Early warning method and system for fan blade state monitoring

Abstract

The invention relates to the technical field of fan blade state monitoring, and discloses an early warning method and system for fan blade state monitoring, and the method comprises the steps: obtaining basic information and environment information of a fan blade, screening out historical similar monitoring logs according to the basic information and the environment information, and storing the historical similar monitoring logs in a database; determining a plurality of feature monitoring schemes and feature monitoring data corresponding to the feature monitoring schemes according to the historical similar monitoring logs; acquiring real-time feature monitoring data of the fan blade according to each feature monitoring scheme, and inputting the real-time feature monitoring data into a preset state evaluation model of the corresponding feature monitoring scheme to obtain a real-time state evaluation value; the comprehensive state evaluation value is generated according to the real-time state evaluation values of all the characteristic monitoring schemes, whether the early warning instruction is generated or not is judged according to the comprehensive state evaluation value and the state evaluation change characteristics, the early warning accuracy and operation and maintenance timeliness of blade state monitoring are improved, and normal and stable operation of the fan blade is guaranteed.

Description

Technical Field

[0001] The present application relates to the technical field of fan blade condition monitoring, and particularly to an early warning method and system for fan blade condition monitoring. Background Art

[0002] As one of the key components of a fan, the condition of the fan blade directly affects the power generation efficiency of the fan. Therefore, higher requirements are put forward for the condition monitoring of the fan blade.

[0003] In the prior art, the condition monitoring of fan blades mostly relies on single-sensor monitoring and early warning, without selecting a reasonable monitoring scheme according to the actual situation, resulting in low monitoring efficiency, inability to accurately feedback the actual condition of the fan blade and timely formulate a reasonable operation and maintenance strategy, reducing the early warning accuracy of blade condition monitoring and the timeliness of operation and maintenance, and unable to ensure the normal and stable operation of the fan blade. Summary of the Invention

[0004] To solve the above technical problems, the present application provides an early warning method and system for fan blade condition monitoring. By selecting a characteristic monitoring scheme and characteristic monitoring data, constructing a corresponding preset state evaluation model, obtaining a real-time state evaluation value and a comprehensive state evaluation value, judging whether to send an early warning signal according to the comprehensive state evaluation value and the state evaluation change characteristic, and performing timely operation according to the early warning signal, the early warning accuracy of blade condition monitoring and the timeliness of operation and maintenance are improved, and the normal and stable operation of the fan blade is ensured.

[0005] In some embodiments of the present application, an early warning method for fan blade condition monitoring is provided, including: Obtain the basic information and environmental information of the fan blade, screen out historical similar monitoring logs according to the basic information and environmental information, and determine a number of characteristic monitoring schemes and the characteristic monitoring data corresponding to the characteristic monitoring schemes according to the historical similar monitoring logs; Obtain the real-time characteristic monitoring data of the fan blade according to each characteristic monitoring scheme, and input it into the preset state evaluation model corresponding to the characteristic monitoring scheme to obtain a real-time state evaluation value; Generate a comprehensive state evaluation value according to the real-time state evaluation values of all characteristic monitoring schemes, and judge whether to generate an early warning instruction according to the comprehensive state evaluation value and the state evaluation change characteristic.

[0006] In some embodiments of the present application, screening out historical similar monitoring logs according to the basic information and environmental information includes: The basic information of the fan blade includes structural information, life information, and operation and maintenance information; The environmental information of the fan blade includes weather condition information, wind speed information, temperature information, and terrain information; Obtain the historical monitoring logs of fan blades of the same type, extract the historical basic information and historical environmental information in each historical monitoring log, and perform similarity analysis with the current basic information and environmental information to obtain a comprehensive similarity coefficient; Preset a similarity coefficient threshold; Set the historical monitoring logs with a comprehensive similarity coefficient greater than the similarity coefficient threshold as the historical similar monitoring logs of the current fan blade; Obtain the historical status evaluation value of each historical similar monitoring log and the credibility of the corresponding historical status evaluation value, and set the monitoring scheme corresponding to the historical similar monitoring log with a credibility greater than the preset credibility threshold as the feature monitoring scheme; Analyze the historical status evaluation value and historical monitoring data in the historical similar monitoring log of the feature monitoring scheme, and determine the feature monitoring data corresponding to the feature monitoring scheme according to the analysis result.

