A Method and System for Monitoring the Health Status of Wind Turbine Generators Based on Sensor Networks

Abstract

This invention provides a method and system for monitoring the health status of wind turbine generators based on sensor networks, belonging to the field of sensor network technology. By acquiring the mechanical structure and energy transfer path of the generator set, a hierarchical index is constructed for the blade layer, transmission chain layer, power generation layer, and support layer. Vibration and temperature data of each layer are collected, and steady-state and dynamic disturbance conditions are identified based on wind speed and rotational speed change rates. Vibration response intensity, temperature change, and vibration-temperature coupling characteristics are extracted, and a condition correction factor is constructed. A feature matrix is ​​constructed using a sliding time window, and a benchmark matrix is ​​calculated based on steady-state data. Local health deviation is obtained through norm differences, and weighted fusion is performed to obtain a comprehensive health index for the entire generator. The comprehensive index is dynamically corrected according to real-time operating conditions, and an early warning is issued after comparison with a threshold. This solves the problems of health assessment being easily affected by disturbances and insufficient utilization of multi-physics coupling under varying operating conditions, achieving adaptive and highly robust wind turbine generator health status monitoring.

Description

Technical Field

[0001] This invention relates to the field of sensor network technology, and more specifically, to a method and system for monitoring the health status of wind turbine generator sets based on sensor networks. Background Technology

[0002] As the global energy structure transitions towards clean and low-carbon energy, wind power, as one of the most technologically mature and commercially developed renewable energy sources, continues to experience rapid and sustained growth in installed capacity. Wind farms are gradually developing towards deep-sea operations, large single-unit capacities, and intelligent operation and maintenance, which places stringent demands on the operational reliability and life-cycle economics of wind turbine generators. As a typical complex system involving the coupling of multiple physical fields (mechanical, electrical, and control systems), wind turbine generators are subjected to random wind loads, alternating stresses, and harsh environmental erosion over long periods. Their key components (such as blades, main shafts, gearboxes, and generators) are highly susceptible to fatigue damage, wear, loosening, and other performance degradation failures. Traditional monitoring strategies based on periodic maintenance or single-threshold over-limit alarms suffer from problems such as excessive maintenance or both missed and false alarms, making it difficult to achieve proactive early warning of minor faults and accurate assessment of health status. To address this, intelligent condition monitoring technology based on sensor networks has emerged. By deploying multi-parameter sensors such as vibration, temperature, and strain sensors, combined with signal processing and data-driven models, continuous perception and health diagnosis of the unit's operating status can be achieved, making it a research hotspot and core technology in the field of wind power operation and maintenance.

[0003] However, most existing solutions still suffer from the following significant shortcomings: First, existing methods typically treat single physical quantities such as vibration or temperature in isolation, failing to effectively explore the physical coupling mechanism between vibration and temperature in the fault evolution process, resulting in insufficient ability to identify complex faults or early latent faults. Second, the operating conditions of wind turbine generators are highly time-varying and non-steady-state, and vibration and temperature responses can experience non-fault-related, drastic fluctuations due to operating condition disturbances. Most current health assessment models lack the ability to adaptively distinguish between dynamic disturbance scenarios and steady-state operating scenarios, and have not constructed scenario-dependent correction factors, causing health indicators to frequently generate false alarms due to changes in operating conditions, or to mask the true fault trend. Third, existing technologies mostly use feature comparison at a single time point, ignoring the evolution trajectory and matrix structure information of features within continuous time windows, making it difficult to characterize the gradual process of the unit from health to deterioration. Fourth, research on hierarchical and systematic health modeling of the complete energy transfer chain of wind turbine blades-drive chain-power generation-support is relatively weak, and there is a lack of a hierarchical index structure for the energy flow path of the entire unit and a method for constructing a comprehensive health index by weighted integration of the contribution of each level. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for monitoring the health status of wind turbine generator sets based on sensor networks, in order to solve the technical problem of how to integrate the coupling characteristics of multiple physical quantities such as vibration and temperature under highly time-varying and multi-disturbance working conditions, and to construct a dynamic health assessment model that is adaptive to the working conditions, so as to achieve accurate and robust health status monitoring of each level of the entire energy transfer chain of the generator set.

[0005] In view of the above-mentioned technical problems, the present invention provides a method and system for monitoring the health status of wind turbine generator sets based on sensor networks.

