Intelligent early warning method for primary air fan based on fan operation state evaluation

Abstract

This invention discloses an intelligent early warning method for primary wind turbines based on wind turbine operating status assessment, comprising the following steps: data acquisition, data preprocessing, operating condition identification, feature parameter extraction, benchmark value determination, deviation calculation, status assessment, trend analysis, early warning determination, and early warning output. This invention effectively avoids the problem of high false alarm rates under different operating conditions by automatically identifying operating conditions and dynamically determining benchmark values ​​for each condition, making early warnings more accurate and reliable. Through trend analysis of the slope of health status changes, it can capture early signs of performance degradation before the overall health status drops to the alarm threshold, achieving advanced early warning. The hierarchical early warning mechanism of primary and advanced warnings provides operators with differentiated response times, helping to rationally arrange maintenance plans and avoid unplanned downtime.

Description

Technical Field

[0001] This invention relates to the field of wind turbine operating status assessment technology, and specifically to a method for intelligent early warning of primary wind turbines based on wind turbine operating status assessment. Background Technology

[0002] As a core auxiliary equipment on the boiler side of a thermal power generating unit, the operational reliability of the primary air fan directly affects the safe and stable operation of the unit. Traditional primary air fan condition monitoring mainly relies on operators periodically inspecting and comparing thresholds for single parameters such as bearing temperature, vibration amplitude, and motor current. An alarm is triggered when a parameter exceeds a preset alarm value. This early warning method based on a single parameter threshold has significant shortcomings:

[0003] On the one hand, a single parameter is difficult to fully reflect the overall health status of the equipment, and it is easy to miss reports; on the other hand, fixed thresholds cannot adapt to the differences in the operating characteristics of the fan under different operating conditions, and the false alarm rate is high when the load changes.

[0004] In recent years, some existing technologies have attempted to incorporate multi-source monitoring data into wind turbine condition assessment, achieving fault early warning through data fusion and model analysis. However, most of these methods have not fully considered the differences in parameter benchmarks of primary wind turbines under different operating conditions, nor have they established a complete technical chain from multi-parameter deviation assessment to comprehensive health measurement and trend early warning, resulting in significant room for improvement in the accuracy and timeliness of early warnings. Summary of the Invention

[0005] To address this, the present invention provides an intelligent early warning method for primary wind turbines based on wind turbine operating status assessment. This method solves the problem that existing technologies do not fully consider the differences in parameter benchmarks of primary wind turbines under different operating conditions, nor do they establish a complete technical chain from multi-parameter deviation assessment to comprehensive health measurement and trend early warning, resulting in significant room for improvement in the accuracy and timeliness of early warnings.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment includes the following steps:

[0008] Data acquisition steps: Acquire multi-source monitoring data of the primary fan during operation at a preset sampling period. The multi-source monitoring data includes bearing vibration amplitude, bearing temperature, motor current, fan inlet pressure, and fan outlet pressure.

[0009] Data preprocessing steps: outlier removal and missing value imputation are performed on the multi-source monitoring data to obtain a preprocessed monitoring data sequence;

[0010] Operating condition identification steps: Based on the motor current and the fan inlet pressure, identify the current operating condition of the primary air fan, including startup condition, stable operation condition, and variable load condition;

[0011] Feature parameter extraction step: For the operating condition, extract feature parameters corresponding to the current operating condition from the preprocessed monitoring data sequence. The feature parameters include vibration feature values, temperature feature values, current feature values, and pressure feature values.

[0012] The benchmark value determination step involves determining the benchmark value and benchmark fluctuation range of each characteristic parameter under the operating condition based on historical monitoring data corresponding to the current operating condition of the primary air turbine under historical normal operating conditions.

[0013] Deviation calculation steps: Calculate the deviation between the real-time measured value of each characteristic parameter and the corresponding benchmark value, and determine the deviation level of each characteristic parameter according to the benchmark fluctuation range of each characteristic parameter;

[0014] Status assessment steps: Based on the deviation levels of each of the aforementioned characteristic parameters, calculate the overall health of the primary air fan according to the following formula:

[0015] H = 100 - ∑(ωi × Li)

[0016] Where H represents the overall health score, ωi represents the weight coefficient of the i-th feature parameter, Li represents the deduction value corresponding to the deviation level of the i-th feature parameter, and ∑ represents the sum of the weighted deduction values ​​of all feature parameters;

[0017] Trend analysis steps: Construct a health time series from the comprehensive health scores at the current sampling time and the previous N sampling times, perform linear regression analysis on the health time series, and calculate the slope of the health score change;

[0018] Warning determination steps: When the overall health level is lower than the first health level threshold and the slope of the health level change is negative, a primary warning signal is issued; when the overall health level is lower than the second health level threshold, a high-level warning signal is issued, wherein the second health level threshold is less than the first health level threshold.

