and the corresponding updated turbulence correction coefficient is 1.107.

[0132] Step SB3: Determine the adjustment cost factor. That is, according to the health state vector determined in step SA4, obtain the corresponding wind turbine adjustment cost factor, specifically:

[0133]

[0134] in: is the system adjustment cost factor of the j-th fan, is the grid frequency deviation, is the total number of fans, is the index of the fan, is the loss coefficient matrix element between the i-th fan and the j-th fan, is the output power of the jth wind turbine, is the linear loss coefficient of the i-th fan.

[0135] Step SB4: Determine the composite impact value. That is, based on the updated impact factor obtained in step SB2 and the wind turbine adjustment cost factor obtained in step SB3, the corresponding composite impact value is obtained through the LSTM network model. Specifically, it is:

[0136]

[0137] in: is the composite impact value of the j-th wind turbine, is the wake influence factor of the j-th wind turbine, is the system adjustment cost factor of the j-th fan, is the aerodynamic weight coefficient, is the grid regulation cost weight coefficient, is the neighbor coupling influence weight coefficient, is the neighbor coupling influence factor of the j-th wind turbine.

[0138] In this embodiment, the wake impact factor of the wind turbine is 0.25, the system regulation cost factor of the wind turbine is 0.1, the neighbor coupling impact factor of the wind turbine is 0.4, and the aerodynamic weight coefficient is 0.7, the grid regulation cost weight coefficient is 0.1, and the neighbor coupling impact weight coefficient is 0.2. The corresponding wind turbine composite impact value is 0.235.

[0139] In this embodiment, the weight coefficients in the composite influence value acquisition formula are specifically:

[0140]

[0141] in: is the aerodynamic weight coefficient, is the grid regulation cost weight coefficient, is the neighbor coupling influence weight coefficient, is the Sigmoid function, 、 is the inversion of the LSTM network weight matrix, is the hidden state of the LSTM network, 、 is the bias term.

[0142] In this embodiment, the coordinated maintenance module is used to determine the credit score corresponding to each wind turbine based on the composite impact value obtained in step SB4, and sort the maintenance order of each wind turbine based on the credit score of each wind turbine. The details are as follows:

[0143] Step SC1: Determine the penalty result. That is, based on the composite impact value obtained in step SB4, the corresponding penalty result is obtained through the blockchain smart contract. The details are as follows:

[0144] Step SC1.1: Determine the pledge rate. That is, adjust the pledge rate of the on-chain asset pledge mechanism in the blockchain smart contract using the composite impact value obtained in step SB4. Specifically,

[0145]

[0146] in: is the adjusted pledge rate, is the initial pledge rate, is the composite impact value of the j-th wind turbine, is the growth rate of the pledge rate.

[0147] During the specific implementation process, the composite impact value of the third wind turbine is 18. At the same time, the initial pledge rate is 5%, and it increases at a rate of 2%. The corresponding adjusted pledge rate is 9%.

[0148] Step SC1.2: Determine the penalty. Compare the composite impact value obtained in SB4 with the preset impact threshold, and compare the neighboring aggregate efficiency loss between adjacent wind turbines with the preset efficiency loss threshold. Based on the comparison results, determine the corresponding penalty. Specifically:

[0149] If the obtained composite impact value exceeds the preset impact threshold, and the aggregate efficiency loss between adjacent wind turbines exceeds the preset efficiency loss threshold, the corresponding penalty will be deducted from the pledged assets through the automatic arbitration logic in the blockchain smart contract. Otherwise, no penalty will be deducted from the pledged assets.

[0150] Furthermore, the corresponding credit score is determined based on the deducted fine and the current pledge amount, specifically:

[0151]

[0152] in: For the final credit score, is the weight coefficient of fine proportion, For fines, is the current total amount of pledge, is the weight coefficient of delayed maintenance times, The number of delayed maintenance.