[0007] In some embodiments of the present application, determining the feature monitoring data corresponding to the feature monitoring scheme according to the analysis result includes: Establish a time reference line based on the historical monitoring duration of the historical similar monitoring log of each feature monitoring scheme, and set data collection nodes at preset time intervals; Obtain the historical monitoring data and historical status evaluation value in the corresponding historical similar monitoring log according to the data collection nodes, and map them to the corresponding time reference line to obtain a historical status evaluation value - monitoring data analysis chart of several historical similar monitoring logs of the same feature monitoring scheme; Obtain the curve change trend of the historical status evaluation value of the same historical status evaluation value - monitoring data analysis chart, and calculate the similarity degree between the curve change trend of each historical monitoring data and the curve change trend of the historical status evaluation value; Screen out the historical monitoring data with a similarity degree greater than the similarity degree threshold, and calculate the trend delay duration between the curve change trend of the screened historical monitoring data and the curve change trend of the historical status evaluation value; If the trend delay duration is less than the preset duration threshold, set the corresponding historical monitoring data as the pending feature monitoring data of the corresponding historical similar monitoring log; Compare the pending feature monitoring data of several historical similar monitoring logs of the same feature monitoring scheme to obtain the repetition ratio of the same pending feature monitoring data in different historical similar monitoring logs; Set the pending feature monitoring data with a repetition ratio greater than the preset ratio threshold as the feature monitoring data of the corresponding feature monitoring scheme; Generate a weight coefficient corresponding to the feature monitoring data according to the similarity degree, trend delay duration, and repetition ratio of each feature monitoring data of the same feature monitoring scheme.

[0008] In some embodiments of the present application, the calculation formula for the weight coefficient of the feature monitoring data is as follows: ; where Q is the weight coefficient of the feature monitoring data, z1 is the weight coefficient of the similarity degree, n1 is the number of historical similarity monitoring logs where the feature monitoring data with a similarity degree greater than the similarity degree threshold is located, N is the total number of historical similarity monitoring logs of the same feature monitoring scheme, is the similarity degree of the feature monitoring data in the i1-th historical similarity monitoring log, X0 is the similarity degree threshold, a1 is the first weight conversion coefficient, z2 is the weight coefficient of the trend delay duration, n2 is the number of historical similarity monitoring logs where the feature monitoring data with a trend delay duration less than the preset duration threshold is located, is the trend delay duration of the feature monitoring data in the i2-th historical similarity monitoring log, X0 is the preset duration threshold, a2 is the second weight conversion coefficient, z3 is the weight coefficient of the repetition ratio, b is the repetition ratio, B0 is the preset ratio threshold, and a3 is the third weight conversion coefficient.

[0009] In some embodiments of the present application, a comprehensive state evaluation value is generated based on the real-time state evaluation values of all feature monitoring schemes, including: Neural network training is performed according to the feature monitoring data of the same feature monitoring scheme, the weight coefficient of the corresponding feature monitoring data, and the historical state evaluation value to obtain a preset state evaluation model for the corresponding feature monitoring scheme; Based on each feature monitoring scheme, the real-time feature monitoring data of the corresponding wind turbine blade is obtained, and the real-time feature monitoring data is input into the preset state evaluation model of the corresponding feature monitoring scheme to obtain a real-time state evaluation value; Calculate the credibility difference with a credibility greater than the credibility threshold for each feature monitoring scheme; Generate the weight coefficient of each feature monitoring scheme according to the number of all feature monitoring schemes and the corresponding credibility difference; Generate a comprehensive state evaluation value according to the real-time state evaluation value of each feature monitoring scheme and the corresponding weight coefficient.

[0010] In some embodiments of the present application, a preset monitoring time node for the current monitoring period is preset in advance, and a state evaluation value change curve is constructed according to the time sequence of the preset monitoring time nodes, the historical comprehensive state evaluation values corresponding to the preset monitoring time nodes, and the comprehensive state evaluation value of the current monitoring time node; Perform curve trend extrapolation on the state evaluation value change curve to obtain a predicted state evaluation value change curve for the future period; Obtain the predicted change characteristics of the predicted state evaluation value change curve, and the predicted change characteristics include a predicted change trend and a predicted change rate; Compare the predicted state evaluation value change curve with the comprehensive state evaluation value at the current monitoring time node to obtain the first comprehensive state evaluation value difference and the second comprehensive state evaluation value difference, and calculate the difference magnitude between the first comprehensive state evaluation value difference and the second comprehensive state evaluation value difference; Set the predicted change trend, predicted change rate, and difference magnitude as state evaluation features, and generate a compensation coefficient for the comprehensive state evaluation value at the current monitoring time node according to the state evaluation features.

[0011] In some embodiments of the present application, determining whether to generate a warning instruction according to the comprehensive state evaluation value and the state evaluation change characteristics includes: Correct the comprehensive state evaluation value according to the compensation coefficient to obtain the corrected comprehensive state evaluation value; Preset a state evaluation threshold; If the corrected comprehensive state evaluation value is greater than the state evaluation threshold, do not generate a warning instruction; If the corrected comprehensive state evaluation value is less than the state evaluation threshold, generate a warning instruction.