[0019] Early warning output steps: Output the primary early warning signal or the advanced early warning signal to the monitoring terminal, and simultaneously record the multi-source monitoring data and the comprehensive health status at the time of the early warning.

[0020] Preferably, in the data preprocessing step, the outlier removal is performed using the Raida criterion, and monitoring data exceeding the mean ± 3 times the standard deviation are identified as outliers and removed; the missing value imputation is performed using linear interpolation, which uses monitoring data from adjacent times before and after the missing time to imput the missing value.

[0021] Preferably, in the operating condition identification step, identifying the current operating condition of the primary air blower based on the motor current and the blower inlet pressure specifically includes: when the real-time value of the motor current is less than the starting current threshold, it is determined to be the starting condition; when the real-time value of the motor current is greater than or equal to the starting current threshold and the absolute value of the rate of change of the motor current over M consecutive sampling periods is less than the rate of change threshold, it is determined to be the stable operating condition; when the real-time value of the motor current is greater than or equal to the starting current threshold and the absolute value of the rate of change of the motor current over M consecutive sampling periods is greater than or equal to the rate of change threshold, it is determined to be the variable load condition.

[0022] Preferably, in the feature parameter extraction step, the vibration feature value is taken as the root mean square value of the bearing vibration amplitude within a preset time window, the temperature feature value is taken as the average value of the bearing temperature within the preset time window, the current feature value is taken as the average value of the motor current within the preset time window, and the pressure feature value is taken as the average value of the difference between the fan outlet pressure and the fan inlet pressure within the preset time window.

[0023] Preferably, in the benchmark value determination step, the historical monitoring data under normal operating conditions is the monitoring data of the primary wind turbine within the first K days after it is put into operation, which is confirmed by manual verification to be operating without faults and under the operating conditions; the benchmark value is the arithmetic mean of the corresponding characteristic parameters in the historical monitoring data, and the benchmark fluctuation range is the average value of the corresponding characteristic parameters in the historical monitoring data ± 2 times the standard deviation.

[0024] Preferably, in the deviation calculation step, the deviation is calculated as follows: when the real-time measured value of the feature parameter is greater than the benchmark value, the deviation is equal to (real-time measured value - benchmark value) / benchmark value × 100%; when the real-time measured value of the feature parameter is less than the benchmark value, the deviation is equal to (benchmark value - real-time measured value) / benchmark value × 100%; the deviation level is determined according to the preset deviation range into which the deviation falls, different preset deviation ranges correspond to different deduction values, and the deduction value increases as the deviation increases.

[0025] Preferably, the preset deviation range includes a first deviation range, a second deviation range, and a third deviation range. The deviation range corresponding to the first deviation range is greater than or equal to 0 and less than 10%, with a corresponding deduction of 5 points. The deviation range corresponding to the second deviation range is greater than or equal to 10% and less than 20%, with a corresponding deduction of 15 points. The deviation range corresponding to the third deviation range is greater than or equal to 20%, with a corresponding deduction of 30 points. When the deviation direction of the real-time measured value of each feature parameter relative to the benchmark value is positive or negative, the preset deviation range into which it falls is determined according to the absolute value of the deviation.

[0026] Preferably, the weighting coefficient ωi satisfies ∑ωi=1, and the weighting coefficient corresponding to the vibration characteristic value is the largest, followed by the weighting coefficient corresponding to the temperature characteristic value. The weighting coefficients corresponding to the current characteristic value and the pressure characteristic value are dynamically adjusted according to the actual operating characteristics of the primary air fan. Under the stable operating condition, the weighting coefficients of the vibration characteristic value, the temperature characteristic value, the current characteristic value, and the pressure characteristic value are 0.40, 0.30, 0.15, and 0.15, respectively. Under the variable load condition, the weighting coefficients of the current characteristic value and the pressure characteristic value increase.

