A Fault Diagnosis Method for Wind Turbine Generators

Abstract

The present application discloses a fault diagnosis method for a wind turbine generator set, comprising: determining characteristic operation data and multiple fault data intervals of different fault types of a component; determining associated data and historical impact characteristics of the fault data interval based on the influence relationship between the multiple fault data intervals of the characteristic operation data and the characteristic operation data of other components; if the real-time characteristic operation data is in the fault data interval, setting it as abnormal characteristic operation data and generating a predicted fault evaluation value; obtaining the actual impact characteristics of the preceding associated data and the predicted impact characteristics of the succeeding associated data of the fault data interval where the abnormal characteristic operation data is located, and generating a corrected fault evaluation value in combination with the predicted fault evaluation value; generating a comprehensive fault evaluation value of the corresponding component according to all the corrected fault evaluation values, and generating a fault diagnosis result based on the comprehensive fault evaluation value, thereby improving the fault diagnosis accuracy of the wind turbine generator set, formulating a reasonable operation and maintenance strategy, and increasing the service life of the components.

Description

Technical Field

[0001] The present application relates to the technical field of wind turbines, and in particular to a fault diagnosis method for wind turbines. Background Art

[0002] A wind turbine is a system that converts the kinetic energy of wind into electrical energy. A wind turbine consists of blades, hubs, generators, reinforcements and other components. Currently, the number of installed wind turbines is increasing year by year, and they are often located in harsh environments such as deserts and mountainous areas. Therefore, the fault diagnosis of wind turbines has attracted widespread attention.

[0003] In the prior art, due to the large number of components in wind turbines and the complex working environment, traditional monitoring methods are used to determine whether a component has failed. Traditional monitoring methods cannot guarantee the accuracy of fault diagnosis, nor can they predict the impact of each component failure on other components, thereby reducing the fault diagnosis accuracy, operation and maintenance efficiency, and component service life of wind turbines. Summary of the invention

[0004] To solve the above technical problems, the present application provides a fault diagnosis method for a wind turbine generator set, by determining the characteristic operating data of each component and the corresponding multiple fault data intervals, and determining the associated data and the associated characteristics of the associated data according to the influence relationship of each fault data interval on the characteristic operating data of other components. If the real-time characteristic operating data is abnormal characteristic operating data, a predicted fault evaluation value and the actual influence characteristics and fault influence characteristics of the associated data are obtained, and a corrected fault evaluation value is generated. A fault diagnosis result is generated according to all corrected fault evaluation values ​​of each component of the wind turbine generator set, thereby improving the fault diagnosis accuracy of the wind turbine generator set, formulating a reasonable operation and maintenance strategy, and increasing the service life of the components.

[0005] In some embodiments of the present application, a fault diagnosis method for a wind turbine is provided, comprising:

[0006] Obtaining a historical fault log of each component in the wind turbine generator set, and determining characteristic operation data of different fault types of the corresponding component and multiple fault data intervals of the characteristic operation data according to the historical fault log;

[0007] Determine the associated data of each fault data interval of the characteristic operation data and the historical impact characteristics of the associated data based on the influence relationship between the multiple fault data intervals of the characteristic operation data and the characteristic operation data of other components, wherein the associated data includes the previous associated data and the subsequent associated data;

[0008] Obtaining the real-time characteristic operation data of each component, determining whether the real-time characteristic operation data is in the fault data interval, and if so, setting the corresponding real-time characteristic operation data as abnormal characteristic operation data, and generating a predicted fault evaluation value corresponding to the abnormal characteristic operation data;

[0009] Obtaining actual impact characteristics of preceding associated data and predicted impact characteristics of subsequent associated data in the fault data interval where the abnormal characteristic operation data is located, and generating a corrected fault evaluation value according to the actual impact characteristics, the predicted impact characteristics, and the predicted fault evaluation value;

[0010] A comprehensive fault evaluation value of a corresponding component is generated according to all corrected fault evaluation values ​​of the same component, and a fault diagnosis result is generated based on the comprehensive fault evaluation values ​​of all components.

[0011] In some embodiments of the present application, determining characteristic operation data of a corresponding component according to a historical fault log and generating an operation evaluation value of the characteristic operation data includes:

[0012] Obtain the historical fault log of each component in the wind turbine generator set, and determine the fault type corresponding to each historical fault log;

[0013] Preset several fault evaluation indicators for each component's fault type;

[0014] Perform correlation analysis on the historical operation data in the historical fault log of each component and multiple fault evaluation indicators corresponding to the fault type in the historical fault log to obtain the correlation between the historical operation data and each fault evaluation indicator;

[0015] The historical operation data with a correlation greater than a preset correlation threshold is set as the marked operation data of the fault type corresponding to the fault evaluation index;

[0016] Compare the marked operation data in all historical fault logs of the same fault type of each component to obtain the marked ratio of the same marked operation data, and set the marked operation data with a marked ratio greater than a preset marked ratio threshold in the same fault type as the characteristic operation data of the corresponding fault type;

[0017] Construct a characteristic operation data sequence W of the corresponding component according to the characteristic operation data of all fault types of the same component;

[0018]

[0019] Among them, x1′ (1) is the first characteristic operation data of the first fault type, n1 is the total number of characteristic operation data of the first fault type, m is the total number of fault types of the current component, xm′ (1)is the first characteristic operation data of the mth fault type, nm is the total number of characteristic operation data of the mth fault type;

[0020] Obtain multiple data intervals of characteristic operation data of each fault type in the corresponding historical fault log, and compare the multiple data intervals with preset standard data intervals of the characteristic operation data in the fault evaluation index of the corresponding fault type to obtain multiple data interval difference amounts;

[0021] The data intervals whose data interval difference is greater than a preset difference threshold are set as fault data intervals of the characteristic operation data, and each fault data interval is configured with a corresponding preset fault coefficient.