[0153] Step SC2: Tiered Management. This involves classifying wind turbines into multiple tiers based on the credit scores obtained in step SC1.2. Specifically, all wind turbines with a credit score less than 70 are classified as tier D, all wind turbines with a credit score between 70 and 79 are classified as tier C, all wind turbines with a credit score between 80 and 89 are classified as tier B, and all wind turbines with a credit score between 90 and 100 are classified as tier A.

[0154] Step SC3: Determine the maintenance score. That is, based on the levels determined in step SC2, the composite impact value obtained in step SB4 and the neighbor impact weight coefficient obtained are used to determine the maintenance score of each wind turbine, and the wind turbines in each level are prioritized.

[0155] In this embodiment, the formula for obtaining the maintenance priority score of the wind turbine is specifically:

[0156]

[0157] in: Score the maintenance priority, is the weight coefficient of the composite impact value, is the composite impact value of the j-th wind turbine, For the owner's credit score, is the neighbor influence weight coefficient, Efficiency loss for neighbors.

[0158] refer to Figure 4 , Figure 4 is the relationship diagram between credit score and penalty in this embodiment, Figure 4 It can be seen that credit points decrease as penalties accumulate, and some wind turbines are penalized because their combined impact value and neighboring efficiency loss exceed the threshold, resulting in a decrease in credit points. Furthermore, penalties only apply to wind turbines whose combined impact value and neighboring efficiency loss both exceed the threshold. In other words, the technical solution in this embodiment uses the threshold logic of the credit penalty mechanism to classify wind turbine credit levels.

[0159] refer to Figure 5 , Figure 5 This is the final maintenance sorting graph in this embodiment, Figure 5 It can be seen that after the wind turbines are grouped according to their credit ratings, the wind turbines in the group can be further sorted in descending order according to their maintenance priority scores, so that wind turbines with high maintenance priority scores can be maintained first.

[0160] Example 2

[0161] This embodiment provides a maintenance plan management system for clustered micro-wind power generation terminals. The specific implementation method is the same as that of Example 1, except that, in step SB1, the influencing factors are obtained by constructing a Jensen model. The present invention is described below with reference to the specific implementation methods of this embodiment.

[0162] In this embodiment, the impact factors are obtained by constructing the Jensen model, as follows:

[0163] Step SB1.1: Data processing. This involves obtaining the corresponding turbulence intensity and upstream freestream wind speed based on the SCADA data and the LiDAR wake scan data. Specifically,

[0164]

[0165] in: is the turbulence intensity of the fan, is the upstream free stream wind speed, is the statistical time window, is the instantaneous radial wind speed, is the average radial wind speed, is the fan output power, is the air density, is the impeller swept area, is the wind energy utilization coefficient.

[0166] During the specific implementation, the average radial wind speed was 8m / s within 10 minutes, and the standard deviation of the radial wind speed was 1.2m / s, which corresponds to a turbulence intensity of 15%. Furthermore, the wind turbine output power is 2MW, and the air density is 1.225kg / m 3 , the impeller swept area is 11310m 2 , the wind energy utilization coefficient is set to 0.42, and the corresponding upstream free stream wind speed is 8.3m / s.

[0167] Step SB1.2: Construct the wake field. That is, based on the turbine wake radius and the constructed wind column coordinate system, the turbulence intensity and upstream free stream wind speed obtained in step SB1.1 are used to obtain the wake velocity distribution corresponding to each grid point. Specifically, it is:

[0168]

[0169] in: is the wake velocity distribution, is the upstream free stream wind speed, is the initial wake radius, is the downstream axial distance, is the turbulence intensity of the fan, is the fan rotor swept diameter, is the vertical distance that the target position deviates from the centerline of the wake.

[0170] During the specific implementation process, the upstream free stream wind speed is 10m / s, the turbulence intensity is 10%, the vertical distance of the target position from the centerline of the wake is 60m, the initial radius of the wake is 36m, and the turbulence correction term is 1.06. The corresponding wake velocity distribution is 1.1m / s.