[0027] Preferably, in the trend analysis step, linear regression analysis is performed on the health time series to calculate the slope of health change. Specifically, this includes: using the sampling time number t as the independent variable and the comprehensive health H(t) at the corresponding sampling time as the dependent variable, fitting a straight line H(t) = a·t + b using the least squares method, where a is the slope of health change; when a < 0 and the comprehensive health continues to decline within P consecutive sampling periods, the health change trend is determined to be a deteriorating trend.

[0028] Preferably, in the early warning determination step, the first health threshold is 85 points, and the second health threshold is 70 points; when the overall health score is lower than 85 points and the slope of the health score change is negative, the primary early warning signal is issued; when the overall health score is lower than 70 points, regardless of the slope of the health score change, the advanced early warning signal is issued; after the advanced early warning signal is issued, the sampling frequency of the multi-source monitoring data is automatically increased, and an early warning report containing the early warning type, early warning time, real-time values ​​of each characteristic parameter, and the overall health score is generated.

[0029] This invention has the following advantages: By automatically identifying operating conditions and dynamically determining benchmark values ​​for different operating conditions, this invention effectively avoids the problem of high false alarm rates under different operating conditions caused by fixed thresholds, making early warnings more accurate and reliable; through weighted comprehensive evaluation of multiple feature parameters, it comprehensively reflects the health status of the primary wind turbine from multiple dimensions, overcoming the defect of easy missed alarms in single-parameter judgment; through trend analysis of the slope of health change, it can capture early signs of performance degradation before the overall health level drops to the alarm threshold, achieving advanced early warning; through the graded early warning mechanism of primary and advanced early warning, it provides operators with differentiated response times, which helps to rationally arrange maintenance plans and avoid unplanned shutdowns. Attached Figure Description

[0030] To more intuitively illustrate the prior art and this application, exemplary drawings are provided below. It should be understood that the specific shapes and structures shown in the drawings should not generally be regarded as limiting conditions for implementing this application; for example, based on the technical concept disclosed in this application and the exemplary drawings, those skilled in the art are able to easily make conventional adjustments or further optimizations to the addition / reduction / classification, specific shapes, positional relationships, connection methods, size ratios, etc. of certain units (components).

[0031] Figure 1 A flowchart of a primary wind turbine intelligent early warning method based on wind turbine operating status assessment provided in this application embodiment. Detailed Implementation

[0032] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that these embodiments are merely for further explanation of the present invention and should not be construed as limiting the scope of protection of the present invention. Technical engineers in the field can make some non-essential improvements and adjustments to the present invention based on the above-described content. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] Please see Figure 1 A primary wind turbine intelligent early warning method based on wind turbine operating status assessment includes the following steps:

[0034] Data acquisition steps: Acquire multi-source monitoring data of the primary fan during operation at a preset sampling period. The multi-source monitoring data includes bearing vibration amplitude, bearing temperature, motor current, fan inlet pressure, and fan outlet pressure.

[0035] Data preprocessing steps: outlier removal and missing value imputation are performed on the multi-source monitoring data to obtain a preprocessed monitoring data sequence;

[0036] Operating condition identification steps: Based on the motor current and the fan inlet pressure, identify the current operating condition of the primary air fan, including startup condition, stable operation condition, and variable load condition;

[0037] Feature parameter extraction step: For the operating condition, extract feature parameters corresponding to the current operating condition from the preprocessed monitoring data sequence. The feature parameters include vibration feature values, temperature feature values, current feature values, and pressure feature values.

[0038] The benchmark value determination step involves determining the benchmark value and benchmark fluctuation range of each characteristic parameter under the operating condition based on historical monitoring data corresponding to the current operating condition of the primary air turbine under historical normal operating conditions.

[0039] Deviation calculation steps: Calculate the deviation between the real-time measured value of each characteristic parameter and the corresponding benchmark value, and determine the deviation level of each characteristic parameter according to the benchmark fluctuation range of each characteristic parameter;

[0040] Status assessment steps: Based on the deviation levels of each of the aforementioned characteristic parameters, calculate the overall health of the primary air fan according to the following formula:

[0041] H = 100 - ∑(ωi × Li)

[0042] Where H represents the overall health score, ωi represents the weight coefficient of the i-th feature parameter, Li represents the deduction value corresponding to the deviation level of the i-th feature parameter, and ∑ represents the sum of the weighted deduction values ​​of all feature parameters;

[0043] Trend analysis steps: Construct a health time series from the comprehensive health scores at the current sampling time and the previous N sampling times, perform linear regression analysis on the health time series, and calculate the slope of the health score change;

[0044] Warning determination steps: When the overall health level is lower than the first health level threshold and the slope of the health level change is negative, a primary warning signal is issued; when the overall health level is lower than the second health level threshold, a high-level warning signal is issued, wherein the second health level threshold is less than the first health level threshold.