[0022] In some embodiments of the present application, determining the associated data of each fault data interval of the characteristic operation data and the influence characteristics of the associated data includes:

[0023] Set target characteristic operation data in sequence according to the characteristic operation data sequence W of each component;

[0024] Determine the historical time period in which each fault data interval of the target characteristic operation data is located, and construct a time reference line for each fault data interval of the target characteristic operation data according to each historical time period and the adjacent time periods of the historical time period, and set multiple data collection nodes at preset time intervals;

[0025] Acquire characteristic operation data of other components and current target characteristic operation data according to multiple data acquisition nodes of each time reference line, and map them to the time reference line of the corresponding fault data interval to obtain the characteristic operation data influence relationship diagram of the corresponding fault data interval;

[0026] According to the fault start node of the corresponding fault data interval in the characteristic operation data impact relationship diagram, the characteristic operation data impact relationship diagram is divided into a first impact relationship diagram and a second impact relationship diagram;

[0027] Analyze the influence relationship between the target characteristic operation data and the characteristic operation data of other components in the corresponding fault data interval according to the first influence relationship diagram of each time reference line, and determine the previous correlation data of the target characteristic operation data corresponding to the fault data interval and the historical influence characteristics of the previous correlation data according to the analysis result;

[0028] According to the second influence relationship diagram of each time reference line, the influence relationship between the target characteristic operation data and the characteristic operation data of other components in the corresponding fault data interval is analyzed, and the post-correlated data of the target characteristic operation data corresponding to the fault data interval and the historical influence characteristics of the post-correlated data are determined according to the analysis results.

[0029] In some embodiments of the present application, determining the associated data of each fault data interval of the characteristic operation data and the historical impact characteristics of the associated data further includes:

[0030] Obtain the degree of dependence between the current target characteristic operation data and the characteristic operation data of other components at each data collection node in the first influence relationship diagram, and filter out the characteristic operation data of other components whose degree of dependence is greater than a preset degree of dependence;

[0031] Obtain the fluctuation degree of the characteristic operation data of other selected components at the preset data collection node of the first influence relationship diagram; if the fluctuation degree is greater than the preset fluctuation degree threshold, set the corresponding characteristic operation data of other components as the previous correlation data of the fault data interval corresponding to the current target characteristic operation data, and set the dependence degree and the fluctuation degree as the historical influence characteristics of the corresponding previous correlation data;

[0032] Obtaining the influence degree of the current target characteristic operation data at each data acquisition node in the second influence relationship diagram on the characteristic operation data of other components, screening out the characteristic operation data of other components whose influence degree is greater than the preset influence degree, and setting them as the post-correlated data of the fault data interval corresponding to the current target characteristic operation data;

[0033] The change rate, change value, and change trend of the subsequent associated data in the second impact relationship diagram are obtained, and the impact degree and the change rate, change value, and change trend are set as historical impact features of the corresponding subsequent associated data.

[0034] In some embodiments of the present application, if yes, setting the corresponding real-time characteristic operation data as abnormal characteristic operation data, and generating a predicted fault evaluation value corresponding to the abnormal characteristic operation data, includes:

[0035] Acquire the real-time characteristic operation data of each component, determine whether the real-time characteristic operation data is in the fault data interval, and if so, set the current real-time characteristic operation data as the abnormal characteristic operation data of the corresponding component and determine the fault type of the abnormal characteristic operation data, and obtain the preset fault coefficient of the fault data interval where the abnormal characteristic operation data is located;

[0036] Generate a first predicted fault sub-evaluation value of the abnormal characteristic operation data according to a preset fault coefficient of the fault data interval in which the abnormal characteristic operation data is located;

[0037] Convert the historical fault levels and historical operation and maintenance costs in the historical fault logs of the fault data interval where the abnormal characteristic operation data is located into a first value and a second value of the same dimension, respectively, and generate a second predicted fault sub-evaluation value corresponding to the abnormal characteristic operation data according to the first value and the second value;

[0038] Obtaining a predicted fault evaluation value of the abnormal characteristic operation data according to the first predicted fault sub-evaluation value and the second predicted fault sub-evaluation value of the abnormal characteristic operation data;

[0039] P = p1 × a1 + p2 × a2;

[0040] Among them, P is the predicted fault evaluation value, p1 is the first predicted fault sub-evaluation value, a1 is the weight coefficient of the first predicted fault sub-evaluation value, p2 is the second predicted fault sub-evaluation value, and a2 is the weight coefficient of the second predicted fault sub-evaluation value.

[0041] In some embodiments of the present application, obtaining actual impact features of preceding correlation data and predicted impact features of subsequent correlation data of the fault data interval in which the abnormal characteristic operation data is located includes:

[0042] Acquire the previous correlation data of the fault data interval where the abnormal characteristic operation data is located, and obtain the predicted fault start node of the abnormal characteristic operation data according to the fault data interval where the abnormal characteristic operation data is located;

[0043] According to the preset data collection node adjacent to the node where the fault starts to be predicted, the actual impact feature of each previous associated data of the abnormal characteristic operation data is obtained, and the actual fluctuation degree in each actual impact feature is subtracted from the fluctuation degree in the historical impact feature of the corresponding previous associated data to obtain multiple actual fluctuation degree difference values;

[0044] Generate a comprehensive actual fluctuation degree difference value of all previous associated data of the fault data interval where the abnormal characteristic operation data is located according to multiple actual fluctuation degree differences, and set the credibility of the fault data interval where the abnormal characteristic operation data is located according to the comprehensive actual fluctuation degree difference value;

[0045] Acquire post-correlation data of the fault data interval where the abnormal characteristic operation data is located, set the historical impact characteristics of the post-correlation data as the predicted impact characteristics of the post-correlation data of the future period, and generate the predicted data interval of the post-correlation data of the future period according to the predicted impact characteristics of the post-correlation data of the future period and the real-time post-correlation data;

[0046] Compare the predicted data interval of each post-correlation data in the future time period with the preset standard data interval of the corresponding post-correlation data to obtain the predicted failure trend and predicted failure coefficient of the predicted data interval of each post-correlation data in the future time period;

[0047] Generate a predicted fault impact coefficient of the corresponding post-correlated data according to the predicted fault trend and predicted fault coefficient of each post-correlated data in the predicted data interval of the future period;

[0048] A predicted comprehensive fault influence coefficient is generated according to the predicted fault influence coefficients of all post-correlated data in the fault data interval where the abnormal characteristic operation data is located, and a compensation coefficient of the predicted fault evaluation value of the abnormal characteristic operation data is set based on the predicted comprehensive fault influence coefficient.

