On-line monitoring method and system for state of variable-pitch bearing

Abstract

The invention provides a variable-pitch bearing state online monitoring method and system, and relates to the technical field of variable-pitch bearings, and the method comprises the following steps: calculating the wind intensity coefficient of each variable-pitch bearing in each prediction time period through a wind speed prediction device, and marking the corresponding prediction time period; the state prediction device predicts each prediction time period according to the historical operation data and the wind intensity coefficient to obtain a prediction state data set; and constructing a wind field planning total health function, and generating a global optimal maintenance path through an iterative algorithm. By adopting the method, the nonlinear influence of the wind speed can be quantified, so that the system can pre-judge the risk in real time. And a global optimal maintenance path is generated by an iterative algorithm, so that the wind field planning total health function obtains an optimal solution, and the economic maximization extension of the bearing life is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to a variable pitch bearing, in particular to a variable pitch bearing state online monitoring method and system. BACKGROUND

[0002] The variable pitch bearing is the core transmission component of the variable pitch system of the wind turbine, and its service life is directly affected by the amplitude and frequency characteristics caused by the dynamic load of wind speed. Wind speed fluctuation causes the aerodynamic force acting on the blade to change nonlinearly, generating alternating inertial load, and then exciting bearing vibration; when the amplitude increases or the vibration frequency approaches the natural frequency of the system, resonance effect will be triggered, accelerating the fatigue spalling of the raceway and the loosening of the bolts, so it is necessary to monitor the state of the variable pitch bearing online.

[0003] However, the existing monitoring technology has significant defects: the traditional method analyzes the vibration effective value, which can detect bearing damage, but ignores the dynamic correlation between wind speed and vibration parameters, and lacks predictability, and cannot generate an optimal maintenance path through prediction. SUMMARY

[0004] To overcome the existing technical problems, the present application provides a variable pitch bearing state online monitoring method and system capable of generating an optimal maintenance path.

[0005] The present application adopts the following technical solutions.

[0006] A variable pitch bearing state online monitoring method, comprising a plurality of variable pitch bearings, each of which is provided with a state monitoring device, a state prediction device and a database;

[0007] Further comprising the following steps:

[0008] The wind speed prediction device calculates the wind intensity coefficient of each variable pitch bearing in each prediction time period and marks the corresponding prediction time period, and synchronizes the wind intensity coefficient to the cloud and each state prediction device;

[0009] The state monitoring device monitors the variable pitch bearing to obtain a bearing state data set, which is stored in the database and uploaded to the cloud. The database stores historical operation data, and the state prediction device predicts each prediction time period according to the historical operation data and the wind intensity coefficient to obtain a prediction state data set and upload it to the cloud;

[0010] The cloud combines the bearing state data set to the historical data set corresponding to the variable pitch bearing, constructs a wind farm planning total health function according to the prediction state data set, the historical data set and the wind intensity coefficient, and generates a global optimal maintenance path through an iterative algorithm to obtain an optimal solution of the wind farm planning total health function.

[0011] The historical data set includes maintenance time data, and a plurality of historical time periods, and historical average wind receiving intensity, historical average amplitude data and historical average vibration frequency data corresponding to the historical time periods one by one;

[0012] The maintenance time data includes a last maintenance time stamp, a last maintenance due time stamp, a last maintenance delay duration and an average maintenance duration, and the last maintenance delay duration is a duration between the last maintenance due time stamp and the last maintenance time stamp;

[0013] The specific steps for constructing the total health function of the wind farm planning include:

[0014] S11, selecting a historical time period between the last maintenance due time stamp and the last maintenance time stamp, and integrating the historical time period into a delay time period, calculating an actual risk degree and an actual dynamic penalty amplification coefficient of the variable pitch bearing in the delay time period according to the historical data set, and repeating the step to obtain the actual risk degree and the actual dynamic penalty amplification coefficient of each variable pitch bearing;

[0015] S12, calculating a device life coefficient of each variable pitch bearing according to the actual risk degree, the actual dynamic penalty amplification coefficient and the last maintenance delay duration, and if it is the first maintenance, the device life coefficient is equal to a preset initial health decay value E health ;

[0016] S13, calculating a predicted risk degree and a predicted dynamic penalty amplification coefficient of the variable pitch bearing in the current iteration according to the predicted state data set, and combining the last maintenance time stamp, the device life coefficient and the average maintenance duration to obtain the total health function of the wind farm planning in the current iteration.

[0017] Further improvement of the application, the expression of the total health function of the wind farm planning is as follows:

[0018]

[0019] Wherein, ψ main，i is the health decay value of the i th variable pitch bearing, β1 is the equipment aging weight factor, T now is the current time stamp, T last is the last maintenance time stamp, δ healt is the device life coefficient, β2 is the maintenance delay weight factor, T delay，pre is the duration between the current time stamp and the maintenance due time stamp, ω amplify，pre is the predicted dynamic penalty amplification coefficient, β3 is the predicted risk weight factor, W pre is the predicted risk degree.

[0020] The further improvement of the present application comprises the specific steps of calculating the wind intensity coefficient of each pitch bearing in each prediction time period and marking the corresponding prediction time period by the wind speed prediction device, which includes:

[0021] The wind speed prediction device predicts the future average wind speed and the future average wind direction in each prediction time period, and the wind speed prediction device is provided with a layout map having a plurality of pitch bearing position data and arrangement direction data;

[0022] The pitch bearing located at the tail end of the future average wind direction is selected as the wind speed intensity reference, the wind intensity coefficient of each pitch bearing in each prediction time period is calculated and the corresponding prediction time period is marked, and the expression of the wind intensity coefficient is as follows:

[0023]

[0024] Wherein, δ W，i，q is the wind intensity coefficient of the ith pitch bearing in the qth prediction time period, W wind，q is the future average wind speed of the qth prediction time period, ε i is the dynamic wind attenuation coefficient of the ith pitch bearing and the wind speed intensity reference, cosθ i，q is the angle between the future average wind direction of the qth prediction time period and the arrangement direction of the ith pitch bearing.

[0025] The further improvement of the present application comprises the specific steps of calculating the wind intensity coefficient of each pitch bearing in each prediction time period and marking the corresponding prediction time period by the wind speed prediction device, which includes:

[0026] The prediction state data includes prediction time period, wind intensity coefficient, prediction average amplitude data and prediction average vibration frequency data;

[0027] The specific steps of predicting each prediction time period according to the historical operation data and the wind intensity coefficient by the state prediction device to obtain the prediction state data set include:

[0028] S21, if the wind intensity coefficient is greater than the cut-out wind speed threshold, the prediction state data marked as shutdown is obtained;

[0029] S22, if the wind intensity coefficient is less than the cut-in wind speed threshold, the prediction state data marked as non-start is obtained;

[0030] S23, if the wind intensity coefficient is between the cut-out wind speed threshold and the cut-in wind speed threshold, calculating a preselected reference range according to the wind intensity coefficient and a preset similar intensity range, selecting a number of historical average wind intensities falling into the preselected reference range, extracting the wind intensity coefficient and the historical average wind intensity, and integrating the historical average wind intensity, historical average amplitude data corresponding to the historical average wind intensity, and historical average vibration frequency data corresponding to the historical average wind intensity into a historical reference data set, and calculating prediction state data according to the entire historical reference data set;

[0031] S24, repeating steps S21 to S23 for each prediction time period to obtain a plurality of prediction state data corresponding to the prediction time period one by one, and integrating the prediction state data into a prediction state data set.