[0045] Early warning output steps: Output the primary early warning signal or the advanced early warning signal to the monitoring terminal, and simultaneously record the multi-source monitoring data and the comprehensive health status at the time of the early warning.

[0046] The present invention provides a method for intelligent early warning of primary wind turbine based on wind turbine operation status assessment. This method collects multi-source monitoring data in real time through a sensor network deployed on the primary wind turbine body and its auxiliary equipment. The data is then collected and preprocessed by a data acquisition and preprocessing module, an operating condition identification and feature extraction module, a benchmark value management and deviation calculation module, a comprehensive health assessment module, a trend analysis module, and an early warning judgment and output module. This enables online assessment and intelligent early warning of the primary wind turbine operation status.

[0047] I. Data Collection

[0048] Vibration sensors are installed at the bearing housing of the primary air fan to collect bearing vibration amplitude signals, with a sampling frequency of no less than 1kHz. Thermocouples or resistance temperature sensors are installed at the bearing to collect bearing temperature signals. A current transformer is installed on the motor power supply line of the primary air fan to collect motor current signals. Pressure transmitters are installed on the inlet and outlet ducts of the primary air fan to collect inlet and outlet pressure signals, respectively. All sensors are connected to a data acquisition terminal, which collects monitoring data from each sensor at a preset sampling period (e.g., once every 10 seconds) to form multi-source monitoring data. The collected multi-source monitoring data includes monitoring data in five dimensions: bearing vibration amplitude, bearing temperature, motor current, fan inlet pressure, and fan outlet pressure.

[0049] II. Data Preprocessing

[0050] The data acquisition terminal transmits the collected multi-source monitoring data to the data processing server. The data processing server first performs outlier removal on the received multi-source monitoring data. Outlier removal is performed using the Raida criterion (i.e., the 3σ criterion): for a monitoring data sequence of a certain characteristic parameter, its arithmetic mean μ and standard deviation σ are calculated, and monitoring data exceeding the range of μ ± 3σ are identified as outliers and removed.

[0051] After outliers are removed, missing points remain in the data sequence. For these missing values, linear interpolation is used for imputation. This involves using monitoring data from adjacent time points before and after the missing point to calculate an estimated value for the missing time point based on a linear relationship and then filling it into the data sequence. After outlier removal and missing value imputation, a preprocessed monitoring data sequence is obtained. This data sequence is continuous and complete in time, providing a reliable data foundation for subsequent operational condition identification and feature extraction.

[0052] III. Operating Condition Identification

[0053] The operating condition identification module receives the pre-processed monitoring data sequence and identifies the current operating condition of the primary air blower based on the motor current and the blower inlet pressure. The specific identification logic is as follows:

[0054] First, determine if the real-time value of the motor current is less than the preset starting current threshold. The starting current threshold is preset based on the rated current and starting characteristics of the primary air fan, for example, it can be set to 1.2 times the rated current. When the real-time value of the motor current is less than the starting current threshold, it is determined that the primary air fan is currently in the starting condition.

[0055] When the real-time value of the motor current is greater than or equal to the starting current threshold, it is further determined whether the primary air fan is in a stable operating condition or a variable load condition. Specifically, the absolute value of the rate of change of the motor current over M consecutive sampling periods is calculated (M can be, for example, 6, corresponding to a 60-second time window). When the absolute value of this rate of change is less than a preset rate of change threshold, the primary air fan is determined to be in a stable operating condition; when the absolute value of this rate of change is greater than or equal to the preset rate of change threshold, the primary air fan is determined to be in a variable load condition. The rate of change threshold can be preset according to the normal operating fluctuation range of the primary air fan, for example, it can be set to 2% of the rated current per minute.

[0056] Through the above operating condition identification logic, the operation process of the primary air turbine can be dynamically divided into three types: start-up condition, stable operation condition, and variable load condition, providing a basis for subsequent feature extraction and benchmark value matching for each operating condition.