[0171] Step SB1.3: Generate wake impact factor. That is, based on the wake velocity distribution and the total number of wind turbines obtained in step SB1.2, determine the corresponding wake impact factor, specifically:

[0172]

[0173] in: is the wake influence factor of the j-th wind turbine, is the total number of fans, is the equivalent wake velocity of the jth affected wind turbine, is the upstream free stream wind speed, The index of the fan.

[0174] Example 3

[0175] This embodiment provides a maintenance plan management system for clustered micro-wind power generation terminals. Its specific implementation method is the same as that of Example 1, except that, in step SC1.2, the corresponding credit score is obtained based on the determined fine. The present invention is described below with reference to the specific implementation of this embodiment.

[0176] In this embodiment, the corresponding credit score is obtained by determining the fine, as follows:

[0177] Step SC1.2.1: Determine the penalty coefficient. That is, based on the composite impact value obtained in SB4, obtain the corresponding penalty coefficient, specifically:

[0178]

[0179] in: is the penalty coefficient, is the weight coefficient of the composite impact value, is the composite impact value of the j-th wind turbine, is the progressive penalty base, The number of delayed maintenance.

[0180] Step SC1.2.2: Determine the penalty. That is, based on the penalty coefficient obtained in step SC1.2.1 and the basic fee of the pledged assets, determine the corresponding penalty. Specifically:

[0181]

[0182] in: For fines, As the basic fee, is the penalty coefficient.

[0183] Step SC1.2.3: Determine the credit score change. That is, based on the fine and pledge amount obtained in step SC1.2.2, determine the corresponding credit score change, specifically:

[0184]

[0185] in: is the credit score change, is the weight coefficient of fine proportion, For fines, is the current total amount of pledge, is the weight coefficient of delayed maintenance times, The number of delayed maintenance.

[0186] During the specific implementation process, the weight coefficient of the penalty ratio is set to -10, and the weight coefficient of the delayed maintenance times is set to -5. At the same time, the penalty is 10,000 yuan, and the current total pledge amount is 100,000 yuan. The corresponding credit score change is -7.

[0187] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is limited by the accompanying embodiments and their equivalents.

Claims

1. A cluster-type wind power generation terminal maintenance plan management system, characterized in that: Includes: The data fusion module fuses SCADA data, lidar wake scans, and power grid PMU data to obtain the health status vector; An impact assessment module obtains a fan adjustment cost factor of the fan according to the health state vector, and determines a composite impact value of each fan through an LSTM network model; The coordinated maintenance module determines the credit score of each wind turbine based on the composite impact value and sorts the maintenance order of each wind turbine based on the credit score, including: SC1: Determine the penalty result: Based on the composite impact value, obtain the credit score of each wind turbine through the blockchain smart contract; SC2: Classification management: Wind turbines are classified into different levels according to the credit score; SC3: Determine maintenance score: Obtain a maintenance score for each wind turbine based on the composite impact value and the neighbor impact weight coefficient, and rank all wind turbines in each level based on the maintenance score. The formula for obtaining the maintenance score is as follows: in: Score the maintenance priority, is the weight coefficient of the composite impact value, is the composite impact value of the j-th wind turbine, For the owner's credit score, is the neighbor influence weight coefficient, Efficiency loss for neighbors.