[0006] In a first aspect, the present invention provides a method for monitoring the health status of wind turbine generator sets based on sensor networks. The method includes: Step S1: acquiring the mechanical structure and energy transfer path of the wind turbine generator set, and constructing a hierarchical index set; deploying sensor modules and constructing historical data acquisition time periods; Step S2: constructing normalized data sets for the blade layer, transmission chain layer, generator layer, and support layer at the historical data acquisition time points; acquiring wind speed and rotational speed data, and analyzing the operating conditions of the wind turbine generator set at the historical data acquisition time points; Step S3: analyzing the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of each layer at the historical data acquisition time points; and constructing a system based on operating condition factors. Step S4: Construct the feature matrix of the level under the sliding time window; calculate the benchmark feature matrix of the level; construct the feature matrix of the level under the current sliding time window; based on the feature matrix and the benchmark feature matrix, calculate the local health deviation of the level and calculate the overall health index of the machine; Step S5: Obtain the working condition scenario factor and working condition correction factor under the real-time data acquisition time point; dynamically correct the overall health index of the machine; preset the health warning index threshold, analyze and issue a health warning.

[0007] As a preferred embodiment of the sensor network-based wind turbine health status monitoring method of the present invention, the mechanical structure and energy transfer path of the wind turbine are obtained, and the wind turbine is divided into a blade layer, a transmission chain layer, a power generation layer, and a support layer based on the mechanical structure and energy transfer path, constructing a hierarchical index set, denoted as . ,in, , , and These respectively represent the blade layer, transmission chain layer, power generation layer, and support layer of a wind turbine generator set;

[0008] In the blade layer Transmission chain layer Power generation layer and support layer An internal sensor module is installed, including a vibration sensor and a temperature sensor; historical data acquisition time periods are established, denoted as... ,in, Let t represent the t-th historical data collection time point, and T represent the total number of historical data collection time points.

[0009] As a preferred embodiment of the sensor network-based wind turbine health status monitoring method described in this invention, historical data collection time points are constructed. Lower blade layer Transmission chain layer Power generation layer and support layer The normalized dataset (normalized using the min-max method) is denoted as follows: , , and ,in, and These represent the historical data collection time points. Leaf layer collected from the bottom Vibration data and temperature data, and These represent the historical data collection time points. Lower-level transmission chain layer Vibration data and temperature data, and These represent the historical data collection time points. Lower collection of power generation layer Vibration data and temperature data, and These represent the historical data collection time points. Support layer for lower collection Vibration data and temperature data.

[0010] As a preferred embodiment of the sensor network-based wind turbine health status monitoring method of the present invention, historical data collection time points are collected. The operating parameter data of the wind turbine generator set, including wind speed data and rotational speed data, are normalized and denoted as follows: and Based on historical data collection time points Wind speed data for wind turbine generators and speed data Analyze historical data collection time points The specific operating scenarios for wind turbine generators are as follows:

[0011] ;

[0012] in, Indicates the time point of historical data collection Operating scenario factors for wind turbine generators and These represent the historical data collection time points. Wind speed and rotational speed data for the wind turbine generator set. This represents a preset constant, typically with a value of 10. -6 Used to avoid division by zero. This represents a threshold value preset by experts based on historical experience. When t=1, it is defined as... ;

[0013] like This indicates the historical data collection time point. The operating scenario for the wind turbine generator set is a dynamic disturbance scenario; if This indicates the historical data collection time point. The operating scenario for the wind turbine generator set is a steady-state operation scenario.

[0014] As a preferred embodiment of the sensor network-based wind turbine health status monitoring method of the present invention, based on historical data acquisition time points... Lower blade layer Transmission chain layer Power generation layer and support layer Vibration and temperature data, and analysis of historical data acquisition time points. The vibration response intensity characteristics and temperature change characteristics of each level are as follows:

[0015] ;

[0016] in, Indicates the time point of historical data collection Vibration response intensity characteristics at the m-th level. Indicates the time point of historical data collection Temperature change characteristics at the m-th level and These represent the historical data collection time points. Vibration and temperature data at the m-th level. and These represent the historical data collection time points. Vibration and temperature data at the m-th level. This represents a preset constant, typically with a value of 10. -6Used to avoid division by zero. When t=1, define , ;