[0057] IV. Feature Parameter Extraction

[0058] After identifying the current operating condition, the feature extraction module extracts feature parameters corresponding to the current operating condition from the preprocessed monitoring data sequence. The specific extraction method is as follows:

[0059] For bearing vibration amplitude, its root mean square (RMS) value is calculated within a preset time window (e.g., 6 sampling points within the most recent 60 seconds) and used as the vibration characteristic value. The RMS value can effectively reflect the energy level of the vibration signal and is more representative than the instantaneous amplitude.

[0060] For bearing temperatures, their arithmetic mean is calculated within the same preset time window and used as the temperature characteristic value. The average value can filter out random fluctuations in the temperature signal and reflect the overall level of the bearing temperature.

[0061] For motor current, its arithmetic mean is calculated within a preset time window and used as the current characteristic value.

[0062] For the pressure characteristic, the difference between the fan outlet pressure and the fan inlet pressure (i.e., the fan total pressure or pressure rise) is calculated, and the arithmetic mean of this pressure difference is calculated within a preset time window as the pressure characteristic value. The pressure characteristic value directly reflects the output level and aerodynamic performance of the primary fan.

[0063] Through the above feature extraction, the original five-dimensional monitoring data is transformed into four feature parameters, namely vibration feature value, temperature feature value, current feature value and pressure feature value. Each feature parameter is characterized by a statistical value within a preset time window, which has strong stability and anti-interference ability.

[0064] V. Determination of Benchmark Values

[0065] The benchmark management module pre-stores historical monitoring data of the primary air turbine under normal operating conditions and corresponding to various operating conditions. This historical monitoring data is acquired as follows: during the first K days after the primary air turbine is put into operation (K can be, for example, 30 days), monitoring data is continuously collected and manually confirmed to be free of any faults and in good operating condition during this period. This data is then stored in the database as benchmark data for historical normal operating conditions. For each operating condition (start-up condition, stable operation condition, variable load condition), data segments under that condition are selected from the benchmark data, and the arithmetic mean and standard deviation of each characteristic parameter (vibration characteristic value, temperature characteristic value, current characteristic value, pressure characteristic value) are calculated.

[0066] When a primary wind turbine is currently operating under a certain condition, the benchmark value management module retrieves the arithmetic mean of each characteristic parameter corresponding to that operating condition from the database as the benchmark value, and retrieves the range of the corresponding average value ± 2 standard deviations as the benchmark fluctuation range. The benchmark fluctuation range represents the reasonable fluctuation range of that characteristic parameter under normal operating conditions.

[0067] VI. Deviation Calculation and Deviation Grade Determination

[0068] The deviation calculation module receives the real-time measured values ​​of each feature parameter (i.e., the feature parameter values ​​extracted at the current sampling time) and the corresponding baseline values, and calculates the deviation of each feature parameter. The deviation is calculated as follows:

[0069] When the real-time measured value of the characteristic parameter is greater than the reference value, the deviation is calculated as (real-time measured value - reference value) / reference value × 100%.

[0070] When the real-time measured value of the characteristic parameter is less than the reference value, the deviation is calculated as (reference value - real-time measured value) / reference value × 100%.

[0071] The deviation calculated in this way is always a non-negative percentage, representing the relative magnitude of the characteristic parameter's deviation from the benchmark value.

[0072] After calculating the deviation of each characteristic parameter, the corresponding deviation level and deduction value are determined based on the preset deviation range into which the deviation falls. In this embodiment, the preset deviation range includes three levels:

[0073] The first deviation interval corresponds to a deviation range of greater than or equal to 0 and less than 10%, with a corresponding deduction of 5 points.

[0074] The second deviation interval corresponds to a deviation range of greater than or equal to 10% and less than 20%, with a corresponding deduction of 15 points.

[0075] The third deviation interval corresponds to a deviation range of 20% or more, with a corresponding deduction of 30 points.

[0076] Regardless of whether the deviation of the real-time measured value of a feature parameter from the benchmark value is positive or negative (i.e., whether the measured value is higher or lower than the benchmark value), the preset deviation range into which it falls is determined by the absolute value of the deviation. For example, if the benchmark value of a feature parameter is 100 and the real-time measured value is 85, then the deviation is 15%, falling into the second deviation range, and the deduction value is 15 points.