[0049] In some embodiments of the present application, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set according to the comprehensive actual fluctuation degree difference, including:

[0050] Presetting a first preset comprehensive actual fluctuation degree difference interval, a second preset comprehensive actual fluctuation degree difference interval, a third preset comprehensive actual fluctuation degree difference interval, and a fourth preset comprehensive actual fluctuation degree difference interval;

[0051] When the comprehensive actual fluctuation degree difference is within the first preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval where the abnormal characteristic operation data is located is set to a fourth preset credibility;

[0052] When the comprehensive actual fluctuation degree difference is in the second preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set to a third preset credibility;

[0053] When the comprehensive actual fluctuation degree difference is in the third preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set to the second preset credibility;

[0054] When the comprehensive actual fluctuation degree difference is in the fourth preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set to the first preset credibility.

[0055] In some embodiments of the present application, generating a modified fault evaluation value according to the actual impact feature, the predicted impact feature, and the predicted fault evaluation value includes:

[0056] The credibility of the fault data interval in which the abnormal characteristic operation data is located is compared with the preset credibility threshold. If the credibility is greater than the preset credibility threshold, the predicted fault evaluation value is corrected according to the compensation coefficient to obtain a corrected fault evaluation value;

[0057] The calculation formula of the modified fault evaluation value is:

[0058] B = P × c1 × c2;

[0059] Among them, B is the corrected fault evaluation value, c1 is the compensation coefficient, and c2 is the preset constant corresponding to the fault type of the current abnormal characteristic operation data;

[0060] If the credibility is less than the preset credibility threshold, re-determine whether the current abnormal characteristic operation data is in the fault data interval;

[0061] If so, re-determine the fault data interval where the abnormal characteristic operation data is located and recalculate the predicted fault evaluation value, set the credibility according to the actual impact characteristics of the previous associated data in the fault data interval where the abnormal characteristic operation data is located, until the credibility is greater than the preset credibility threshold, set the compensation coefficient according to the predicted impact characteristics of the subsequent associated data in the fault data interval where the abnormal characteristic operation data is located, and correct the recalculated predicted fault evaluation value to obtain a corrected fault evaluation value;

[0062] If not, set the current abnormal characteristic operation data to normal characteristic operation data.

[0063] In some embodiments of the present application, a fault diagnosis result is generated based on the comprehensive fault evaluation values ​​of all components, including:

[0064] Generate a fault diagnosis result according to the comprehensive fault evaluation value of each component, wherein the fault diagnosis result includes the fault level, fault type and operation and maintenance strategy of the corresponding component;

[0065] All components of the wind turbine are sorted according to the comprehensive fault evaluation value of each component and the weight coefficient of the corresponding component, the component operation and maintenance sequence of the wind turbine is set according to the sorting result, and the operation and maintenance instructions of the wind turbine are generated based on the component operation and maintenance sequence and the fault diagnosis result.

[0066] Compared with the prior art, the fault diagnosis method for a wind turbine generator system in the embodiment of the present application has the following beneficial effects:

[0067] By determining the characteristic operation data of each component and the corresponding multiple fault data intervals, and according to the influence relationship of each fault data interval on the characteristic operation data of other components, the associated data and the associated characteristics of the associated data are determined. If the real-time characteristic operation data is abnormal characteristic operation data, the predicted fault evaluation value and the actual influence characteristics and fault influence characteristics of the associated data are obtained, and a corrected fault evaluation value is generated. The fault diagnosis result is generated according to all the corrected fault evaluation values ​​of each component of the wind turbine, which improves the fault diagnosis accuracy of the wind turbine, formulates a reasonable operation and maintenance strategy, and increases the service life of the components. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a flow chart of a fault diagnosis method for a wind turbine generator system in a preferred embodiment of the present application. DETAILED DESCRIPTION

[0069] The specific implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present application but are not intended to limit the scope of the present application.

[0070] In the description of the present application, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply 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 understood as a limitation on the present application.

[0071] The terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0072] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0073] like Figure 1 As shown, a fault diagnosis method for a wind turbine generator system according to a preferred embodiment of the present application includes:

[0074] Step S101: obtaining a historical fault log of each component in the wind turbine generator set, and determining characteristic operation data of different fault types of the corresponding component and a plurality of fault data intervals of the characteristic operation data according to the historical fault log;

[0075] Step S102: determining the associated data of each fault data interval of the characteristic operation data and the historical impact characteristics of the associated data based on the influence relationship between the multiple fault data intervals of the characteristic operation data and the characteristic operation data of other components, wherein the associated data includes the previous associated data and the subsequent associated data;

[0076] Step S103: acquiring the real-time characteristic operation data of each component, determining whether the real-time characteristic operation data is in the fault data interval, and if so, setting the corresponding real-time characteristic operation data as abnormal characteristic operation data, and generating a predicted fault evaluation value corresponding to the abnormal characteristic operation data;

[0077] Step S104: obtaining actual impact features of the preceding associated data and predicted impact features of the following associated data of the fault data interval where the abnormal characteristic operation data is located, and generating a modified fault evaluation value according to the actual impact features, the predicted impact features and the predicted fault evaluation value;

[0078] Step S105: Generate a comprehensive fault evaluation value of the corresponding component according to all corrected fault evaluation values ​​of the same component, and generate a fault diagnosis result based on the comprehensive fault evaluation values ​​of all components.