[0012] In some embodiments of the present application, the warning instruction includes: Compare the real-time feature monitoring data obtained by each feature monitoring scheme with the corresponding standard monitoring data to obtain the real-time feature monitoring data difference; Determine the abnormal feature monitoring data and the abnormal level in each feature monitoring scheme according to the real-time feature monitoring data difference; Compare the abnormal feature monitoring data with the fault monitoring data set of the historical fault type of the corresponding feature monitoring scheme to obtain the correlation degree between the current abnormal feature monitoring data and the fault monitoring data set of the historical fault type; Set the historical fault type with the largest correlation degree as the predicted first fault type of the abnormal feature monitoring data of the corresponding feature monitoring scheme; Analyze multiple predicted first fault types according to the credibility of all feature monitoring schemes to determine the predicted second fault type; Screen out the abnormal feature monitoring data of the feature monitoring scheme where the predicted first fault type is consistent with the predicted second fault type, and determine multiple predicted first abnormal areas of the fan blade according to the data source and sensor type of the screened abnormal feature monitoring data; Analyze multiple predicted first abnormal areas according to the weight coefficient and abnormal level of the screened abnormal feature monitoring data to determine the predicted second abnormal area; Predict the predicted time interval when obvious fault characteristics appear in the corresponding abnormal feature monitoring data according to the data change characteristics of the abnormal feature monitoring data in the current monitoring period; Based on the preset operation and maintenance analysis model of the wind turbine blade, analyze the predicted second fault type, the predicted second abnormal area, and the predicted time interval when obvious fault characteristics appear in the monitoring data of each abnormal feature, and obtain multiple preset operation and maintenance strategies; Based on the objective function of maximizing the service life of the wind turbine blade and the objective function of minimizing the operation and maintenance cost, analyze multiple preset operation and maintenance strategies to determine the optimal operation and maintenance strategy; Generate a warning instruction according to the predicted second fault type, the predicted second abnormal area, and the optimal operation and maintenance strategy.

[0013] In some embodiments of the present application, there is also a warning system for monitoring the state of the wind turbine blade: An acquisition module, configured to acquire the basic information and environmental information of the wind turbine blade, screen out historical similar monitoring logs according to the basic information and environmental information, and determine several feature monitoring schemes and the feature monitoring data corresponding to the feature monitoring schemes; A state evaluation module, configured to obtain the real-time feature monitoring data of the wind turbine blade according to each feature monitoring scheme, and input it into the preset state evaluation model corresponding to the feature monitoring scheme to obtain the real-time state evaluation value; A warning module, configured to generate a comprehensive state evaluation value according to the real-time state evaluation values of all feature monitoring schemes, and determine whether to generate a warning instruction according to the comprehensive state evaluation value and the state evaluation change characteristics..

[0014] A warning method and system for monitoring the state of a wind turbine blade according to an embodiment of the present application, compared with the prior art, its beneficial effects are as follows: By selecting the feature monitoring scheme and the feature monitoring data, constructing the corresponding preset state evaluation model, obtaining the real-time state evaluation value and the comprehensive state evaluation value, determining whether to send a warning signal according to the comprehensive state evaluation value and the state evaluation change characteristics, and performing timely operation according to the warning signal, the warning accuracy of the blade state monitoring and the operation and maintenance timeliness are improved, and the normal and stable operation of the wind turbine blade is ensured. Description of the Drawings

[0015] Figure 1 is a flowchart of a warning method for monitoring the state of a wind turbine blade in an embodiment of the present application; Figure 2 is a schematic diagram of a warning system for monitoring the state of a wind turbine blade in an embodiment of the present application. Detailed Embodiments

[0016] The following combines the drawings and embodiments to further describe in detail the specific embodiments of the present application. The following embodiments are used to illustrate the present application, but are not used to limit the scope of the present application.

[0017] In the description of the present application, it should be understood that the orientation or positional relationship indicated by terms such as "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present application.

[0018] The terms "first" and "second" are only used for descriptive purposes and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present application, unless otherwise specified, the meaning of "a plurality" is two or more.

[0019] In the description of the present application, it should be noted that unless otherwise clearly specified and limited, the terms "installation", "connection", and "coupling" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific circumstances.

[0020] As Figure 1 shown, a warning method for monitoring the state of a wind turbine blade according to an embodiment of the present application includes: Step S101: Obtain the basic information and environmental information of the wind turbine blade, screen out the historical similar monitoring logs according to the basic information and environmental information, and determine several characteristic monitoring schemes and the characteristic monitoring data corresponding to the characteristic monitoring schemes according to the historical similar monitoring logs; Step S102: Obtain the real-time characteristic monitoring data of the wind turbine blade according to each characteristic monitoring scheme, and input it into the preset state evaluation model corresponding to the characteristic monitoring scheme to obtain the real-time state evaluation value; Step S103: Generate a comprehensive state evaluation value according to the real-time state evaluation values of all the characteristic monitoring schemes, and determine whether to generate a warning instruction according to the comprehensive state evaluation value and the state evaluation change characteristics.