[0032] Further improvement of the present application, after selecting a number of historical average wind intensities falling into the preselected reference range, further comprising the steps of:

[0033] If the number of historical average wind intensities falling into the preselected reference range exceeds the preset historical reference number value, calculate the difference between the current timestamp and the timestamp of each historical average wind intensity, if greater than the preset spatiotemporal reference threshold, do not select the historical average wind intensity;

[0034] After integrating into the historical reference data set, further comprising the step of: extracting the timestamp of the historical average wind intensity and merging it into the corresponding historical reference data set;

[0035] The expression for calculating the prediction state data according to the entire historical reference data set is as follows:

[0036]

[0037] wherein P l is the prediction value of the lth prediction data, N is the number of historical reference data sets, T n is the time decay weight coefficient of the nth historical reference data set, λ t is the time decay weight factor, is the timestamp of the nth historical reference data set, T now is the current timestamp, T allow is the dynamic reference time threshold, ω n is the dynamic decay weight of the nth historical reference data set, λ l is the distance decay factor of the lth prediction data, d n is the difference between the historical average wind intensity and the wind intensity coefficient in the nth historical reference data set, is the historical value of the same type as the lth prediction data in the nth historical reference data set.

[0038] Further improvement of the present application, the actual risk degree expression in step S11 is as follows:

[0039]

[0040] wherein, W Risk is the actual risk degree, T his is the number of historical time periods in the delay time period, W wind，O is the historical average wind intensity of the Oth historical time period, is the historical amplitude-frequency influence coefficient of the Oth historical time period, P hz，O is the historical average frequency data of the Oth historical time period, P hz，stand is the safe frequency value, P amp，O is the historical average amplitude data of the Oth historical time period, δ amp，stand is the safe amplitude value, and α1is the amplitude-frequency weight factor;

[0041] if the value of the actual risk degree is less than the preset health operation value, the actual dynamic penalty amplification coefficient is 0;

[0042] if the value of the actual risk degree is greater than the preset health operation value, the expression of the actual dynamic penalty amplification coefficient is as follows:

[0043]

[0044] wherein, λ pun is the penalty amplification weight factor, W Risk，stand is the health operation value;

[0045] The expression for calculating the equipment life coefficient of each pitch bearing is as follows:

[0046]

[0047] wherein, δ health，i is the equipment life coefficient of the ith pitch bearing, is the equipment life coefficient of the ith pitch bearing at the last maintenance, if it is the first maintenance, the item is equal to the preset initial health decay value E health , T delay，i is the last maintenance delay time of the ith pitch bearing, W Risk，i is the actual risk degree of the ith pitch bearing, ω amplify，i is the actual dynamic penalty amplification coefficient of the ith pitch bearing, and M is the number of pitch bearings.

[0048] Further improvement of the present application, the expression for predicting the risk degree in step 13 is as follows:

[0049]

[0050] Wherein, W pre is the predicted risk degree, T first is the number of the current iteration prediction time period, W wind，U is the wind intensity coefficient of the Uth prediction time period, is the predicted amplitude-frequency influence coefficient of the Uth prediction time period, P hz，U is the predicted average frequency data of the Uth prediction time period, P hz，stand is the safe frequency value, P amp，O is the predicted average amplitude data of the Uth prediction time period, δ amp，stand is the safe amplitude value, and α1 is the amplitude-frequency weight factor;

[0051] If the value of the predicted risk degree is less than the preset health running value, the predicted dynamic penalty amplification coefficient is 0.

[0052] If the value of the predicted risk degree is greater than the preset health running value, the expression of the predicted dynamic penalty amplification coefficient is as follows:

[0053]

[0054] Wherein, λ pun is the penalty amplification weight factor, W Risk，stand is the health running value.

[0055] Further improvement of the application, the specific steps of generating the global optimal maintenance path by the iteration algorithm include:

[0056] S31, according to the wind intensity coefficient and the position data of each variable pitch bearing, a virtual aging center is established, and the expression of the virtual aging center is as follows:

[0057]

[0058] Wherein, M is the number of variable pitch bearings, ψ main，i is the health attenuation value of the ith variable pitch bearing in the current iteration, POS i is the position of the ith variable pitch bearing;

[0059] S32, according to the virtual aging center, the global search is iterated to any variable pitch bearing, and the expression of the global search is as follows:

[0060]

[0061] Wherein, POS t+1 is the path position in the t+1th iteration, fit c is the current position POS t in the tth iteration to the position POS c of the cth variable pitch bearing.The cost value calculated by the cost function, POS c is the path position of the pitch bearing with the lowest cost value in the tth iteration, r is a random disturbance number, T t，c is the path length from the current position POS t to the cth pitch bearing position POS c , T c，ave is the average maintenance time of the cth pitch bearing, ψ main，c is the health degradation value of the cth pitch bearing in the current iteration, γ1, γ2, γ3 are cost weight factors, respectively.

[0062] According to the path length of the pitch bearing, the going-to time is calculated, and the daily working time is calculated in combination with the average maintenance time of the pitch bearing, if the daily working time is less than the preset working time threshold, then the local search iteration is performed to any one pitch bearing, the sum of the going-to time, the average maintenance time and the daily working time of the pitch bearing is calculated, and the daily working time is covered, the expression of the local search is as follows:

[0063]

[0064] If the daily working time is greater than the preset working time threshold, the working of the day is recorded whether it is completed or not at the end of the day, if it is completed, the global search iteration is performed to any one pitch bearing the next day, if it is not completed, the remaining working time is converted into the daily working time of the next day, and the local search iteration is performed to any one pitch bearing.