[0079] In some embodiments of the present application, determining characteristic operation data of a corresponding component according to a historical fault log and generating an operation evaluation value of the characteristic operation data includes:

[0080] Obtain the historical fault log of each component in the wind turbine generator set, and determine the fault type corresponding to each historical fault log;

[0081] Preset several fault evaluation indicators for each component's fault type;

[0082] Perform correlation analysis on the historical operation data in the historical fault log of each component and multiple fault evaluation indicators corresponding to the fault type in the historical fault log to obtain the correlation between the historical operation data and each fault evaluation indicator;

[0083] The historical operation data with a correlation greater than a preset correlation threshold is set as the marked operation data of the fault type corresponding to the fault evaluation index;

[0084] Compare the marked operation data in all historical fault logs of the same fault type of each component to obtain the marked ratio of the same marked operation data, and set the marked operation data with a marked ratio greater than a preset marked ratio threshold in the same fault type as the characteristic operation data of the corresponding fault type;

[0085] Construct a characteristic operation data sequence W of the corresponding component according to the characteristic operation data of all fault types of the same component;

[0086]

[0087] Among them, x1′ (1) is the first characteristic operation data of the first fault type, n1 is the total number of characteristic operation data of the first fault type, m is the total number of fault types of the current component, xm′ (1)is the first characteristic operation data of the mth fault type, nm is the total number of characteristic operation data of the mth fault type;

[0088] Obtain multiple data intervals of characteristic operation data of each fault type in the corresponding historical fault log, and compare the multiple data intervals with preset standard data intervals of the characteristic operation data in the fault evaluation index of the corresponding fault type to obtain multiple data interval difference amounts;

[0089] The data intervals whose data interval difference is greater than a preset difference threshold are set as fault data intervals of the characteristic operation data, and each fault data interval is configured with a corresponding preset fault coefficient.

[0090] In this embodiment, marking operation data refers to marking historical operation data for each fault type whose correlation degree of multiple fault evaluation indicators is greater than a preset correlation degree threshold value differently, and the marking ratio = the number of marked operation data in all historical fault logs of the same fault type / the number of all historical fault logs of the same fault type, and the preset marking ratio threshold is 2 / 3.

[0091] In this embodiment, the arrangement order of the fault types in the characteristic operation data sequence W is set according to the fault frequency and fault level of the fault type. When the fault frequency is higher and the fault level is larger, the arrangement order of the fault type is higher. The arrangement order of the characteristic operation data in each fault type is set according to the correlation with the fault evaluation index of the corresponding fault type. If the correlation is larger, the arrangement order of the characteristic operation data is higher.

[0092] In this embodiment, the preset fault coefficient is set based on the difference between the data intervals and the preset difference threshold. When the difference is larger, the preset fault coefficient is larger, and when the difference is smaller, the preset fault coefficient is smaller, and the value range of the preset fault coefficient is (0, 1).

[0093] In some embodiments of the present application, determining the associated data of each fault data interval of the characteristic operation data and the influence characteristics of the associated data includes:

[0094] Set target characteristic operation data in sequence according to the characteristic operation data sequence W of each component;

[0095] Determine the historical time period in which each fault data interval of the target characteristic operation data is located, and construct a time reference line for each fault data interval of the target characteristic operation data according to each historical time period and the adjacent time periods of the historical time period, and set multiple data collection nodes at preset time intervals;

[0096] Acquire characteristic operation data of other components and current target characteristic operation data according to multiple data acquisition nodes of each time reference line, and map them to the time reference line of the corresponding fault data interval to obtain the characteristic operation data influence relationship diagram of the corresponding fault data interval;

[0097] According to the fault start node of the corresponding fault data interval in the characteristic operation data impact relationship diagram, the characteristic operation data impact relationship diagram is divided into a first impact relationship diagram and a second impact relationship diagram;

[0098] Analyze the influence relationship between the target characteristic operation data and the characteristic operation data of other components in the corresponding fault data interval according to the first influence relationship diagram of each time reference line, and determine the previous correlation data of the target characteristic operation data corresponding to the fault data interval and the historical influence characteristics of the previous correlation data according to the analysis result;

[0099] According to the second influence relationship diagram of each time reference line, the influence relationship between the target characteristic operation data and the characteristic operation data of other components in the corresponding fault data interval is analyzed, and the post-correlated data of the target characteristic operation data corresponding to the fault data interval and the historical influence characteristics of the post-correlated data are determined according to the analysis results.

[0100] In this embodiment, the target characteristic operation data are selected in sequence to obtain the pre-correlation data, post-correlation data and historical impact characteristics of the pre-correlation data and post-correlation data of the characteristic operation data of each component, thereby laying a foundation for the subsequent generation of corrected fault evaluation values, improving the fault diagnosis accuracy of each component, and thus ensuring the fault diagnosis accuracy of the wind turbine.

[0101] In some embodiments of the present application, determining the associated data of each fault data interval of the characteristic operation data and the historical impact characteristics of the associated data further includes:

[0102] Obtain the degree of dependence between the current target characteristic operation data and the characteristic operation data of other components at each data collection node in the first influence relationship diagram, and filter out the characteristic operation data of other components whose degree of dependence is greater than a preset degree of dependence;

[0103] Obtain the fluctuation degree of the characteristic operation data of other selected components at the preset data collection node of the first influence relationship diagram; if the fluctuation degree is greater than the preset fluctuation degree threshold, set the corresponding characteristic operation data of other components as the previous correlation data of the fault data interval corresponding to the current target characteristic operation data, and set the dependence degree and the fluctuation degree as the historical influence characteristics of the corresponding previous correlation data;

[0104] Obtaining the influence degree of the current target characteristic operation data at each data acquisition node in the second influence relationship diagram on the characteristic operation data of other components, screening out the characteristic operation data of other components whose influence degree is greater than the preset influence degree, and setting them as the post-correlated data of the fault data interval corresponding to the current target characteristic operation data;

[0105] The change rate, change value, and change trend of the subsequent associated data in the second impact relationship diagram are obtained, and the impact degree and the change rate, change value, and change trend are set as historical impact features of the corresponding subsequent associated data.