[0017] Calculate historical data collection time points The vibration-temperature coupling characteristics of the m-th level are calculated using the following formula:

[0018] ;

[0019] in, Indicates the time point of historical data collection The vibration-temperature coupling characteristics of the m-th level (positive values ​​indicate that vibration and temperature change in the same direction (which may indicate friction, increased load, etc.), and negative values ​​indicate that they change in opposite directions (which may indicate abnormal cooling, loosening, etc.)). This represents a preset constant;

[0020] Based on historical data collection time points Operating scenario factors of wind turbine generator sets Construct the operating condition correction factor, denoted as ,in, , and These represent the preset modulation coefficients (used to adjust the weight of the impact of dynamic disturbances on health indicators).

[0021] As a preferred embodiment of the sensor network-based wind turbine health status monitoring method of the present invention, based on historical data acquisition time points... Vibration response intensity characteristics of the m-th level Temperature change characteristics and vibration-temperature coupling characteristics The vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the m-th level at all historical data acquisition time points are obtained, and the original feature sequence matrix is ​​constructed, denoted as... ,in, ;

[0022] Set the sliding time window length, denoted as L (if L=10, it means the sliding time window length includes 10 consecutive historical data collection time points), and denote the starting time point index of the k-th sliding time window as... Let the index of the end time of the k-th sliding time window be denoted as . ,in, , (To cover all time points without omission, the step size of adjacent windows is usually taken as 1, that is, the sliding step size is 1 time point).

[0023] As a preferred embodiment of the sensor network-based wind turbine generator health status monitoring method of the present invention, a feature matrix of the m-th level under the k-th sliding time window is constructed based on the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the m-th level under the k-th sliding time window, denoted as... ,in, ;

[0024] We obtain all historical data acquisition time points for the wind turbine generator set under steady-state operation, and acquire the feature matrix corresponding to all sliding time windows (all steady-state scenarios) constructed with each historical data acquisition time point as the starting point. We then calculate the baseline feature matrix for the m-th level using the following formula:

[0025] ;

[0026] in, This represents the baseline feature matrix of the m-th level. When the operating scenario of a wind turbine generator is considered to be a steady-state operation scenario, it represents the set of all sliding time windows constructed with each historical data acquisition time point as the starting point. This represents the total number of sliding time windows when the operating scenario of the wind turbine generator is a steady-state operation scenario.

[0027] During real-time monitoring, the current real-time data acquisition time point is used as the end point of the window. The current sliding time window is constructed by taking the previous L-1 consecutive historical data acquisition time points, and the feature matrix of the m-th level under the current sliding time window is constructed. ;

[0028] Based on the feature matrix of the m-th level under the current sliding time window Using the baseline feature matrix of the m-th level, calculate the local health deviation of the m-th level using the following formula:

[0029] ;

[0030] in, This represents the local health deviation at the m-th level. Denotes the Frobenius norm. This represents a preset constant (a very small constant to prevent division by zero errors).

[0031] The local health deviations at all levels are weighted and fused to calculate the overall health index of the machine, denoted as . ,in, , This represents the preset influence factor for the m-th level (determined using the analytic hierarchy process or based on historical fault data statistics; default recommended value:). =0.35 (blade layer) =0.35 (transmission chain layer) =0.20 (power generation layer) =0.10 (support layer)), and, .

[0032] As a preferred embodiment of the sensor network-based wind turbine health status monitoring method of the present invention, based on historical data acquisition time points... Operating scenario factors of wind turbine generator sets and operating condition correction factor During real-time monitoring, wind speed and rotational speed data are acquired at the real-time data acquisition time point, and the operating condition scenario factor and operating condition correction factor of the wind turbine generator at the real-time data acquisition time point are recorded as follows: and ;

[0033] Overall health indicators of the machine Dynamic corrections will be made, as follows:

[0034] like This indicates that the operating scenario of the wind turbine generator at the real-time data acquisition time point is a steady-state operating scenario, let ,in, This indicates the overall health index of the entire machine after dynamic correction;

[0035] like If , it indicates that the operating scenario of the wind turbine generator at the data acquisition time point is a dynamic disturbance scenario, then let ,in, This indicates the overall health index of the entire machine after dynamic correction;

[0036] Preset health warning indicator thresholds (the health warning indicator thresholds are based on the overall health indicators of the whole machine under historical steady-state data) Distribution determination: Take the overall health index of the whole machine during the historical steady-state operation period. (95th percentile), if the overall health index of the machine is dynamically corrected If the health warning indicator threshold is greater than or equal to the threshold, a health warning will be issued to the relevant staff.