[0065] The application also proposes a pitch bearing state online monitoring system, which uses the method as described above, comprising:

[0066] A wind speed prediction device, which calculates the wind intensity coefficient of each pitch bearing in each prediction time period and marks the corresponding prediction time period, and synchronizes the wind intensity coefficient to the cloud and each state prediction device;

[0067] A state monitoring device, which is used for receiving the wind intensity coefficient marked with the corresponding prediction time period by the wind speed prediction device, and is used for monitoring the pitch bearing to obtain a bearing state data set and store it in a database and upload it to the cloud;

[0068] A database, which stores historical operation data and transmits the historical operation data to the state prediction device;

[0069] A state prediction device, which predicts each prediction time period according to the historical operation data and the wind intensity coefficient, obtains a prediction state data set and uploads it to the cloud;

[0070] Cloud, for merging the bearing state data set to the historical data set corresponding to the variable pitch bearing, constructing the wind farm planning total health function by the predicted state data set, the historical data set, the position data of the variable pitch bearing and the wind intensity coefficient, generating the global optimal maintenance path by the iterative algorithm, so that the wind farm planning total health function obtains the optimal solution.

[0071] The beneficial effects of the present application are: the traditional vibration effective value analysis can only detect the damage of the current bearing, while the present scheme predicts the wind intensity coefficient through the wind speed prediction device, quantifies the nonlinear influence of the wind speed, and enables the system to predict the vibration life risk of the variable pitch bearing under a specific wind speed in real time. The iterative algorithm generates the global optimal maintenance path to make the wind farm planning total health function obtain the optimal solution, realizing the economic maximum extension of the bearing life. BRIEF DESCRIPTION OF DRAWINGS

[0072] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0073] Fig. 1 is the main flowchart of the present application;

[0074] Fig. 2 is the calculation diagram of the wind intensity coefficient of the present application. DETAILED DESCRIPTION

[0075] The drawings are only used for illustrative description and cannot be understood as a limitation of the present patent; in order to better illustrate the present embodiment, some components in the drawings may be omitted, enlarged or reduced, and do not represent the average product size.

[0076] For those skilled in the art, it is understandable that some known structures and their descriptions in the drawings may be omitted. The technical solutions of the present application will be further described below in combination with the drawings and embodiments.

[0077] Reference Figs. 1-2 It can be seen that an online monitoring method for variable pitch bearing state includes a plurality of variable pitch bearings, and each variable pitch bearing is provided with a state monitoring device, a state prediction device and a database;

[0078] Further comprising the following steps:

[0079] The wind speed prediction device calculates the wind intensity coefficient of each variable pitch bearing in each predicted time period and marks the corresponding predicted time period, and synchronizes the wind intensity coefficient to the cloud and each state prediction device;

[0080] Further improvement of the present application, the specific steps of calculating the wind intensity coefficient of each pitch bearing in each prediction time period and marking the corresponding prediction time period by the wind speed prediction device include:

[0081] The future average wind speed and the future average wind direction in each prediction time period are predicted by the wind speed prediction device, and the wind speed prediction device is provided with a layout diagram having a plurality of pitch bearing position data and arrangement direction data;

[0082] The pitch bearing located at the tail end of the future average wind direction is selected as the wind speed intensity reference, the wind intensity coefficient of each pitch bearing in each prediction time period is calculated and the corresponding prediction time period is marked, and the expression of the wind intensity coefficient is as follows:

[0083]

[0084] Wherein, δ W，i，q is the wind intensity coefficient of the i th pitch bearing in the q th prediction time period, W wind，q is the future average wind speed in the q th prediction time period, ε i is the dynamic wind force attenuation coefficient of the i th pitch bearing and the wind speed intensity reference, cosθ i，q is the included angle between the future average wind direction in the q th prediction time period and the arrangement direction of the i th pitch bearing.

[0085] Each prediction time period is determined according to the predicted time range, for example, the work of maintaining all pitch bearings takes one week from start to finish, in the conventional prediction means, the prediction time period needs to predict at least one week (to plan the overall maintenance path in advance), and the longer the prediction time, the lower the prediction accuracy. For predicting the future average wind speed and the future average wind direction in the next week, each prediction time period is recommended to be 3 to 6 hours. However, since the global optimal maintenance path of the present application is calculated by an iterative algorithm, the next pitch bearing to be maintained can be calculated by the iterative algorithm when the i th pitch bearing is maintained, at this time, the prediction time period basically only needs to be one day, and under this premise, each prediction time period is recommended to be one hour, and the prediction accuracy is greatly improved.

[0086] The arrangement direction data is the orientation direction of the pitch bearing corresponding to the wind turbine blade. Specifically, as Fig. 2 The wind turbine located at the right upper corner is located at the tail end of the future average wind direction, so the wind turbine at the right upper corner is taken as the wind speed intensity reference, that is, the pitch bearing at the right upper corner is taken as the wind speed intensity reference. By calculating the wind intensity coefficient, the effective wind speed blowing through the wind turbine blade can be effectively determined.

[0087] For the dynamic wind force attenuation coefficient ε iThe specific value of the dynamic wind force attenuation coefficient ε of the pitch bearing will change for different wind speed intensity references, because the value is calculated using historical operation data, and if there is no historical value to be referenced, the current pitch bearing dynamic wind force attenuation coefficient ε i The value of the dynamic wind force attenuation coefficient ε is 1 by default. In the historical operation data of the pitch bearing, the historical average wind intensity with similar historical average wind direction is selected to calculate the dynamic wind force attenuation coefficient ε i It should be pointed out that the wind speed prediction device used to predict the future average wind speed and the future average wind direction is not installed on the wind turbine, and the anemometer used to detect the real-time wind speed is installed on the wind turbine, and the historical average wind intensity is calculated based on the wind speed measurement of the anemometer.

[0088] As a specific embodiment, if the i-th pitch bearing has the same historical average wind intensity as the pitch bearing serving as the wind speed intensity reference, the value of the dynamic wind force attenuation coefficient ε i is equal to 1. In addition, it should be pointed out that for wind turbines in the same area, there is usually a dominant wind direction (which is the reason for installing the wind turbine here), so their arrangement directions are basically consistent, their future average wind directions are basically consistent, and the pitch bearing serving as the wind speed intensity reference is basically determined, so that the value of the dynamic wind force attenuation coefficient ε i is basically unchanged, and a large amount of historical data is not needed for calculation.

[0089] The state monitoring device is used to monitor the pitch bearing, and the bearing state data set is stored in the database and uploaded to the cloud. The database stores historical operation data, and the state prediction device predicts each prediction time period based on the historical operation data and the wind intensity coefficient to obtain the prediction state data set and upload it to the cloud.