[0106] In this embodiment, the degree of dependence refers to the degree to which the target characteristic operation data changes with the change of the characteristic operation data of other components, and the degree of influence refers to the degree to which the characteristic operation data of other components changes due to the change of the target characteristic operation data.

[0107] In this embodiment, the preset data collection nodes of the first impact relationship diagram refer to several data collection nodes adjacent to the fault start node.

[0108] In this embodiment, by judging whether the characteristic operation data of other screened components show a large degree of fluctuation in several data collection nodes adjacent to the fault start node, the previous associated data in the characteristic operation data of other components that affects the target characteristic operation data to generate the corresponding fault data interval is determined, thereby laying a foundation for subsequently improving the fault diagnosis accuracy of the characteristic operation data.

[0109] In this embodiment, by determining the historical impact characteristics of the subsequent associated data of the fault data interval corresponding to the target characteristic operation data, a foundation is laid for predicting the characteristic operation data of other components after a subsequent characteristic operation data fails, thereby improving the prediction accuracy, and formulating operation and maintenance plans in a timely manner, thereby improving the fault diagnosis accuracy of the wind turbine and the service life of the components, and reducing the operation and maintenance costs.

[0110] In some embodiments of the present application, if yes, setting the corresponding real-time characteristic operation data as abnormal characteristic operation data, and generating a predicted fault evaluation value corresponding to the abnormal characteristic operation data, includes:

[0111] Acquire the real-time characteristic operation data of each component, determine whether the real-time characteristic operation data is in the fault data interval, and if so, set the current real-time characteristic operation data as the abnormal characteristic operation data of the corresponding component and determine the fault type of the abnormal characteristic operation data, and obtain the preset fault coefficient of the fault data interval where the abnormal characteristic operation data is located;

[0112] Generate a first predicted fault sub-evaluation value of the abnormal characteristic operation data according to a preset fault coefficient of the fault data interval in which the abnormal characteristic operation data is located;

[0113] Convert the historical fault levels and historical operation and maintenance costs in the historical fault logs of the fault data interval where the abnormal characteristic operation data is located into a first value and a second value of the same dimension, respectively, and generate a second predicted fault sub-evaluation value corresponding to the abnormal characteristic operation data according to the first value and the second value;

[0114] Obtaining a predicted fault evaluation value of the abnormal characteristic operation data according to the first predicted fault sub-evaluation value and the second predicted fault sub-evaluation value of the abnormal characteristic operation data;

[0115] P = p1 × a1 + p2 × a2;

[0116] Among them, P is the predicted fault evaluation value, p1 is the first predicted fault sub-evaluation value, a1 is the weight coefficient of the first predicted fault sub-evaluation value, p2 is the second predicted fault sub-evaluation value, and a2 is the weight coefficient of the second predicted fault sub-evaluation value.

[0117] In this embodiment, the historical fault level and the historical operation and maintenance cost are recorded in advance in each historical fault log. When the historical fault level of the fault data interval corresponding to the characteristic operation data in each historical fault log is higher and the historical operation cost is higher, the larger the first value and the second value are, the larger the corresponding second predicted fault sub-evaluation value is, and the larger the preset fault coefficient is, the larger the corresponding first predicted fault sub-evaluation value is.

[0118] In some embodiments of the present application, obtaining actual impact features of preceding correlation data and predicted impact features of subsequent correlation data of the fault data interval in which the abnormal characteristic operation data is located includes:

[0119] Acquire the previous correlation data of the fault data interval where the abnormal characteristic operation data is located, and obtain the predicted fault start node of the abnormal characteristic operation data according to the fault data interval where the abnormal characteristic operation data is located;

[0120] According to the preset data collection node adjacent to the node where the fault starts to be predicted, the actual impact feature of each previous associated data of the abnormal characteristic operation data is obtained, and the actual fluctuation degree in each actual impact feature is subtracted from the fluctuation degree in the historical impact feature of the corresponding previous associated data to obtain multiple actual fluctuation degree difference values;

[0121] Generate a comprehensive actual fluctuation degree difference value of all previous associated data of the fault data interval where the abnormal characteristic operation data is located according to multiple actual fluctuation degree differences, and set the credibility of the fault data interval where the abnormal characteristic operation data is located according to the comprehensive actual fluctuation degree difference value;

[0122] Acquire post-correlation data of the fault data interval where the abnormal characteristic operation data is located, set the historical impact characteristics of the post-correlation data as the predicted impact characteristics of the post-correlation data of the future period, and generate the predicted data interval of the post-correlation data of the future period according to the predicted impact characteristics of the post-correlation data of the future period and the real-time post-correlation data;

[0123] Compare the predicted data interval of each post-correlation data in the future time period with the preset standard data interval of the corresponding post-correlation data to obtain the predicted failure trend and predicted failure coefficient of the predicted data interval of each post-correlation data in the future time period;

[0124] Generate a predicted fault impact coefficient of the corresponding post-correlated data according to the predicted fault trend and predicted fault coefficient of each post-correlated data in the predicted data interval of the future period;

[0125] A predicted comprehensive fault influence coefficient is generated according to the predicted fault influence coefficients of all post-correlated data in the fault data interval where the abnormal characteristic operation data is located, and a compensation coefficient of the predicted fault evaluation value of the abnormal characteristic operation data is set based on the predicted comprehensive fault influence coefficient.