[0021] In some embodiments of the present application, screening out the historical similar monitoring logs according to the basic information and environmental information includes: The basic information of the wind turbine blade includes structural information, life information, and operation and maintenance information; The environmental information of the wind turbine blade includes weather condition information, wind speed information, temperature information, and terrain information; Obtain the historical monitoring logs of fan blades of the same type, extract the historical basic information and historical environmental information in each historical monitoring log, and perform similarity analysis with the current basic information and environmental information to obtain a comprehensive similarity coefficient; Preset a similarity coefficient threshold in advance; Set the historical monitoring logs with a comprehensive similarity coefficient greater than the similarity coefficient threshold as the historical similar monitoring logs of the current fan blade; Obtain the historical status evaluation values of each historical similar monitoring log and the credibility of the corresponding historical status evaluation value, and set the monitoring scheme corresponding to the historical similar monitoring log with a credibility greater than the preset credibility threshold as the feature monitoring scheme; Analyze the historical status evaluation values and historical monitoring data in the historical similar monitoring logs of the feature monitoring scheme, and determine the feature monitoring data corresponding to the feature monitoring scheme according to the analysis results.

[0022] In this embodiment, the credibility refers to the historical accuracy of the feature monitoring scheme for the state evaluation of the fan blade. When the historical accuracy is greater, the credibility is greater, and vice versa.

[0023] In this embodiment, similarity analysis is performed on the historical basic data, historical environmental information and the current basic information and environmental information to obtain historical similar monitoring logs that are highly similar to the structure, life, weather conditions, and terrain of the current fan blade, so as to obtain the corresponding feature monitoring scheme and feature monitoring data, ensure the accuracy of the monitoring scheme, lay a foundation for the accurate evaluation and early warning of the subsequent fan blade state, and improve the accuracy of the fan blade state early warning.

[0024] In some embodiments of the present application, determining the feature monitoring data corresponding to the feature monitoring scheme according to the analysis results includes: Establish a time reference line based on the historical monitoring duration of the historical similar monitoring logs of each feature monitoring scheme, and set data acquisition nodes at preset time intervals; Obtain the historical monitoring data and historical status evaluation values in the corresponding historical similar monitoring logs according to the data acquisition nodes, and map them to the corresponding time reference line to obtain a historical status evaluation value - monitoring data analysis diagram of several historical similar monitoring logs of the same feature monitoring scheme; Obtain the curve change trend of the historical status evaluation values of the same historical status evaluation value - monitoring data analysis diagram, and calculate the similarity degree between the curve change trend of each historical monitoring data and the curve change trend of the historical status evaluation value; Screen out the historical monitoring data with a similarity degree greater than the similarity degree threshold, and calculate the trend delay duration between the curve change trend of the screened historical monitoring data and the curve change trend of the historical status evaluation value; If the trend delay duration is less than the preset duration threshold, the corresponding historical monitoring data is set as the pending feature monitoring data of the corresponding historical similar monitoring log; Compare the pending feature monitoring data of several historical similar monitoring logs of the same feature monitoring scheme to obtain the repetition ratio of the same pending feature monitoring data in different historical similar monitoring logs; Set the pending feature monitoring data with a repetition ratio greater than the preset ratio threshold as the feature monitoring data of the corresponding feature monitoring scheme; Generate the weight coefficient of the corresponding feature monitoring data according to the similarity degree, trend delay duration and repetition ratio of each feature monitoring data of the same feature monitoring scheme.

[0025] In this embodiment, the curve change trend includes but is not limited to the curve form, the degree of fluctuation and the change direction. The trend delay duration refers to the time interval of the curve change trend in which the historical monitoring data has a large similarity degree with the historical state evaluation value, and the curve change trend of the historical state evaluation value changes with the curve change trend of the historical monitoring data.

[0026] In this embodiment, the similarity degree threshold, the preset duration threshold and the preset ratio threshold are all set in advance. When the similarity degree is greater than the similarity degree threshold and the trend extension duration is less than the preset duration threshold, it indicates that the corresponding historical monitoring data has a great influence on the historical state evaluation value, that is, the pending feature monitoring data in the current historical similar monitoring log.

[0027] In this embodiment, the repetition ratio refers to the ratio of the number of repeated occurrences of the same pending feature monitoring data in different historical similar monitoring logs to the total number of historical similar monitoring logs.

[0028] In this embodiment, the feature monitoring data is determined by calculating the influence degree of each historical monitoring data on the state evaluation value of the fan blade, and the corresponding preset state evaluation model is constructed, which lays a foundation for subsequent data collection and analysis, reduces the data processing volume and analysis volume, and improves the state evaluation and early warning efficiency of the fan blade.

[0029] In some embodiments of the present application, the calculation formula for the weight coefficient of the feature monitoring data is: ; where Q is the weight coefficient of the feature monitoring data, z1 is the weight coefficient of the similarity degree, n1 is the number of historical similar monitoring logs where the feature monitoring data with a similarity degree greater than the similarity degree threshold is located, and N is the total number of historical similar monitoring logs of the same feature monitoring scheme. Let \(s_i\) be the similarity degree of the feature monitoring data in the \(i_1\)-th historical similar monitoring log, \(X_0\) be the similarity degree threshold, \(a_1\) be the first weight conversion coefficient, \(z_2\) be the weight coefficient of the trend delay duration, and \(n_2\) be the number of historical similar monitoring logs where the feature monitoring data with a trend delay duration less than the preset duration threshold is located. Let \(t_{i_2}\) be the trend delay duration of the feature monitoring data in the \(i_2\)-th historical similar monitoring log, \(X_0\) be the preset duration threshold, \(a_2\) be the second weight conversion coefficient, \(z_3\) be the weight coefficient of the repetition ratio, \(b\) be the repetition ratio, \(B_0\) be the preset ratio threshold, and \(a_3\) be the third weight conversion coefficient.