[0090] Further improvement of the application, the historical operation data includes a plurality of historical time periods, and historical average wind intensity, historical average amplitude data and historical average vibration frequency data corresponding to the historical time periods one by one;

[0091] The prediction state data includes prediction time period, wind intensity coefficient, prediction average amplitude data and prediction average vibration frequency data;

[0092] The specific steps of the state prediction device predicting each prediction time period based on the historical operation data and the wind intensity coefficient to obtain the prediction state data set include:

[0093] S21, if the wind intensity coefficient is greater than the cut-out wind speed threshold, the prediction state data marked as shutdown is obtained;

[0094] S22, if the wind intensity coefficient is less than the cut-in wind speed threshold, the prediction state data marked as non-start is obtained;

[0095] S23, if the wind intensity coefficient is between the cut-out wind speed threshold and the cut-in wind speed threshold, calculating a preselected reference range according to the wind intensity coefficient and a preselected similar intensity range, selecting a plurality of historical average wind intensities falling into the preselected reference range, extracting the wind intensity coefficient and the historical average wind intensities, and integrating the historical average wind intensities, the wind intensity coefficient, and historical average amplitude data and historical average vibration frequency data corresponding to the historical average wind intensities into a historical reference data set, and calculating prediction state data according to the entire historical reference data set;

[0096] S24, repeating steps S21 to S23 for each prediction time period to obtain a plurality of prediction state data corresponding to the prediction time period one by one, and integrating them into a prediction state data set.

[0097] Specifically, the cut-in wind speed refers to the minimum wind speed at which the wind turbine can start and begin generating electricity. The cut-out wind speed refers to the maximum wind speed at which the wind turbine stops running for safety considerations. Therefore, once the cut-out wind speed threshold is exceeded, the future variable pitch bearing will be in a shutdown state, and for the same reason, once the cut-in wind speed threshold is exceeded, the variable pitch bearing will be in an unstarted state. When the variable pitch bearing is in a shutdown or unstarted state, the value of the predicted dynamic penalty amplification coefficient ω amplify，pre is zero, and the value of the predicted risk degree W pre is also zero, greatly reducing the value of the health degradation value ψ main，i .

[0098] Further improvements of the present application include the following steps after selecting a plurality of historical average wind intensities falling into the preselected reference range:

[0099] If the number of historical average wind intensities falling into the preselected reference range exceeds a preselected historical reference number value, calculate the difference between the current timestamp and the timestamp of each historical average wind intensity, and if it is greater than a preselected spatiotemporal reference threshold, do not select the historical average wind intensity.

[0100] Since the state of the variable pitch bearing will continuously decline with the increase of the service life, when there are enough historical average wind intensities falling into the preselected reference range, the historical average wind intensities that are relatively old in time should be removed. Specifically, the historical reference number value is five, and the spatiotemporal reference threshold is one month. At this time, it needs to be reminded that if the wind intensity coefficient is high, resulting in a low number of historical average wind intensities falling into the preselected reference range, even if the difference between the historical average wind intensity timestamp and the current timestamp is greater than the preselected spatiotemporal reference threshold, the historical average wind intensity is still selected because in the case of high wind intensity coefficient, the influence of the wind intensity coefficient will exceed the influence of the service life of the variable pitch bearing.

[0101] After integrating into the historical reference dataset, a step of merging the time stamp of the historical average wind intensity into the corresponding historical reference dataset is further included;

[0102] The expression of the predicted state data according to the whole historical reference dataset is as follows:

[0103]

[0104] wherein P l is the predicted value of the lth prediction data, N is the number of the historical reference dataset, T n is the time decay weight coefficient of the nth historical reference dataset, λ t is the time decay weight factor, is the time stamp of the nth historical reference dataset, T now is the current time stamp, T allow is the dynamic reference time threshold, ω n is the dynamic decay weight of the nth historical reference dataset, λ l is the distance decay factor of the lth prediction data, d n is the difference between the historical average wind intensity and the wind intensity coefficient in the nth historical reference dataset, is the historical value of the same type as the lth prediction data in the nth historical reference dataset.

[0105] When the number of the historical reference dataset is less than or equal to the preset historical reference number value, and the wind intensity coefficient is greater than the full-load wind speed, the dynamic reference time threshold T allow is equal to the oldest time stamp in the historical reference dataset, so that the time decay weight coefficient of each historical reference dataset is the same. The reason why the wind intensity coefficient is required to be greater than the full-load wind speed is that when the wind intensity coefficient is between the cut-in wind speed and the full-load wind speed, the power increases approximately in a cubic relationship with the wind speed, and the corresponding amplitude and frequency data will have a sharp fluctuation. At this time, the influence of the service life of the variable pitch bearing is similar to the influence caused by the wind intensity coefficient. For other cases, the dynamic reference time threshold T allow is equal to the latest time stamp in the historical reference dataset.

[0106] The setting of the time decay weight factor λ t is used to optimize the value size for the convenience of the system processing data. Therefore, the determination of the value of the time decay weight factor λ t may be determined according to the difference between the oldest time stamp in the historical reference dataset and the latest time stamp in the historical reference dataset. Specifically, the unit of the time stamp is day, and assuming that the oldest time stamp and the latest time stamp are different by one month, the value of the decay weight factor λ t may be 1 / 15, so that T nThe minimum value is 0.135, facilitating the system to process data. l Can be 1.

[0107] The cloud collects the bearing state data into a historical data set corresponding to the variable pitch bearing, constructs a wind farm planning total health function according to the predicted state data set, the historical data set and the wind intensity coefficient, generates a globally optimal maintenance path through an iterative algorithm, so that the wind farm planning total health function obtains an optimal solution.

[0108] The traditional vibration effective value analysis can only detect the damage of the current bearing, while the present scheme predicts the wind intensity coefficient through a wind speed prediction device, quantifies the nonlinear influence of the wind speed, enables the system to predict the vibration life risk of the variable pitch bearing under a specific wind speed in real time, generates a globally optimal maintenance path through an iterative algorithm, so that the wind farm planning total health function obtains an optimal solution, and realizes economic maximization extension of the bearing life.

[0109] Further improvement of the present application, the historical data set includes maintenance time data, a plurality of historical time periods, and historical average wind intensity, historical average amplitude data and historical average vibration frequency data corresponding to the historical time periods one by one;

[0110] The maintenance time data includes the last maintenance timestamp, the last maintenance due timestamp, the last maintenance delay duration and the average maintenance duration, the last maintenance delay duration is the duration between the last maintenance due timestamp and the last maintenance timestamp;

[0111] The specific steps of constructing the wind farm planning total health function include:

[0112] S11, select the historical time period between the last maintenance due timestamp and the last maintenance timestamp, and integrate it into a delay time period, calculate the actual risk degree and the actual dynamic penalty amplification coefficient of the variable pitch bearing in the delay time period according to the historical data set, and repeat the step to obtain the actual risk degree and the actual dynamic penalty amplification coefficient of each variable pitch bearing;

[0113] Due to insufficient manpower, economic benefits and pitch bearing characteristics, etc., there is a long period of time without maintenance after the maintenance of all pitch bearings, and if some pitch bearings are maintained again in this period of time, over-maintenance will occur, such as new lubricating grease will react with old grease to form acidic substances to corrode the bearing raceway. Therefore, for wind turbines in the same area, periodic unified maintenance is usually used for maintenance, but when the equipment life coefficient of the pitch bearing is reduced, the maintenance frequency is higher and higher, for example, regular maintenance is usually once every six months, but after several years of use, it may need to be maintained once every two or three months. Therefore, if maintenance is directly performed according to the installation order of the wind turbine, the state of each pitch bearing cannot be fully considered, such as the wind turbine installed earliest, the time of which is far longer than that of other wind turbines without starting due to the long-term wind speed not reaching the cut-in wind speed, thereby the life loss of the pitch bearing is too small, and the wind turbine installed last is almost in full power wind speed for a long time, the high temperature generated by the rotation of the pitch bearing leads to the reduction of the viscosity of the lubricating oil, reduces the protection of the bearing, and accelerates the oxidation and evaporation of the lubricating oil, if maintenance is performed according to the installation order of the wind turbine, the wind turbine installed last but should be maintained first will be maintained last.