2. A clustered wind power generation terminal maintenance plan management system according to claim 1, characterized in that: Get the health status vector, including: SA1: Spatiotemporal alignment processing: Based on the SCADA data, LiDAR wake scan and power grid PMU data, a coordinate system conversion matrix is ​​established to obtain the aligned data coordinates, specifically: in: is the coordinate of the target point in the global coordinate system, is the rotation matrix, is the radial distance of the target point measured by the lidar, is the azimuth of the target point measured by the lidar, is the height of the target point measured by the lidar, is the translation vector; SA2: Kalman filter processing: By improving the Kalman filter algorithm and aligning the data coordinates, the Mahalanobis distance is obtained, specifically: in: is the Mahalanobis distance, is the inverse of the residual vector, is the residual vector, is the residual covariance matrix, is the observation matrix, is the inversion of the observation matrix, is the state estimation covariance matrix, is the observation noise covariance matrix; SA3: Determine parameter weights: Based on the Mahalanobis distance, determine the final comprehensive health of each wind turbine, specifically: in: is the comprehensive health of the j-th wind turbine, is the Mahalanobis distance, is the preset distance threshold, is the mechanical health of the j-th fan, is the electrical health of the j-th wind turbine, is the turbulence intensity of the j-th wind turbine, is the baseline turbulence intensity; SA4: Determine the health state vector: Determine the final health state vector based on the final comprehensive health level, specifically: in: is the health status vector, is the final comprehensive health of the first wind turbine, is the final comprehensive health of the nth wind turbine.

3. A clustered wind power generation terminal maintenance plan management system according to claim 2, characterized in that: The Mahalanobis distance is compared with a preset distance threshold, and the weight of the comprehensive health is adjusted according to the comparison result, specifically: When the Mahalanobis distance is greater than a preset distance threshold, the weight of the comprehensive health is reduced; otherwise, the weight of the comprehensive health remains unchanged.

4. A clustered wind power generation terminal maintenance plan management system according to claim 2 or 3, characterized in that: According to the adjusted weights in the comprehensive health, the mechanical health and electrical health of the wind turbine are obtained, specifically: in: is the mechanical health of the fan, The electrical health of the wind turbine, is the actual temperature rise, is the maximum allowable temperature rise, is the effective value of vibration, is the vibration critical value, is 3 times the frequency energy, is the baseline band energy, is the current insulation resistance, is the insulation resistance of the new equipment, is the current harmonic distortion rate, is the maximum allowable current harmonic distortion rate.

5. A clustered wind power generation terminal maintenance plan management system according to claim 1, characterized in that: Determine the weight distribution, including: SB1: Determine the impact factor: Use the health status vector as the input of the constructed Jensen model and output the impact factor, specifically: in: is the wake influence factor of the j-th wind turbine, is the total number of fans, is the equivalent wake velocity of the jth affected wind turbine, is the upstream free stream wind speed, is the index of the fan; SB2: Impact Factor Update: The turbulence correction coefficient of the Jensen model is modified through real-time CFD verification model, and the Jensen model is updated based on the modified turbulence correction coefficient. At the same time, an updated wake impact factor is obtained based on the updated Jensen model; SB3: Determine the adjustment cost factor: According to the health state vector, obtain the fan adjustment cost factor, specifically: in: is the system adjustment cost factor of the j-th fan, is the grid frequency deviation, is the total number of fans, is the index of the fan, is the loss coefficient matrix element between the i-th fan and the j-th fan, is the output power of the jth wind turbine, is the linear loss coefficient of the i-th fan; SB4: Determine the composite impact value: Based on the updated wake impact factor and wind turbine adjustment cost factor, the composite impact value is obtained through the LSTM network model, specifically: in: is the composite impact value of the j-th wind turbine, is the wake influence factor of the j-th wind turbine, is the system adjustment cost factor of the j-th fan, is the aerodynamic weight coefficient, is the grid regulation cost weight coefficient, is the neighbor coupling influence weight coefficient, is the neighbor coupling influence factor of the j-th wind turbine.

6. A clustered wind power generation terminal maintenance plan management system according to claim 5, characterized in that: Output acquisition impact factors, including: SB1.1: Data processing: Based on SCADA data and lidar wake scan data, obtain turbulence intensity and upstream free stream wind speed, specifically: in: is the turbulence intensity of the fan, is the upstream free stream wind speed, is the statistical time window, is the instantaneous radial wind speed, is the average radial wind speed, is the fan output power, is the air density, is the impeller swept area, is the wind energy utilization coefficient; SB1.2: Construct wake field: Based on the turbine wake radius and the constructed wind column coordinate system, obtain the wake velocity distribution at each grid point, specifically: in: is the wake velocity distribution, is the upstream free stream wind speed, is the initial wake radius, is the downstream axial distance, is the turbulence intensity of the fan, is the fan rotor swept diameter, is the vertical distance of the target position from the centerline of the wake; SB1.3: Generate wake impact factor: Determine the wake impact factor based on the wake velocity distribution and the total number of wind turbines.