[0030] In this embodiment, the first weight conversion coefficient, the second weight conversion coefficient, and the third weight conversion coefficient respectively refer to the weight coefficients that convert the similarity degree, the trend delay duration, and the repetition ratio into the same dimension. When the similarity degree is larger, the trend delay duration is smaller, and the repetition ratio is larger, the weight coefficient of the corresponding feature monitoring data is larger, improving the accuracy of the state evaluation of the fan blade.

[0031] In some embodiments of the present application, a comprehensive state evaluation value is generated according to the real-time state evaluation values of all feature monitoring schemes, including: Neural network training is performed according to the feature monitoring data of the same feature monitoring scheme, the weight coefficient of the corresponding feature monitoring data, and the historical state evaluation value to obtain the preset state evaluation model of the corresponding feature monitoring scheme; Based on each feature monitoring scheme, the real-time feature monitoring data of the corresponding fan blade is obtained, and the real-time feature monitoring data is input into the preset state evaluation model of the corresponding feature monitoring scheme to obtain the real-time state evaluation value; Calculate the credibility difference with a credibility greater than the credibility threshold for each feature monitoring scheme; Generate the weight coefficient of each feature monitoring scheme according to the number of all feature monitoring schemes and the corresponding credibility differences; Generate a comprehensive state evaluation value according to the real-time state evaluation value of each feature monitoring scheme and the corresponding weight coefficient.

[0032] In this embodiment, calculations are performed according to the real-time state evaluation values of multiple feature monitoring schemes and the corresponding weight coefficients to obtain a comprehensive state evaluation value, reducing the false alarm rate of the state evaluation and early warning of a single monitoring scheme. By screening reasonable monitoring schemes and comprehensively calculating the state evaluation value, the accuracy of the state evaluation and early warning of the fan blade is improved.

[0033] In some embodiments of the present application, a preset monitoring time node for the current monitoring period is preset, and a state evaluation value change curve is constructed according to the time sequence of the preset monitoring time nodes, the historical comprehensive state evaluation values corresponding to the preset monitoring time nodes, and the comprehensive state evaluation value of the current monitoring time node. Perform curve trend extrapolation on the curve of the state evaluation value change to obtain the predicted curve of the state evaluation value change in the future time period; Obtain the predicted change characteristics of the predicted curve of the state evaluation value change, where the predicted change characteristics include the predicted change trend and the predicted change rate; Compare the predicted curve of the state evaluation value change with the comprehensive state evaluation value at the current monitoring time node to obtain the first comprehensive state evaluation value difference and the second comprehensive state evaluation value difference, and calculate the difference magnitude between the first comprehensive state evaluation value difference and the second comprehensive state evaluation value difference; Set the predicted change trend, the predicted change rate, and the difference magnitude as the state evaluation characteristics, and generate a compensation coefficient for the comprehensive state evaluation value at the current monitoring time node according to the state evaluation characteristics.

[0034] In this embodiment, the predicted change trends include a general downward trend, an upward trend, a normal trend, and an abnormal trend. Corresponding selection coefficients are assigned to different predicted change trends, and corresponding quantization values are assigned to the predicted change rate. Among them, the selection coefficients assigned to the general downward trend and the abnormal trend are negative values, the selection coefficients assigned to the upward trend and the normal trend are positive values, and the value range of the selection coefficient is (-1, 1).

[0035] In this embodiment, the first comprehensive state evaluation value difference and the second comprehensive state evaluation value difference respectively refer to the maximum difference and the minimum difference between the predicted state evaluation value in the predicted curve of the state evaluation value change and the comprehensive state evaluation value, and the difference magnitude refers to the absolute value of the difference between the maximum difference and the minimum difference.

[0036] In this embodiment, when the corresponding selection coefficient assigned to the change trend is a positive value, the larger the quantization value assigned to the predicted change rate and the smaller the difference magnitude, the larger the corresponding compensation coefficient, and vice versa. The value range of the compensation coefficient is (-1, 1).

[0037] In some embodiments of the present application, judging whether to generate a warning instruction according to the comprehensive state evaluation value and the state evaluation change characteristics includes: Correct the comprehensive state evaluation value according to the compensation coefficient to obtain the corrected comprehensive state evaluation value; Preset a state evaluation threshold; If the corrected comprehensive state evaluation value is greater than the state evaluation threshold, no warning instruction is generated; If the corrected comprehensive state evaluation value is less than the state evaluation threshold, a warning instruction is generated.