[0114] Therefore, in the process of maintenance, some pitch bearings will be delayed in maintenance due to the maintenance of other pitch bearings when they reach the time of needing maintenance, thereby the pitch bearings have additional losses, and when the total health function of the wind farm planning is constructed, the additional losses should be fully considered.

[0115] Further improvement of the present application, the actual risk degree expression in step S11 is as follows:

[0116]

[0117] Wherein, W Risk is the actual risk degree, T his is the number of historical time periods in the delay time period, W wind，O is the historical average wind intensity of the Oth historical time period, is the historical amplitude-frequency influence coefficient of the Oth historical time period, P hz，O is the historical average frequency data of the Oth historical time period, P hz，stand is the safe frequency value, P amp，O is the historical average amplitude data of the Oth historical time period, δ amp，stand is the safe amplitude value, and α1 is the amplitude-frequency weight factor.

[0118] If the value of the actual risk degree is less than the preset health running value, the actual dynamic penalty amplification coefficient is 0;

[0119] If the value of the actual risk degree is greater than the preset health operation value, the expression of the actual dynamic penalty amplification coefficient is as follows:

[0120]

[0121] wherein λ pun is a penalty amplification weight factor, W Risk，stand is the health operation value;

[0122] Specifically, the preset health operation value is equal to the full-load wind speed, because although the pitch system will fine-tune the blade angle according to the control system instruction to maintain stable output, when the actual risk degree is large, it will cause the mechanical load and aerodynamic load borne by the unit to increase significantly, and the blade angle needs to be adjusted frequently, causing the rollers inside the bearing to constantly contact and slide with the raceway, generating more wear and fretting corrosion, and causing the overall life of the pitch bearing to be seriously wasted. Specifically, the value of the penalty amplification weight factor λ pun is 1.5.

[0123] S12, calculate the equipment life coefficient of each pitch bearing according to the actual risk degree, the actual dynamic penalty amplification coefficient, and the last maintenance delay length, if it is the first maintenance, the equipment life coefficient is equal to the preset initial health attenuation value E health .

[0124] Since the amplitude and frequency will cause great damage to the pitch bearing, calculating the historical amplitude and frequency influence coefficient according to the historical average amplitude data and the historical average frequency data can effectively provide a safety margin for the actual risk degree. Specifically, the value of α1 is 1. If the pitch bearing does not have a last maintenance delay length, the actual risk degree is considered to be 0.

[0125] The expression for calculating the equipment life coefficient of each pitch bearing is as follows:

[0126]

[0127] wherein δ health，i is the equipment life coefficient of the i-th pitch bearing, is the equipment life coefficient of the i-th pitch bearing at the last maintenance, if it is the first maintenance, the term is equal to the preset initial health attenuation value E health , T delay，i is the last maintenance delay length of the i-th pitch bearing, W Risk，i is the actual risk degree of the i-th pitch bearing, ω amplify，i is the actual dynamic penalty amplification coefficient of the i-th pitch bearing, and M is the number of pitch bearings.

[0128] The last maintenance delay length T delay，i The unit is day. When the actual risk degree is greater, the value of the actual dynamic penalty amplification coefficient is greater, the degree of change of the equipment life coefficient of the variable pitch bearing device is improved, and the health degradation value is indirectly improved, so that maintenance is more preferentially performed (the optimal solution of the total health function of the wind farm means that the health degradation value ψ main，i of the wind turbine of the entire wind farm is the lowest). The formula is processed by normalization, which can comprehensively balance the values of the equipment life coefficient changes of all variable pitch bearings, so that the equipment life coefficient is not sharply reduced due to the increase of the actual risk degree. Specifically, the initial health degradation value E health is determined according to the number of maintenance times in the whole cycle of the wind turbine. According to the full-size structure test of the wind turbine blade in Wind Turbine Generator and actual verification, the number of maintenance times in the whole life cycle is nearly 150 times, and the initial health degradation value E health may be 200.

[0129] S13, calculating the predicted risk degree and the predicted dynamic penalty amplification coefficient of the variable pitch bearing in the current iteration according to the predicted state data set, and combining the last maintenance time stamp, the equipment life coefficient and the average maintenance length to obtain the total health function of the wind farm planning in the current iteration.

[0130] Further improvement of the application, the expression of the predicted risk degree in step 13 is as follows:

[0131]

[0132] Wherein, W pre is the predicted risk degree, T first is the number of the current iteration prediction period, W wind，U is the wind intensity coefficient of the Uth prediction period, is the predicted amplitude and frequency influence coefficient of the Uth prediction period, P hz，U is the predicted average frequency data of the Uth prediction period, P hz，stand is the safe frequency value, P amp，O is the predicted average amplitude data of the Uth prediction period, δ amp，stand is the safe amplitude value, and α1 is the amplitude and frequency weight factor.

[0133] Specifically, as planned to maintain the i-th variable pitch bearing today, the next variable pitch bearing to be maintained is selected by the iterative cost algorithm while maintaining, before the iterative cost algorithm calculation, the wind speed prediction device updates the wind strength coefficient of the day, selects the longest dynamic average maintenance time in the un-maintained variable pitch bearing, if the longest variable pitch bearing needs to be maintained day by day, the number of iterative prediction time periods is the prediction time period in a day, in an embodiment of the present application, the prediction time period is one hour, so the number of iterative prediction time periods is 24*1=24.

[0134] If the value of the predicted risk degree is less than the preset health running value, the predicted dynamic penalty amplification coefficient is 0;

[0135] If the value of the predicted risk degree is greater than the preset health running value, the expression of the predicted dynamic penalty amplification coefficient is as follows:

[0136]

[0137] Wherein, λ pun is a penalty amplification weight factor, W Risk，stand is a health running value.