[0077] VII. Comprehensive Health Assessment

[0078] The comprehensive health assessment module calculates the overall health of the primary air fan according to the following formula, based on the deduction values ​​corresponding to the deviation levels of each characteristic parameter:

[0079] H = 100 - ∑(ωi × Li)

[0080] Wherein, H represents the overall health score, ranging from 0 to 100, with higher values ​​indicating better health of the primary air blower; ωi represents the weight coefficient of the i-th feature parameter, and the sum of the weight coefficients of all feature parameters equals 1, i.e., ∑ωi=1; Li represents the deduction value corresponding to the deviation level of the i-th feature parameter, with values ​​of 5, 15, or 30; Σ represents the sum of the weighted deduction values ​​for all four feature parameters.

[0081] The weighting coefficients reflect the varying degrees of importance of different characteristic parameters to the health status of the primary air fan. Under stable operating conditions, the weighting coefficients for vibration characteristic values, temperature characteristic values, current characteristic values, and pressure characteristic values ​​are set to 0.40, 0.30, 0.15, and 0.15, respectively.

[0082] Among them, the vibration characteristic value has the largest weighting coefficient because bearing vibration is the most direct and sensitive indicator of the mechanical state of the rotating parts of the fan; the temperature characteristic value is the second largest, reflecting the bearing lubrication and cooling state; the current characteristic value and the pressure characteristic value reflect the electrical performance of the motor and the aerodynamic performance of the fan.

[0083] Under variable load conditions, since load changes have a significant impact on motor current and fan pressure, the weighting coefficients of current characteristic values ​​and pressure characteristic values ​​are appropriately increased (for example, adjusted to 0.20 respectively), while the weighting coefficients of vibration characteristic values ​​and temperature characteristic values ​​are slightly decreased (for example, adjusted to 0.35 and 0.25 respectively) to more accurately reflect the equipment status under variable load conditions.

[0084] For example, assuming that under stable operating conditions, in a certain evaluation, the deviation of the vibration characteristic value falls into the first deviation range (deduction of 5 points), the deviation of the temperature characteristic value falls into the second deviation range (deduction of 15 points), the deviation of the current characteristic value falls into the first deviation range (deduction of 5 points), and the deviation of the pressure characteristic value falls into the first deviation range (deduction of 5 points), then the overall health score H = 100 - (0.40×5 + 0.30×15 + 0.15×5 + 0.15×5) = 100 - (2 + 4.5 + 0.75 + 0.75) = 92 points.

[0085] VIII. Trend Analysis

[0086] The trend analysis module constructs a health time series by combining the health status of the current sampling time and the previous N sampling times. The value of N can be set according to actual needs. For example, N=30, which includes the comprehensive health status data of the most recent 30 sampling times (corresponding to the most recent 300 seconds), including the current time.

[0087] Linear regression analysis was performed on the health status time series to calculate the slope of health status change. Specifically, the sampling time number t was used as the independent variable (t=1,2,…,N+1), and the comprehensive health status H(t) at the corresponding sampling time was used as the dependent variable. The least squares method was used to fit the straight line:

[0088] H(t) = a·t + b

[0089] Where 'a' represents the slope of the health status change, and 'b' represents the intercept. When 'a' < 0, it indicates that the overall health status is decreasing over time; when 'a' > 0, it indicates that the overall health status is increasing over time; and when 'a' = 0, it indicates that the overall health status is basically stable.

[0090] Furthermore, when a < 0 and the overall health score continuously decreases over P consecutive sampling periods (P can be, for example, 5), the health score trend is determined to be a deteriorating trend. A deteriorating trend indicates that the health status of the primary wind turbine is continuously deteriorating, suggesting a potential risk of failure even if the current overall health score is acceptable.

[0091] IX. Early Warning Judgment and Output

[0092] The early warning determination module performs graded early warning determinations based on the overall health score and the slope of the health score change. In this embodiment, the first health score threshold is set to 85 points, and the second health score threshold is set to 70 points.

[0093] The trigger condition for a primary warning is: the overall health score is below 85 points and the slope of the health score change is negative. That is, when the overall health score of the primary wind turbine has dropped below 85 points and continues to decline, a primary warning signal is issued. The primary warning signal prompts operators to pay attention to the operating status of the primary wind turbine and recommends increasing the frequency of inspections and closely monitoring the changing trends of relevant parameters.