[0038] In some embodiments of the present application, the warning instruction includes: Compare the real-time feature monitoring data obtained by each feature monitoring scheme with the corresponding standard monitoring data to obtain the difference in real-time feature monitoring data; Determine the abnormal feature monitoring data and the abnormal level in each feature monitoring scheme according to the difference in real-time feature monitoring data; Compare the abnormal feature monitoring data with the fault monitoring data set of the historical fault type corresponding to the feature monitoring scheme to obtain the degree of association between the current abnormal feature monitoring data and the fault monitoring data set of the historical fault type; Set the historical fault type with the largest degree of association as the predicted first fault type of the abnormal feature monitoring data corresponding to the feature monitoring scheme; Analyze multiple predicted first fault types according to the credibility of all feature monitoring schemes to determine the predicted second fault type; Screen out the abnormal feature monitoring data of the feature monitoring scheme where the predicted first fault type is consistent with the predicted second fault type, and determine multiple predicted first abnormal areas of the wind turbine blade according to the data source and sensor type of the screened abnormal feature monitoring data; Analyze multiple predicted first abnormal areas according to the weight coefficient and abnormal level of the screened abnormal feature monitoring data to determine the predicted second abnormal area; Predict the predicted time interval when obvious fault features appear for the corresponding abnormal feature monitoring data according to the data change characteristics of the abnormal feature monitoring data in the current monitoring period; Analyze the predicted second fault type, the predicted second abnormal area, and the predicted time interval when obvious fault features appear for each abnormal feature monitoring data based on the preset operation and maintenance analysis model of the wind turbine blade to obtain multiple preset operation and maintenance strategies; Analyze multiple preset operation and maintenance strategies based on the objective function of maximizing the service life of the wind turbine blade and the objective function of minimizing the operation and maintenance cost to determine the optimal operation and maintenance strategy; Generate a warning instruction according to the predicted second fault type, the predicted second abnormal area, and the optimal operation and maintenance strategy.

[0039] In this embodiment, when the credibility of the feature monitoring scheme is greater, it means that the accuracy of the corresponding predicted first fault type is higher. Screen out the predicted first fault type with the highest accuracy according to the repetition ratio of the predicted first fault type and the credibility of the corresponding feature monitoring scheme, and set it as the predicted second fault type, where the repetition ratio is the ratio of the number of occurrences of the same predicted first fault type to the number of all feature monitoring schemes.

[0040] In this embodiment, when the weight coefficient of the screened abnormal feature monitoring data is larger and the abnormal level is higher, it means that the accuracy of the corresponding predicted first abnormal area is larger. Set the predicted first abnormal area with the highest accuracy as the predicted second abnormal area.

[0041] In this embodiment, the prediction time interval refers to the time interval during which obvious faults occur in the abnormal feature monitoring data or cause obvious faults in the wind turbine blades.

[0042] In this embodiment, the preset operation and maintenance analysis model is constructed based on the historical fault types, historical abnormal areas of the wind turbine blades, and the corresponding historical operation and maintenance strategies, and combined with the corresponding prediction time interval. The multiple preset operation and maintenance strategies refer to the operation and maintenance strategies that meet the requirements of completing the prediction of the corresponding second fault type and the prediction of the second abnormal area before the prediction time interval.

[0043] In this embodiment, by determining the optimal operation and maintenance strategy, the operation and maintenance objectives of predicting the second fault type and the second abnormal area are completed as much as possible before the prediction time interval, the service life of the wind turbine blades is maximally maintained, the operation and maintenance costs are reduced, the use efficiency of the wind turbine blades is improved, and the early warning efficiency is maximized.

[0044] In some embodiments of the present application, as Figure 2 shown, there is also a warning system for monitoring the state of wind turbine blades: An acquisition module, configured to acquire the basic information and environmental information of the wind turbine blades, screen out the historical similar monitoring logs according to the basic information and environmental information, and determine a number of feature monitoring schemes and the feature monitoring data corresponding to the feature monitoring schemes; A state evaluation module, configured to obtain the real-time feature monitoring data of the wind turbine blades according to each feature monitoring scheme, and input the data into the preset state evaluation model corresponding to the feature monitoring scheme to obtain a real-time state evaluation value; A warning module, configured to generate a comprehensive state evaluation value according to the real-time state evaluation values of all feature monitoring schemes, and determine whether to generate a warning instruction according to the comprehensive state evaluation value and the state evaluation change characteristics.

[0045] The above are only the preferred embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present application, several improvements and replacements can be made, and these improvements and replacements should also be regarded as the protection scope of the present application.