7. A clustered wind power generation terminal maintenance plan management system according to claim 5, characterized in that: Get the updated wake impact factors, including: SB2.1: Simulation trigger: Compare the wake impact factor with the simulated impact factor of the real-time CFD verification model to obtain the impact factor difference. At the same time, compare the impact factor difference with the preset difference threshold. Based on the comparison result, simulate the real-time CFD verification model. Specifically: When the difference in the impact factor is greater than a preset difference threshold, the simulation of the real-time CFD verification model is performed; otherwise, the simulation of the real-time CFD verification model is not performed; SB2.2: Determine model error: Based on the simulation impact factor and the wake impact factor, determine the relative model error, specifically: in: is the relative error of the model, Simulation impact factors for real-time CFD validation models, is the wake influence factor of the Jensen model; SB2.3: Determine the updated turbulence correction factor: Compare the model relative error with a preset error threshold and update the turbulence correction factor based on the comparison result, specifically: When the model relative error is greater than a preset error threshold, the turbulence correction coefficient is updated; otherwise, the turbulence correction coefficient is not updated; The updating formula of the turbulence correction coefficient is specifically: in: is the updated turbulence correction coefficient, is the initial turbulence correction coefficient, is the relative error of the model.

8. The clustered wind power generation terminal maintenance plan management system according to claim 1, characterized in that: Get the credit score of each wind turbine, including: SC1.1: Determine the pledge rate: Based on the composite impact value, adjust the pledge rate of the on-chain asset pledge mechanism, specifically: in: is the adjusted pledge rate, is the initial pledge rate, is the composite impact value of the j-th wind turbine, is the growth rate of the pledge rate; SC1.2: Determine Penalty: Compare the composite impact value with the preset impact threshold, and compare the neighbor aggregation efficiency loss with the preset efficiency loss threshold. Based on the comparison results, determine the penalty. Based on the penalty and the current pledge amount, determine the credit score. Specifically: in: For the final credit score, is the weight coefficient of fine proportion, For fines, is the current total amount of pledge, is the weight coefficient of delayed maintenance times, The number of delayed maintenance.

9. A clustered wind power generation terminal maintenance plan management system according to claim 8, characterized in that: The composite impact value is compared with a preset impact threshold, and the neighbor aggregation efficiency loss is compared with a preset efficiency loss threshold. Based on the comparison results, a penalty is determined, specifically: When the composite impact value is greater than the preset impact threshold and the neighbor aggregation efficiency loss is greater than the preset efficiency loss threshold, a fine is deducted through the blockchain smart contract; otherwise, no fine is deducted.

10. A clustered wind power generation terminal maintenance plan management system according to claim 8 or 9, characterized in that: The credit score is determined based on the fine and the current pledge amount, including: SC1.2.1: Determine the penalty coefficient: Based on the composite impact value, obtain the penalty coefficient, specifically: in: is the penalty coefficient, is the weight coefficient of the composite impact value, is the composite impact value of the j-th wind turbine, is the progressive penalty base, The number of delayed maintenance; SC1.2.2: Determine the penalty: The penalty is determined based on the penalty coefficient and the base fee of the pledged assets, specifically: in: For fines, As the basic fee, is the penalty coefficient; SC1.2.3: Determine the credit score change: Based on the fine and pledge amount, determine the credit score change, specifically: in: is the credit score change, is the weight coefficient of fine proportion, For fines, is the current total amount of pledge, is the weight coefficient of delayed maintenance times, The number of delayed maintenance.

Images