[0094] The trigger condition for a high-level warning is an overall health score below 70. When the overall health score is below 70, a high-level warning signal is issued regardless of whether the slope of the health score change is positive, negative, or zero. The high-level warning signal indicates that the primary air turbine is in a poor health condition and has a high risk of failure. It is recommended that operators take immediate intervention measures, such as reducing load or arranging a shutdown for maintenance.

[0095] After the warning signal is issued, the warning output module outputs the primary or advanced warning signal to the display interface of the monitoring terminal and alerts the operators with an audible and visual alarm. At the same time, the warning output module automatically records the multi-source monitoring data at the time of the warning (including bearing vibration amplitude, bearing temperature, motor current, fan inlet pressure, fan outlet pressure), values ​​of various characteristic parameters, overall health status, and warning type, forming a complete warning record.

[0096] After issuing an advanced warning signal, the system automatically increases the sampling frequency of multi-source monitoring data (e.g., from once every 10 seconds to once every 2 seconds) to more intensively track the status changes of primary wind turbines, providing more comprehensive data support for fault diagnosis and emergency response. Simultaneously, the system generates a warning report containing the warning type, warning time, real-time values ​​of various characteristic parameters, and overall health status, and pushes it to the mobile terminals or email addresses of relevant operation and maintenance management personnel.

[0097] By cyclically executing the above nine steps, this invention can perform continuous, online status assessment and intelligent early warning of primary wind turbines, issuing early warning signals in the early stages of performance degradation, thus providing sufficient response time for operation and maintenance and effectively reducing the risk of unplanned shutdowns.

[0098] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent early warning of primary wind turbines based on wind turbine operating status assessment, characterized in that, Includes the following steps: Data acquisition steps; Multi-source monitoring data of a primary fan during operation is acquired at a preset sampling period. The multi-source monitoring data includes bearing vibration amplitude, bearing temperature, motor current, fan inlet pressure, and fan outlet pressure. Data preprocessing steps: outlier removal and missing value imputation are performed on the multi-source monitoring data to obtain a preprocessed monitoring data sequence; Operating condition identification steps: Based on the motor current and the fan inlet pressure, identify the current operating condition of the primary air fan, including startup condition, stable operation condition, and variable load condition; Feature parameter extraction steps; For the aforementioned operating conditions, feature parameters corresponding to the current operating conditions are extracted from the preprocessed monitoring data sequence. These feature parameters include vibration feature values, temperature feature values, current feature values, and pressure feature values. Steps for determining the benchmark value; Based on historical monitoring data of the primary air turbine under historical normal operating conditions and corresponding to the current operating conditions, the baseline values ​​and baseline fluctuation ranges of each of the aforementioned characteristic parameters under the operating conditions are determined. Deviation calculation steps: Calculate the deviation between the real-time measured value of each characteristic parameter and the corresponding benchmark value, and determine the deviation level of each characteristic parameter according to the benchmark fluctuation range of each characteristic parameter; Status assessment steps: Based on the deviation levels of each of the aforementioned characteristic parameters, calculate the overall health of the primary air fan according to the following formula: H = 100 - ∑(ωi × Li) Where H represents the overall health score, ωi represents the weight coefficient of the i-th feature parameter, Li represents the deduction value corresponding to the deviation level of the i-th feature parameter, and ∑ represents the sum of the weighted deduction values ​​of all feature parameters; Trend analysis steps: Construct a health time series from the comprehensive health scores at the current sampling time and the previous N sampling times, perform linear regression analysis on the health time series, and calculate the slope of the health score change; Warning determination steps: When the overall health level is lower than the first health level threshold and the slope of the health level change is negative, a primary warning signal is issued; when the overall health level is lower than the second health level threshold, a high-level warning signal is issued, wherein the second health level threshold is less than the first health level threshold. Early warning output steps: Output the primary early warning signal or the advanced early warning signal to the monitoring terminal, and simultaneously record the multi-source monitoring data and the comprehensive health status at the time of the early warning.

2. The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment according to claim 1, characterized in that, In the data preprocessing step, the outlier removal is performed using the Raida criterion, which identifies and removes monitoring data that exceed the mean ± 3 standard deviations. The missing value imputation is performed using linear interpolation, which uses monitoring data from adjacent times before and after the missing time to imput the missing value.