[0126] In this embodiment, the actual fluctuation degree difference = the actual fluctuation degree in the actual impact feature - the fluctuation degree in the historical impact feature. The predicted fault trend includes no fault, slow fault, and fast fault trends. The predicted fault coefficient refers to the fault degree of the predicted data interval of each post-correlation data in the future time period. When the data difference between the predicted data interval and the preset standard data interval is larger and the rate of change of the data deviation is faster, the predicted fault coefficient is larger and the predicted fault trend is worse, that is, the corresponding predicted fault impact coefficient of the post-correlation data is larger. When the predicted fault impact coefficient is larger, the set compensation coefficient is larger, and vice versa. The value range of the compensation coefficient is (1, 1.5).

[0127] In this embodiment, the predicted comprehensive fault influence coefficient is obtained by summing the predicted fault influence coefficients of all post-correlated data and the weight coefficients of the corresponding post-correlated data in the corresponding components. The predicted comprehensive fault influence coefficient can be used to determine the degree of fault influence of the real-time characteristic operation data on the characteristic operation data of other components, thereby determining the compensation coefficient to correct the predicted fault evaluation value, thereby improving the accuracy of fault diagnosis of the characteristic operation data of each component, improving the fault diagnosis accuracy of the wind turbine set and the service life of the components, and reducing the operation and maintenance costs.

[0128] In some embodiments of the present application, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set according to the comprehensive actual fluctuation degree difference, including:

[0129] Presetting a first preset comprehensive actual fluctuation degree difference interval, a second preset comprehensive actual fluctuation degree difference interval, a third preset comprehensive actual fluctuation degree difference interval, and a fourth preset comprehensive actual fluctuation degree difference interval;

[0130] When the comprehensive actual fluctuation degree difference is within the first preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval where the abnormal characteristic operation data is located is set to a fourth preset credibility;

[0131] When the comprehensive actual fluctuation degree difference is in the second preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set to a third preset credibility;

[0132] When the comprehensive actual fluctuation degree difference is in the third preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set to the second preset credibility;

[0133] When the comprehensive actual fluctuation degree difference is in the fourth preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set to the first preset credibility.

[0134] In this embodiment, the first preset comprehensive actual fluctuation degree difference interval < the second preset comprehensive actual fluctuation degree difference interval < the third preset comprehensive actual fluctuation degree difference interval < the fourth preset comprehensive actual fluctuation degree difference interval, the first preset credibility < the second preset credibility < the third preset credibility < the fourth preset credibility, when the preset comprehensive actual fluctuation degree difference in which the comprehensive actual fluctuation degree difference is located is smaller, it means that the difference between the historical impact characteristics and the actual impact characteristics of the previous correlation data is smaller, and the accuracy of the fault data interval in which the abnormal feature operation data is located is greater, that is, the corresponding credibility is greater.

[0135] In this embodiment, the accuracy of the fault data interval of the real-time characteristic operation data is determined based on the previous correlation data, thereby improving the fault diagnosis accuracy of the characteristic operation data, laying the foundation for the subsequent determination of the corrected fault evaluation value and the formulation of operation and maintenance strategies, improving the fault diagnosis accuracy of the wind turbine and the service life of the components, and reducing the operation and maintenance costs.

[0136] In some embodiments of the present application, generating a modified fault evaluation value according to the actual impact feature, the predicted impact feature, and the predicted fault evaluation value includes:

[0137] The credibility of the fault data interval in which the abnormal characteristic operation data is located is compared with the preset credibility threshold. If the credibility is greater than the preset credibility threshold, the predicted fault evaluation value is corrected according to the compensation coefficient to obtain a corrected fault evaluation value;

[0138] The calculation formula of the modified fault evaluation value is:

[0139] B = P × c1 × c2;

[0140] Among them, B is the corrected fault evaluation value, c1 is the compensation coefficient, and c2 is the preset constant corresponding to the fault type of the current abnormal characteristic operation data;

[0141] If the credibility is less than the preset credibility threshold, re-determine whether the current abnormal characteristic operation data is in the fault data interval;

[0142] If so, re-determine the fault data interval where the abnormal characteristic operation data is located and recalculate the predicted fault evaluation value, set the credibility according to the actual impact characteristics of the previous associated data in the fault data interval where the abnormal characteristic operation data is located, until the credibility is greater than the preset credibility threshold, set the compensation coefficient according to the predicted impact characteristics of the subsequent associated data in the fault data interval where the abnormal characteristic operation data is located, and correct the recalculated predicted fault evaluation value to obtain a corrected fault evaluation value;

[0143] If not, set the current abnormal characteristic operation data to normal characteristic operation data.

[0144] In some embodiments of the present application, a fault diagnosis result is generated based on the comprehensive fault evaluation values ​​of all components, including:

[0145] Generate a fault diagnosis result according to the comprehensive fault evaluation value of each component, wherein the fault diagnosis result includes the fault level, fault type and operation and maintenance strategy of the corresponding component;

[0146] All components of the wind turbine are sorted according to the comprehensive fault evaluation value of each component and the weight coefficient of the corresponding component, the component operation and maintenance sequence of the wind turbine is set according to the sorting result, and the operation and maintenance instructions of the wind turbine are generated based on the component operation and maintenance sequence and the fault diagnosis result.

[0147] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present application. These improvements and substitutions should also be regarded as the scope of protection of the present application.

Claims

1. A fault diagnosis method for a wind turbine generator system, characterized in that: include: Obtaining a historical fault log of each component in the wind turbine generator set, and determining characteristic operation data of different fault types of the corresponding component and multiple fault data intervals of the characteristic operation data according to the historical fault log; Determine the associated data of each fault data interval of the characteristic operation data and the historical impact characteristics of the associated data based on the influence relationship between the multiple fault data intervals of the characteristic operation data and the characteristic operation data of other components, wherein the associated data includes the previous associated data and the subsequent associated data; Obtaining the real-time characteristic operation data of each component, determining whether the real-time characteristic operation data is in the fault data interval, and if so, setting the corresponding real-time characteristic operation data as abnormal characteristic operation data, and generating a predicted fault evaluation value corresponding to the abnormal characteristic operation data; Obtaining actual impact characteristics of preceding associated data and predicted impact characteristics of subsequent associated data in the fault data interval where the abnormal characteristic operation data is located, and generating a corrected fault evaluation value according to the actual impact characteristics, the predicted impact characteristics, and the predicted fault evaluation value; A comprehensive fault evaluation value of a corresponding component is generated according to all corrected fault evaluation values ​​of the same component, and a fault diagnosis result is generated based on the comprehensive fault evaluation values ​​of all components.