[0138] Further improvement of the present application, the expression of the total health function of the wind field planning is as follows:

[0139]

[0140] Wherein, ψ main，i is the health attenuation value of the i-th variable pitch bearing, β1 is a device aging weight factor, T now is the current time stamp, T last is the last maintenance time stamp, δ healt is a device life coefficient, β2 is a maintenance delay weight factor, T delay，pre is the time length between the current time stamp and the time stamp of the maintenance to be performed, ω amplify，pre is a predicted dynamic penalty amplification coefficient, β3 is a predicted risk weight factor, W pre is a predicted risk degree.

[0141] The health attenuation value calculation formula can fully consider the time interval from the last maintenance of the variable pitch bearing, the device life coefficient, the maintenance delay time T delay，pre , the predicted dynamic penalty amplification coefficient and the predicted risk degree, so that the variable pitch bearing with longer maintenance interval, lower device life coefficient and higher predicted risk degree due to maintenance delay is preferentially maintained. Specifically, the value of β3 can refer to the cut-in wind speed and the cut-out wind speed, for example, β2 is 0.5, and since the initial health attenuation value Ehealth Set to 200, so beta1 is set to 66 by default, T now -T last Here, the unit is month, when the equipment life coefficient decreases, the value of beta1 can also be increased accordingly, for example, when the pitch bearing is maintained once every three months on average instead of once every six months on average, the value of beta1 can be adjusted to 132, which can be manually set or automatically set according to the equipment life coefficient, for example, when the equipment life coefficient is 199, the value of beta1 is 99, of course, the maintenance frequency of the pitch bearing is most accurate according to the artificial judgment at the last maintenance.

[0142] At this time, it should be noted that the meaning of the wind field planning total health function obtaining the optimal solution is to make the health degradation value ψ main，i of the entire wind field wind turbine lowest.

[0143] Further improvement of the present application, the specific steps of generating a global optimal maintenance path by an iterative algorithm include:

[0144] S31, a virtual aging center is established according to the wind intensity coefficient and position data of each pitch bearing, and the expression of the virtual aging center is as follows:

[0145]

[0146] Wherein, M is the number of pitch bearings, ψ main，i is the health degradation value of the i-th pitch bearing in the current iteration, POS i is the position of the i-th pitch bearing;

[0147] By establishing the virtual aging center, when the health degradation value ψ main，i is higher, the weight of its POS i position is greater, so that it can be close to the highest health degradation value ψ main，i , and the virtual aging center is updated once after maintaining each pitch bearing, and then a new virtual aging center is obtained.

[0148] S32, according to the virtual aging center, iteratively search globally to any pitch bearing, and the expression of the global search is as follows:

[0149]

[0150] Wherein, POS t+1 is the path position in the t+1 iteration, fit c is the current position POS t in the t-th iteration to the position POS c of the c-th pitch bearing, the value of the cost function calculated by the cost function, POS cis the path position of the pitch bearing with the lowest cost value in the tth iteration, r is a random disturbance number, T t，c is the path length from the current position POS t to the cth pitch bearing position POS c , T c，ave is the average maintenance time of the cth pitch bearing, ψ main，c is the health degradation value of the cth pitch bearing in the current iteration, γ1, γ2, γ3 are cost weight factors, respectively;

[0151] The global search can make the maintenance personnel tend to the area with the highest average health degradation value ψ main，i to perform maintenance. Of course, based on the setting of the cost function, if the health degradation value ψ main，i of the pitch bearing at the virtual aging center is low, and the health degradation values ψ main，i of the pitch bearings around are high, it is very likely that the pitch bearing with the higher health degradation value ψ main，i is given priority to maintenance. The random disturbance number r is a random number between 0 and 1, which can reduce the disturbance intensity after being squared. The logic of the formula is that the cost value of each pitch bearing is calculated with the virtual aging center as the origin, the pitch bearing with the lowest cost value is selected as the next maintenance target, and the lower the cost value is, the greater the influence of the position of the cth pitch bearing is, that is, iteration is performed to the position of the pitch bearing with the lowest cost value.

[0152] The calculation formula of the cost function is that the closer the cth pitch bearing is to the current position, the shorter the average maintenance time of the pitch bearing is, and the higher the health degradation value ψ main，i is, the lower the cost value is, so that it is given priority to maintenance. Specifically, T t，c uses kilometers as the unit, γ1 uses 2, T c，ave uses days as the unit, γ2 uses 0.5, and γ3 uses 6, thereby improving the influence weight of the health degradation value ψ main，i . If there is no pitch bearing with a cost value less than 1, and at least two pitch bearings have a cost value equal to 1, then one pitch bearing is randomly selected.

[0153] The travel time to the pitch bearing is calculated according to the path length of the pitch bearing, and the daily work time is calculated in combination with the average maintenance time of the pitch bearing. If the daily work time is less than the preset work time threshold, then the local search is iterated to any pitch bearing, the sum of the travel time, the average maintenance time and the daily work time of the pitch bearing is calculated, and the daily work time is covered. The expression of the local search is as follows:

[0154]

[0155] If the daily working time is greater than the preset working time threshold, whether the daily work is completed is recorded when the daily work is completed, if completed, the next day is searched iteratively to any one of the pitch bearing, if not completed, the remaining working time is converted to the daily working time of the next day, and the local search is iterated to any one of the pitch bearing.

[0156] Unlike global search, the core of local search is that the value of each pitch bearing is calculated based on the current position, and the pitch bearing with the lowest value is selected to calculate the next maintenance position. The lower the value, the greater the influence of the position of the cth pitch bearing.

[0157] The global optimal maintenance path is generated by the iterative algorithm, which has the advantages of interruptibility, which can pause maintenance in case of sudden situation, and reiterate to calculate the next pitch bearing to be maintained, and can comprehensively consider the distance to each pitch bearing, the time required for maintenance of each pitch bearing, and the health degradation value of each pitch bearing main，i , so that the wind farm planning total health function obtains the optimal solution, that is, the maintenance planning maximizes the economic benefit. In addition, in the technical scheme of the application, the six-month maintenance interval of the newly installed pitch bearing and the three-month maintenance interval of the pitch bearing with a certain age can be comprehensively planned in the algorithm. As mentioned above, the value of each pitch bearing β1 can be manually set according to the last maintenance by artificial judgment, to adjust the health degradation value of each pitch bearing main，i . In the weight of the maintenance interval time. After the maintenance of a certain pitch bearing, the health degradation value of the pitch bearing main，i is approximately equal to zero, and the second term is equal to zero because there is no delay, so the health degradation value of the pitch bearing main，i is much lower than that of other pitch bearings, so other pitch bearings are preferentially maintained, so that the iterative algorithm is performed uninterruptedly for 365 days a year, and the best maintenance strategy for the entire wind farm can be obtained. As a further improvement of the application, in order to avoid the same pitch bearing being maintained multiple times in a short period of time and thus damaging the service life of the pitch bearing, when the second term is equal to 0, no maintenance is performed.