Claims

1. An early warning method for monitoring the status of wind turbine blades, characterized in that: include: Obtain basic information and environmental information of the wind turbine blades, filter out historical similar monitoring logs based on the basic information and environmental information, and determine several characteristic monitoring schemes and characteristic monitoring data corresponding to the characteristic monitoring schemes based on the historical similar monitoring logs; Acquire real-time characteristic monitoring data of the wind turbine blades according to each characteristic monitoring scheme, and input the data into a preset state assessment model of the corresponding characteristic monitoring scheme to obtain a real-time state assessment value; A comprehensive status evaluation value is generated based on the real-time status evaluation values ​​of all feature monitoring schemes, and whether to generate an early warning instruction is determined based on the comprehensive status evaluation value and the status evaluation change characteristics.

2. The early warning method for monitoring the status of wind turbine blades according to claim 1, characterized in that: Filter out historical similar monitoring logs based on basic information and environmental information, including: The basic information of the wind turbine blades includes structural information, life information and operation and maintenance information; The environmental information of the wind turbine blades includes weather condition information, wind speed information, temperature information and terrain information; Obtain historical monitoring logs of the same type of wind turbine blades, extract historical basic information and historical environmental information in each historical monitoring log, and perform similarity analysis with the current basic information and environmental information to obtain a comprehensive similarity coefficient; Pre-set similarity coefficient threshold; The historical monitoring log whose comprehensive similarity coefficient is greater than the similarity coefficient threshold is set as the historical similarity monitoring log of the current fan blade; Obtain the historical status evaluation value of each historical similar monitoring log and the credibility of the corresponding historical status evaluation value, and set the monitoring plan corresponding to the historical similar monitoring log whose credibility is greater than a preset credibility threshold as the characteristic monitoring plan; The historical status evaluation values ​​and historical monitoring data in the historical similar monitoring logs of the characteristic monitoring scheme are analyzed, and the characteristic monitoring data corresponding to the characteristic monitoring scheme is determined according to the analysis results.

3. The early warning method for monitoring the status of wind turbine blades according to claim 2, characterized in that: Determine the characteristic monitoring data of the corresponding characteristic monitoring plan based on the analysis results, including: Establish a time reference line based on the historical monitoring duration of similar historical monitoring logs for each feature monitoring scheme, and set data collection nodes according to preset time intervals; According to the data collection node, the historical monitoring data and the historical status evaluation value in the corresponding historical similar monitoring log are obtained, and mapped to the corresponding time reference line to obtain the historical status evaluation value-monitoring data analysis diagram of several historical similar monitoring logs of the same feature monitoring scheme; Obtaining the curve change trend of the historical state evaluation value of the same historical state evaluation value-monitoring data analysis diagram, and calculating the similarity between the curve change trend of each historical monitoring data and the curve change trend of the historical state evaluation value; Filter out historical monitoring data with a similarity greater than a similarity threshold, and calculate a trend delay time between a curve change trend of the filtered historical monitoring data and a curve change trend of a historical state evaluation value; If the trend delay duration is less than the preset duration threshold, the corresponding historical monitoring data is set as the pending feature monitoring data of the corresponding historical similar monitoring log; Compare the pending feature monitoring data of several historically similar monitoring logs of the same feature monitoring scheme to obtain the repetition ratio of the same pending feature monitoring data in different historically similar monitoring logs; The undetermined characteristic monitoring data with a repetition ratio greater than a preset ratio threshold is set as the characteristic monitoring data of the corresponding characteristic monitoring scheme; The weight coefficient of the corresponding feature monitoring data is generated according to the similarity, trend delay time and repetition ratio of each feature monitoring data of the same feature monitoring scheme.

4. The early warning method for monitoring the status of wind turbine blades according to claim 3, characterized in that: The calculation formula of the weight coefficient of the characteristic monitoring data is: ； Among them, Q is the weight coefficient of the feature monitoring data, z1 is the weight coefficient of the similarity, n1 is the number of historical similar monitoring logs where the feature monitoring data with a similarity greater than the similarity threshold is located, and N is the total number of historical similar monitoring logs of the same feature monitoring scheme. is the similarity of the feature monitoring data in the i1th historical similar monitoring log, X0 is the similarity threshold, a1 is the first weight conversion coefficient, z2 is the weight coefficient of the trend delay time, n2 is the number of historical similar monitoring logs where the feature monitoring data with a trend delay time shorter than the preset time threshold is located, is the trend delay duration of the feature monitoring data in the i2th historical similar monitoring log, X0 is the preset duration threshold, a2 is the second weight conversion coefficient, z3 is the weight coefficient of the repetition ratio, b is the repetition ratio, B0 is the preset ratio threshold, and a3 is the third weight conversion coefficient.

5. The early warning method for monitoring the status of wind turbine blades according to claim 4, characterized in that: Generate a comprehensive status assessment value based on the real-time status assessment values ​​of all feature monitoring schemes, including: A neural network is trained based on the feature monitoring data of the same feature monitoring scheme, the weight coefficient of the corresponding feature monitoring data, and the historical state evaluation value to obtain a preset state evaluation model of the corresponding feature monitoring scheme; Based on each characteristic monitoring scheme, real-time characteristic monitoring data of the corresponding wind turbine blade is obtained, and the real-time characteristic monitoring data is input into a preset state evaluation model of the corresponding characteristic monitoring scheme to obtain a real-time state evaluation value; Calculate the credibility difference of each feature monitoring scheme whose credibility is greater than the credibility threshold; Generate a weight coefficient for each feature monitoring scheme according to the number of all feature monitoring schemes and the corresponding credibility difference; A comprehensive status evaluation value is generated based on the real-time status evaluation value of each feature monitoring scheme and the corresponding weight coefficient.