2. The fault diagnosis method for a wind turbine generator set according to claim 1, characterized in that: Determine the characteristic operation data of the corresponding component according to the historical fault log and generate the operation evaluation value of the characteristic operation data, including: Obtain the historical fault log of each component in the wind turbine generator set, and determine the fault type corresponding to each historical fault log; Preset several fault evaluation indicators for each component's fault type; Perform correlation analysis on the historical operation data in the historical fault log of each component and multiple fault evaluation indicators corresponding to the fault type in the historical fault log to obtain the correlation between the historical operation data and each fault evaluation indicator; The historical operation data with a correlation greater than a preset correlation threshold is set as the marked operation data of the fault type corresponding to the fault evaluation index; Compare the marked operation data in all historical fault logs of the same fault type of each component to obtain the marked ratio of the same marked operation data, and set the marked operation data with a marked ratio greater than a preset marked ratio threshold in the same fault type as the characteristic operation data of the corresponding fault type; Construct a characteristic operation data sequence W of the corresponding component according to the characteristic operation data of all fault types of the same component; Among them, x1′ (1) is the first characteristic operation data of the first fault type, n1 is the total number of characteristic operation data of the first fault type, m is the total number of fault types of the current component, xm′ (1) is the first characteristic operation data of the mth fault type, nm is the total number of characteristic operation data of the mth fault type; Obtain multiple data intervals of characteristic operation data of each fault type in the corresponding historical fault log, and compare the multiple data intervals with preset standard data intervals of the characteristic operation data in the fault evaluation index of the corresponding fault type to obtain multiple data interval difference amounts; The data intervals whose data interval difference is greater than a preset difference threshold are set as fault data intervals of the characteristic operation data, and each fault data interval is configured with a corresponding preset fault coefficient.

3. The fault diagnosis method for a wind turbine generator set according to claim 2, characterized in that: Determine the associated data of each fault data interval of the characteristic operation data and the impact characteristics of the associated data, including: Set target characteristic operation data in sequence according to the characteristic operation data sequence W of each component; Determine the historical time period in which each fault data interval of the target characteristic operation data is located, and construct a time reference line for each fault data interval of the target characteristic operation data according to each historical time period and the adjacent time periods of the historical time period, and set multiple data collection nodes at preset time intervals; Acquire characteristic operation data of other components and current target characteristic operation data according to multiple data acquisition nodes of each time reference line, and map them to the time reference line of the corresponding fault data interval to obtain the characteristic operation data influence relationship diagram of the corresponding fault data interval; According to the fault start node of the corresponding fault data interval in the characteristic operation data impact relationship diagram, the characteristic operation data impact relationship diagram is divided into a first impact relationship diagram and a second impact relationship diagram; Analyze the influence relationship between the target characteristic operation data and the characteristic operation data of other components in the corresponding fault data interval according to the first influence relationship diagram of each time reference line, and determine the previous correlation data of the target characteristic operation data corresponding to the fault data interval and the historical influence characteristics of the previous correlation data according to the analysis result; According to the second influence relationship diagram of each time reference line, the influence relationship between the target characteristic operation data and the characteristic operation data of other components in the corresponding fault data interval is analyzed, and the post-correlated data of the target characteristic operation data corresponding to the fault data interval and the historical influence characteristics of the post-correlated data are determined according to the analysis results.

4. The fault diagnosis method for a wind turbine generator set according to claim 3, characterized in that: Determining the associated data of each fault data interval of the characteristic operation data and the historical impact characteristics of the associated data, further comprising: Obtain the degree of dependence between the current target characteristic operation data and the characteristic operation data of other components at each data collection node in the first influence relationship diagram, and filter out the characteristic operation data of other components whose degree of dependence is greater than a preset degree of dependence; Obtain the fluctuation degree of the characteristic operation data of other selected components at the preset data collection node of the first influence relationship diagram; if the fluctuation degree is greater than the preset fluctuation degree threshold, set the corresponding characteristic operation data of other components as the previous correlation data of the fault data interval corresponding to the current target characteristic operation data, and set the dependence degree and the fluctuation degree as the historical influence characteristics of the corresponding previous correlation data; Obtaining the influence degree of the current target characteristic operation data at each data acquisition node in the second influence relationship diagram on the characteristic operation data of other components, screening out the characteristic operation data of other components whose influence degree is greater than the preset influence degree, and setting them as the post-correlated data of the fault data interval corresponding to the current target characteristic operation data; The change rate, change value, and change trend of the subsequent associated data in the second impact relationship diagram are obtained, and the impact degree and the change rate, change value, and change trend are set as historical impact features of the corresponding subsequent associated data.

5. The fault diagnosis method for a wind turbine generator set according to claim 4, characterized in that: If yes, set the corresponding real-time characteristic operation data as abnormal characteristic operation data, and generate a predicted fault evaluation value corresponding to the abnormal characteristic operation data, including: Acquire the real-time characteristic operation data of each component, determine whether the real-time characteristic operation data is in the fault data interval, and if so, set the current real-time characteristic operation data as the abnormal characteristic operation data of the corresponding component and determine the fault type of the abnormal characteristic operation data, and obtain the preset fault coefficient of the fault data interval where the abnormal characteristic operation data is located; Generate a first predicted fault sub-evaluation value of the abnormal characteristic operation data according to a preset fault coefficient of the fault data interval in which the abnormal characteristic operation data is located; Convert the historical fault levels and historical operation and maintenance costs in the historical fault logs of the fault data interval where the abnormal characteristic operation data is located into a first value and a second value of the same dimension, respectively, and generate a second predicted fault sub-evaluation value corresponding to the abnormal characteristic operation data according to the first value and the second value; Obtaining a predicted fault evaluation value of the abnormal characteristic operation data according to the first predicted fault sub-evaluation value and the second predicted fault sub-evaluation value of the abnormal characteristic operation data; P = p1 × a1 + p2 × a2; Among them, P is the predicted fault evaluation value, p1 is the first predicted fault sub-evaluation value, a1 is the weight coefficient of the first predicted fault sub-evaluation value, p2 is the second predicted fault sub-evaluation value, and a2 is the weight coefficient of the second predicted fault sub-evaluation value.