[0158] In addition, the iterative algorithm can consider the maintenance work of each team, if the maintenance work is not completed on the same day, the maintenance of the pitch bearing is continued the next day, and the local search is performed based on the current position. If the work is completed on the same day, the next day can start from the position of the virtual aging center. It needs to be reminded that when the global search is performed, the current position POS t of the cost function refers to the set departure position of them.

[0159] The application further provides a variable pitch bearing state online monitoring system, which uses the method described above, comprising:

[0160] A wind speed prediction device, which calculates the wind intensity coefficient of each variable pitch bearing in each prediction time period and marks the corresponding prediction time period, and synchronizes the wind intensity coefficient to the cloud and each state prediction device;

[0161] A state monitoring device, which receives the wind intensity coefficient marked by the wind speed prediction device with the corresponding prediction time period, monitors the variable pitch bearing, obtains a bearing state data set, stores the bearing state data set in a database, and uploads the bearing state data set to the cloud;

[0162] A database, which stores historical operation data and transmits the historical operation data to the state prediction device;

[0163] A state prediction device, which predicts each prediction time period according to the historical operation data and the wind intensity coefficient, obtains a prediction state data set, and uploads the prediction state data set to the cloud;

[0164] A cloud, which combines the bearing state data set to a historical data set corresponding to the variable pitch bearing, constructs a wind farm planning total health function by using the prediction state data set, the historical data set, position data of the variable pitch bearing, and the wind intensity coefficient, generates a global optimal maintenance path by using an iterative algorithm, and obtains an optimal solution of the wind farm planning total health function.

[0165] Obviously, the above embodiments of the application are only examples for clearly illustrating the application, and are not intended to limit the implementation modes of the application. Based on the above description, other different forms of changes or variations can be made by those skilled in the art. It is not necessary or possible to exhaust all the implementation modes. Any modification, equivalent replacement, and improvement made within the spirit and principle of the application should be included in the protection scope of the claims of the application.

Claims

1. A method of online monitoring of a pitch bearing condition, characterized by, The wind turbine includes a plurality of pitch bearings, each of which is provided with a state monitoring device, a state prediction device and a database; Further comprising the following steps: The wind speed prediction device calculates the wind intensity coefficient of each pitch bearing in each prediction time period and marks the corresponding prediction time period, and synchronizes the wind intensity coefficient to the cloud and each state prediction device; The state monitoring device monitors the pitch bearing to obtain a bearing state data set, which is stored in the database and uploaded to the cloud. The database stores historical operation data. The state prediction device predicts each prediction time period based on the historical operation data and the wind intensity coefficient to obtain a prediction state data set and upload it to the cloud; The cloud combines the bearing state data set with the historical data set corresponding to the pitch bearing, constructs a wind farm planning total health function based on the prediction state data set, the historical data set and the wind intensity coefficient, and generates a global optimal maintenance path through an iterative algorithm to obtain an optimal solution of the wind farm planning total health function.

2. The method of claim 1, wherein, The historical data set includes maintenance time data, a plurality of historical time periods, and historical average wind intensity, historical average amplitude data and historical average vibration frequency data corresponding to the historical time periods one by one; The maintenance time data includes a last maintenance timestamp, a last maintenance due timestamp, a last maintenance delay duration and an average maintenance duration. The last maintenance delay duration is the duration between the last maintenance due timestamp and the last maintenance timestamp; The specific steps of constructing the wind farm planning total health function include: S11, selecting a historical time period between the last maintenance due timestamp and the last maintenance timestamp, and integrating it into a delay time period. Calculate the actual risk degree and the actual dynamic penalty amplification coefficient of the pitch bearing in the delay time period based on the historical data set. Repeat the step to obtain the actual risk degree and the actual dynamic penalty amplification coefficient of each pitch bearing; S12, calculate the equipment life coefficient of each pitch bearing according to the actual risk degree, the actual dynamic penalty amplification coefficient, and the last maintenance delay length. If it is the first maintenance, the equipment life coefficient is equal to the preset initial health degradation value E health ; S13, calculate the prediction risk degree and the prediction dynamic penalty amplification coefficient of the pitch bearing in the current iteration based on the prediction state data set, and combine the last maintenance timestamp, the equipment life coefficient and the average maintenance duration to obtain the wind farm planning total health function in the current iteration.

3. The method of claim 2, wherein, The expression of the wind farm planning total health function is as follows: wherein ψ main，i is the health degradation value of the i-th pitch bearing, β1is the equipment aging weight factor, T now is the current timestamp, T last is the last maintenance timestamp, is the equipment life coefficient, β2is the maintenance delay weight factor, T delay，pre is the time length between the current timestamp and the timestamp of this maintenance, ω amplify，pre is the predicted dynamic penalty amplification coefficient, β3is the predicted risk weight factor, W pre is the predicted risk degree.

4. The method of claim 1, wherein, The specific steps of calculating the wind intensity coefficient of each pitch bearing in each prediction time period by the wind speed prediction device and marking the corresponding prediction time period include: The wind speed prediction device predicts the future average wind speed and the future average wind direction in each prediction time period. The wind speed prediction device is provided with a layout diagram having a plurality of pitch bearing position data and arrangement direction data; Select a pitch bearing located at the tail end of the future average wind direction as the wind speed intensity reference, calculate the wind intensity coefficient of each pitch bearing in each prediction time period and mark the corresponding prediction time period. The expression of the wind intensity coefficient is as follows: wherein δ W，i，q is the wind intensity coefficient of the ithpitch bearing at the qthprediction time period, W wind，q is the future average wind speed of the qthprediction time period, ε i is the dynamic wind force attenuation coefficient of the ithpitch bearing with the wind speed intensity reference, cos θ i，q is the angle between the future average wind direction of the qthprediction time period and the arrangement direction of the ithpitch bearing.

5. The method of claim 1, wherein, The historical operation data includes a plurality of historical time periods, and historical average wind intensity, historical average amplitude data and historical average vibration frequency data corresponding to the historical time periods one by one; The prediction state data includes prediction time period, wind intensity coefficient, prediction average amplitude data and prediction average vibration frequency data; The specific steps of obtaining the predicted state data set by the state prediction device according to historical operation data and wind intensity coefficients of each prediction time period include: S21, if the wind intensity coefficient is greater than the cut-out wind speed threshold, the predicted state data marked as shutdown is obtained; S22, if the wind intensity coefficient is less than the cut-in wind speed threshold, the predicted state data marked as non-start is obtained; S23, if the wind intensity coefficient is between the cut-out wind speed threshold and the cut-in wind speed threshold, a preselected reference range is calculated according to the wind intensity coefficient and a preselected similar intensity range, a plurality of historical average wind intensities falling within the preselected reference range are selected, the wind intensity coefficient and the historical average wind intensity are extracted, and the historical average amplitude data and the historical average vibration frequency data corresponding to the historical average wind intensity are integrated into a historical reference data set, and the predicted state data is calculated according to the entire historical reference data set; S24, steps S21 to S23 are repeated for each prediction time period to obtain a plurality of predicted state data corresponding to the prediction time period, and the predicted state data is integrated into a predicted state data set.