6. The early warning method for monitoring the status of wind turbine blades according to claim 5, characterized in that: Preset the preset monitoring time nodes of the current monitoring cycle, and construct a status evaluation value change curve according to the time sequence of the preset monitoring time nodes, the historical comprehensive status evaluation values ​​corresponding to the preset monitoring time nodes, and the comprehensive status evaluation values ​​of the current monitoring time nodes; The state assessment value change curve is extrapolated to obtain the predicted state assessment value change curve for the future period; Acquire predicted change characteristics of the predicted state evaluation value change curve, wherein the predicted change characteristics include a predicted change trend and a predicted change rate; Compare the predicted state evaluation value change curve with the comprehensive state evaluation value of the current monitoring time node to obtain a first comprehensive state evaluation value difference and a second comprehensive state evaluation value difference, and calculate the difference value between the first comprehensive state evaluation value difference and the second comprehensive state evaluation value difference; The predicted change trend, predicted change rate and difference value are set as state assessment features, and a compensation coefficient of the comprehensive state assessment value of the current monitoring time node is generated according to the state assessment features.

7. The early warning method for monitoring the status of wind turbine blades according to claim 6, characterized in that: Determine whether to generate an early warning instruction based on the comprehensive status assessment value and the status assessment change characteristics, including: Correcting the comprehensive status evaluation value according to the compensation coefficient to obtain a corrected comprehensive status evaluation value; Pre-set status assessment thresholds; If the corrected comprehensive status assessment value is greater than the status assessment threshold, no warning instruction is generated; If the corrected comprehensive status assessment value is less than the status assessment threshold, a warning instruction is generated.

8. The early warning method for monitoring the status of wind turbine blades according to claim 7, characterized in that: The warning instructions include: Compare the real-time feature monitoring data obtained by each feature monitoring scheme with the corresponding standard monitoring data to obtain the difference of the real-time feature monitoring data; Determine abnormal feature monitoring data and abnormal level in each feature monitoring scheme according to the real-time feature monitoring data difference; Compare the abnormal feature monitoring data with the fault monitoring data set of the historical fault type of the corresponding feature monitoring scheme to obtain the correlation degree between the current abnormal feature monitoring data and the fault monitoring data set of the historical fault type; The historical fault type with the greatest correlation degree is set as the predicted first fault type of the abnormal feature monitoring data of the corresponding feature monitoring scheme; Analyze multiple predicted first fault types according to the credibility of all characteristic monitoring schemes to determine the predicted second fault type; Screening out abnormal characteristic monitoring data of a characteristic monitoring scheme that predicts a first fault type consistent with a second fault type, and determining a plurality of predicted first abnormal regions of the fan blade according to a data source and a sensor type of the screened abnormal characteristic monitoring data; Analyze multiple predicted first abnormal areas according to the weight coefficients of the screened abnormal feature monitoring data and the abnormality levels to determine the predicted second abnormal area; According to the data change characteristics of the abnormal characteristic monitoring data in the current monitoring period, the predicted time interval in which the corresponding abnormal characteristic monitoring data will have obvious fault characteristics is predicted; Based on the preset operation and maintenance analysis model of the wind turbine blades, the predicted second fault type, the predicted second abnormal area, and the predicted time interval in which each abnormal characteristic monitoring data shows obvious fault characteristics are analyzed to obtain multiple preset operation and maintenance strategies; Based on the objective function of maximizing the service life of wind turbine blades and minimizing the operation and maintenance cost, multiple preset operation and maintenance strategies are analyzed to determine the optimal operation and maintenance strategy; An early warning instruction is generated according to the predicted second fault type, the predicted second abnormal area, and the preferred operation and maintenance strategy.

9. An early warning system for monitoring the status of wind turbine blades, characterized in that: include: An acquisition module is used to acquire basic information and environmental information of the wind turbine blades, screen out historical similar monitoring logs based on the basic information and environmental information, and determine several characteristic monitoring schemes and characteristic monitoring data corresponding to the characteristic monitoring schemes based on the historical similar monitoring logs; A state assessment module, used to obtain real-time characteristic monitoring data of the wind turbine blades according to each characteristic monitoring scheme, and input it into a preset state assessment model of the corresponding characteristic monitoring scheme to obtain a real-time state assessment value; The early warning module is used to generate a comprehensive status evaluation value based on the real-time status evaluation values ​​of all feature monitoring schemes, and determine whether to generate an early warning instruction based on the comprehensive status evaluation value and the status evaluation change characteristics.

Images