6. The fault diagnosis method for a wind turbine generator set according to claim 5, characterized in that: The actual impact characteristics of the previous correlation data and the predicted impact characteristics of the subsequent correlation data in the fault data interval where the abnormal characteristic operation data is located are obtained, including: Acquire the previous correlation data of the fault data interval where the abnormal characteristic operation data is located, and obtain the predicted fault start node of the abnormal characteristic operation data according to the fault data interval where the abnormal characteristic operation data is located; According to the preset data collection node adjacent to the node where the fault starts to be predicted, the actual impact feature of each previous associated data of the abnormal characteristic operation data is obtained, and the actual fluctuation degree in each actual impact feature is subtracted from the fluctuation degree in the historical impact feature of the corresponding previous associated data to obtain multiple actual fluctuation degree difference values; Generate a comprehensive actual fluctuation degree difference value of all previous associated data of the fault data interval where the abnormal characteristic operation data is located according to multiple actual fluctuation degree differences, and set the credibility of the fault data interval where the abnormal characteristic operation data is located according to the comprehensive actual fluctuation degree difference value; Acquire post-correlation data of the fault data interval where the abnormal characteristic operation data is located, set the historical impact characteristics of the post-correlation data as the predicted impact characteristics of the post-correlation data of the future period, and generate the predicted data interval of the post-correlation data of the future period according to the predicted impact characteristics of the post-correlation data of the future period and the real-time post-correlation data; Compare the predicted data interval of each post-correlation data in the future time period with the preset standard data interval of the corresponding post-correlation data to obtain the predicted failure trend and predicted failure coefficient of the predicted data interval of each post-correlation data in the future time period; Generate a predicted fault impact coefficient of the corresponding post-correlated data according to the predicted fault trend and predicted fault coefficient of each post-correlated data in the predicted data interval of the future period; A predicted comprehensive fault influence coefficient is generated according to the predicted fault influence coefficients of all post-correlated data in the fault data interval where the abnormal characteristic operation data is located, and a compensation coefficient of the predicted fault evaluation value of the abnormal characteristic operation data is set based on the predicted comprehensive fault influence coefficient.

7. The fault diagnosis method for a wind turbine generator set according to claim 6, characterized in that: The credibility of the fault data interval where the abnormal characteristic operation data is located is set according to the difference in the comprehensive actual fluctuation degree, including: Presetting a first preset comprehensive actual fluctuation degree difference interval, a second preset comprehensive actual fluctuation degree difference interval, a third preset comprehensive actual fluctuation degree difference interval, and a fourth preset comprehensive actual fluctuation degree difference interval; When the comprehensive actual fluctuation degree difference is within the first preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval where the abnormal characteristic operation data is located is set to a fourth preset credibility; When the comprehensive actual fluctuation degree difference is in the second preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set to a third preset credibility; When the comprehensive actual fluctuation degree difference is in the third preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set to the second preset credibility; When the comprehensive actual fluctuation degree difference is in the fourth preset comprehensive actual fluctuation degree difference interval, the credibility of the fault data interval in which the abnormal characteristic operation data is located is set to the first preset credibility.

8. The fault diagnosis method for a wind turbine generator set according to claim 6, characterized in that: Generate a modified fault evaluation value based on the actual impact characteristics, the predicted impact characteristics, and the predicted fault evaluation value, including: The credibility of the fault data interval in which the abnormal characteristic operation data is located is compared with the preset credibility threshold. If the credibility is greater than the preset credibility threshold, the predicted fault evaluation value is corrected according to the compensation coefficient to obtain a corrected fault evaluation value; The calculation formula of the modified fault evaluation value is: B = P × c1 × c2; Among them, B is the corrected fault evaluation value, c1 is the compensation coefficient, and c2 is the preset constant corresponding to the fault type of the current abnormal characteristic operation data; If the credibility is less than the preset credibility threshold, re-determine whether the current abnormal characteristic operation data is in the fault data interval; If so, re-determine the fault data interval where the abnormal characteristic operation data is located and recalculate the predicted fault evaluation value, set the credibility according to the actual impact characteristics of the previous associated data in the fault data interval where the abnormal characteristic operation data is located, until the credibility is greater than the preset credibility threshold, set the compensation coefficient according to the predicted impact characteristics of the subsequent associated data in the fault data interval where the abnormal characteristic operation data is located, and correct the recalculated predicted fault evaluation value to obtain a corrected fault evaluation value; If not, set the current abnormal characteristic operation data to normal characteristic operation data.

9. The fault diagnosis method for a wind turbine generator set according to claim 8, characterized in that: Generate fault diagnosis results based on the comprehensive fault evaluation values ​​of all components, including: Generate a fault diagnosis result according to the comprehensive fault evaluation value of each component, wherein the fault diagnosis result includes the fault level, fault type and operation and maintenance strategy of the corresponding component; All components of the wind turbine are sorted according to the comprehensive fault evaluation value of each component and the weight coefficient of the corresponding component, the component operation and maintenance sequence of the wind turbine is set according to the sorting result, and the operation and maintenance instructions of the wind turbine are generated based on the component operation and maintenance sequence and the fault diagnosis result.

Images