6. The method of claim 5, wherein, After selecting a plurality of historical average wind intensities falling within the preselected reference range, the following steps are further included: If the number of historical average wind intensities falling within the preselected reference range exceeds a preselected historical reference number value, the difference between the current timestamp and the timestamp of each historical average wind intensity is calculated, and if the difference is greater than a preselected spatiotemporal reference threshold, the historical average wind intensity is not selected; After integration into the historical reference data set, the following step is further included: the timestamp of the historical average wind intensity is extracted and merged into the corresponding historical reference data set; The expression of calculating the predicted state data according to the entire historical reference data set is as follows: wherein P l is the predicted value of the lth prediction data, N is the number of historical reference data sets, T n is the n th historical reference data set time decay weight coefficient, λ t is the time decay weight factor, is the timestamp of the n th historical reference data set, T now is the current timestamp, T allow is the dynamic reference time threshold, ω n is the dynamic decay weight of the n th historical reference data set, λ l is the distance decay factor of the lth prediction data, d n is the difference between the historical average wind intensity and the wind intensity coefficient in the n th historical reference data set, is the historical value of the same type as the lth prediction data in the n th historical reference data set.

7. The method of claim 2, wherein, The expression of the actual risk degree in step S11 is as follows: wherein W Risk is the actual risk level, T his is the number of historical time periods in the delay period, W wind，O is the historical average wind intensity of the Oth historical time period, is the historical amplitude-frequency influence coefficient of the Oth historical time period, P hz，O is the historical average frequency data of the Oth historical time period, P hz，stand is the safe frequency value, P amp，O is the historical average amplitude data of the Oth historical time period, δ amp，stand is the safe amplitude value, and α1is the amplitude-frequency weight factor. If the value of the actual risk degree is less than a preselected healthy operation value, the actual dynamic penalty amplification coefficient is 0; If the value of the actual risk degree is greater than the preselected healthy operation value, the expression of the actual dynamic penalty amplification coefficient is as follows: where λ pun is a penalty amplification weight factor, W Risk，stand is a healthy operation value; The expression of calculating the equipment life coefficient of each pitch bearing is as follows: wherein, is the equipment life coefficient of the i-th pitch bearing, is the equipment life coefficient of the i-th pitch bearing at the last maintenance, if this is the first maintenance, then this item is equal to the preset initial health degradation value E health , T delay，i is the last maintenance delay length of the i-th pitch bearing, W Risk，i is the actual risk degree of the i-th pitch bearing, ω amplify，i is the actual dynamic penalty amplification coefficient of the i-th pitch bearing, M is the number of pitch bearings.

8. The method of claim 2, wherein, The expression of the predicted risk degree in step 13 is as follows: wherein W pre is the predicted risk level, T first is the number of current iteration prediction time periods, W wind，U is the wind intensity coefficient of the Uth prediction time period, is the predicted amplitude-frequency influence coefficient of the Uth prediction time period, P hz，U is the predicted average frequency data of the Uth prediction time period, P hz，stand is the safe frequency value, P amp，O is the predicted average amplitude data of the Uth prediction time period, δ amp，stand is the safe amplitude value, and a1is the amplitude-frequency weight factor. If the value of the predicted risk degree is less than a preselected healthy operation value, the predicted dynamic penalty amplification coefficient is 0; If the value of the predicted risk degree is greater than the preselected healthy operation value, the expression of the predicted dynamic penalty amplification coefficient is as follows: where λ pun is a penalty amplification weight factor, W Risk，stand is a healthy operation value.

9. The method of claim 3, wherein, The specific steps of generating a globally optimal maintenance path through an iterative algorithm include: S31, a virtual aging center is established according to the wind intensity coefficient and the position data of each pitch bearing, and the expression of the virtual aging center is as follows: where M is the number of pitch bearings, ψ main，i is the health degradation value of the i-th pitch bearing at the current iteration, POS i is the position of the i-th pitch bearing; S32, global search iteration is performed to any pitch bearing according to the virtual aging center, and the expression of the global search is as follows: Among them, POS t+1 It is the path position in iteration t+1, fit c It is the current position POS in the t-th iteration. t To the c-th pitch bearing position POS c The cost value calculated using the cost function, POS c It is the path position of the pitch bearing with the lowest generation value in the t-th iteration, where r is the number of random disturbances, and T t，c This is the current location (POS). t To the c-th pitch bearing position POS c The path length, T c，ave It is the average maintenance time of the c-th pitch bearing, ψ main，c γ1, γ2, and γ3 are the health degradation values ​​of the c-th pitch bearing in the current iteration, and γ1, γ2, and γ3 are cost weighting factors, respectively. The travel time to the pitch bearing is calculated according to the path length of the pitch bearing, and the daily work time is calculated by combining the average maintenance time of the pitch bearing, if the daily work time is less than a preselected work time threshold, local search iteration is performed to any pitch bearing, the sum of the travel time, the average maintenance time and the daily work time of the pitch bearing is calculated, and the daily work time is covered, and the expression of the local search is as follows: If the daily working time is greater than the preset working time threshold, it is recorded whether the daily work is completed at the end of the day, if completed, the next day is searched globally to any one pitch bearing, if not completed, the remaining working time is converted to the daily working time of the next day, and the local search is iterated to any one pitch bearing.

10. An online monitoring system of a pitch bearing condition, characterized by The method of any one of claims 1-9, comprising: a wind speed prediction device, calculating the wind intensity coefficient of each pitch bearing in each prediction time period and marking the corresponding prediction time period, synchronizing the wind intensity coefficient to the cloud and each state prediction device; a state monitoring device for receiving the wind intensity coefficient marked by the wind speed prediction device with the corresponding prediction time period, for monitoring the pitch bearing, obtaining the bearing state data set and uploading to the cloud; a database, storing historical operation data, transmitting the historical operation data to the state prediction device; a state prediction device, predicting each prediction time period according to the historical operation data and the wind intensity coefficient, obtaining the prediction state data set and uploading to the cloud; a cloud, for merging the bearing state data set to the historical data set corresponding to the pitch bearing, constructing the wind farm planning total health function with the prediction state data set, the historical data set, the location data of the pitch bearing and the wind intensity coefficient, generating the global optimal maintenance path by the iteration algorithm, so as to obtain the optimal solution of the wind farm planning total health function.

Images