3. The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment according to claim 2, characterized in that, In the operating condition identification step, the current operating condition of the primary air fan is identified based on the motor current and the fan inlet pressure. Specifically, this includes: when the real-time value of the motor current is less than the starting current threshold, it is determined to be the starting condition; when the real-time value of the motor current is greater than or equal to the starting current threshold and the absolute value of the rate of change of the motor current over M consecutive sampling periods is less than the rate of change threshold, it is determined to be the stable operating condition; when the real-time value of the motor current is greater than or equal to the starting current threshold and the absolute value of the rate of change of the motor current over M consecutive sampling periods is greater than or equal to the rate of change threshold, it is determined to be the variable load condition.

4. The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment according to claim 3, characterized in that, In the feature parameter extraction step, the vibration feature value is the root mean square value of the bearing vibration amplitude within a preset time window, the temperature feature value is the average value of the bearing temperature within the preset time window, the current feature value is the average value of the motor current within the preset time window, and the pressure feature value is the average value of the difference between the fan outlet pressure and the fan inlet pressure within the preset time window.

5. The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment according to claim 4, characterized in that, In the benchmark value determination step, the historical monitoring data under normal operating conditions is the monitoring data of the primary wind turbine during the first K days after it is put into operation, which is confirmed by manual verification to be fault-free and under the operating conditions; the benchmark value is the arithmetic mean of the corresponding characteristic parameters in the historical monitoring data, and the benchmark fluctuation range is the average value of the corresponding characteristic parameters in the historical monitoring data ± 2 times the standard deviation.

6. The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment according to claim 5, characterized in that, In the deviation calculation step, the deviation is calculated as follows: when the real-time measured value of the feature parameter is greater than the benchmark value, the deviation is equal to (real-time measured value - benchmark value) / benchmark value × 100%; when the real-time measured value of the feature parameter is less than the benchmark value, the deviation is equal to (benchmark value - real-time measured value) / benchmark value × 100%; the deviation level is determined according to the preset deviation range into which the deviation falls, and different preset deviation ranges correspond to different deduction values, and the deduction value increases as the deviation increases.

7. The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment according to claim 6, characterized in that, The preset deviation range includes a first deviation range, a second deviation range, and a third deviation range. The deviation range corresponding to the first deviation range is greater than or equal to 0 and less than 10%, with a corresponding deduction of 5 points. The deviation range corresponding to the second deviation range is greater than or equal to 10% and less than 20%, with a corresponding deduction of 15 points. The deviation range corresponding to the third deviation range is greater than or equal to 20%, with a corresponding deduction of 30 points. When the real-time measured value of each feature parameter deviates from the benchmark value in either a positive or negative direction, the preset deviation range into which it falls is determined according to the absolute value of the deviation.

8. The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment according to claim 7, characterized in that, The weighting coefficient ωi satisfies ∑ωi=1, and the weighting coefficient corresponding to the vibration characteristic value is the largest, followed by the weighting coefficient corresponding to the temperature characteristic value. The weighting coefficients corresponding to the current characteristic value and the pressure characteristic value are dynamically adjusted according to the actual operating characteristics of the primary air blower. Under the stable operating conditions, the weighting coefficients for the vibration characteristic value, the temperature characteristic value, the current characteristic value, and the pressure characteristic value are 0.40, 0.30, 0.15, and 0.15, respectively. Under the variable load condition, the weighting coefficients of the current characteristic value and the pressure characteristic value are increased.

9. The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment according to claim 8, characterized in that, In the trend analysis step, linear regression analysis is performed on the health time series to calculate the slope of health change. Specifically, this includes: using the sampling time number t as the independent variable and the comprehensive health H(t) at the corresponding sampling time as the dependent variable, the least squares method is used to fit a straight line H(t) = a·t + b, where a is the slope of health change; when a < 0 and the comprehensive health continues to decline within P consecutive sampling periods, the health change trend is determined to be a deteriorating trend.

10. The intelligent early warning method for primary wind turbines based on wind turbine operating status assessment according to claim 9, characterized in that, In the aforementioned warning determination step, the first health threshold is 85 points, and the second health threshold is 70 points. When the overall health score is below 85 points and the slope of the health score change is negative, the primary warning signal is issued. When the overall health score is below 70, regardless of the slope of the health score change, the advanced warning signal is issued. After the advanced warning signal is issued, the sampling frequency of the multi-source monitoring data is automatically increased, and a warning report containing the warning type, warning time, real-time values ​​of each characteristic parameter, and the overall health score is generated.

Images