[0037] Secondly, the present invention also provides a health status monitoring system for wind turbine generator sets based on sensor networks. The system includes: an aggregate construction module, an operating condition scenario construction module, a feature calculation and matrix construction module, a comprehensive index calculation module, and an index correction and early warning module.

[0038] The collection construction module: acquires the mechanical structure and energy transfer path of the wind turbine generator set, constructs a hierarchical index set; deploys sensor modules and constructs historical data acquisition time periods;

[0039] The operating condition scenario construction module: constructs a normalized data set of the blade layer, transmission chain layer, power generation layer and support layer at the historical data collection time point; collects wind speed data and rotational speed data, and analyzes the operating condition scenario of the wind turbine generator set at the historical data collection time point.

[0040] The feature calculation and matrix construction module analyzes the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of each level at historical data acquisition time points; constructs a working condition correction factor based on the working condition scenario factor; obtains the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of all levels at all historical data acquisition time points, and constructs the original feature sequence matrix; and sets the sliding time window length.

[0041] The comprehensive index calculation module: constructs the feature matrix of the level under the sliding time window; calculates the benchmark feature matrix of the level; constructs the feature matrix of the level under the current sliding time window; and calculates the local health deviation of the level based on the feature matrix and the benchmark feature matrix, and calculates the overall health index of the whole machine.

[0042] The indicator correction and early warning module: acquires the operating condition scenario factors and operating condition correction factors at the real-time data acquisition time point; dynamically corrects the overall health indicators of the machine; presets health early warning indicator thresholds, analyzes and issues health early warnings.

[0043] Furthermore, the feature calculation and matrix construction module includes a feature calculation unit and a matrix construction unit;

[0044] The feature calculation unit analyzes the vibration response intensity characteristics and temperature change characteristics of each layer at the historical data acquisition time point based on the vibration and temperature data of the blade layer, transmission chain layer, power generation layer, and support layer at the historical data acquisition time point; calculates the vibration-temperature coupling characteristics of each layer at the historical data acquisition time point; and constructs an operating condition correction factor based on the operating condition scenario factors of the wind turbine generator at the historical data acquisition time point.

[0045] The matrix construction unit: based on the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the lower level at all historical data acquisition time points, obtains the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the lower level at all historical data acquisition time points, and constructs the original feature sequence matrix; and sets the sliding time window length.

[0046] The present invention provides one or more technical solutions, which have at least the following technical effects or advantages: A hierarchical index set is constructed based on the energy transfer path, overcoming the defects of black-box monitoring of the entire machine and providing a structural foundation for accurate fault location and differentiated weight fusion; by introducing operating condition scenario factors to distinguish steady-state and dynamic disturbance scenarios, the problem of false alarms and missed alarms caused by wind condition changes is solved from the data source; based on the extraction of vibration and temperature change features, vibration-temperature coupling features and operating condition correction factors are creatively constructed, which not only deeply mines the fault correlation information between multiple physical fields, but also realizes the quantifiable modeling of non-stationary disturbances; by using a sliding time window to upgrade the one-dimensional feature sequence into a feature matrix, and constructing a benchmark matrix based on pure steady-state data, the local health deviation is calculated through norm differences and weighted and fused into a comprehensive indicator of the entire machine, significantly improving the smoothness, robustness, and system-level characterization capability of the evaluation results; step S5 dynamically corrects the comprehensive health indicator according to the real-time operating condition scenario factors, maintaining a sensitive response under steady state and actively reducing weight under disturbance, ultimately achieving adaptive early warning decision-making. This closed-loop process, progressing step by step from physical layering, scene perception, deep coupling, matrix-based baseline learning to dynamic correction, comprehensively solves the core problems of existing technologies such as inaccurate evaluation under varying operating conditions, insufficient utilization of multi-physical quantity correlation, and lack of system-level health representation.

[0047] The above description is merely an overview of the technical solution of the present invention. To better understand the technical means of the present invention and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of the present invention more apparent, specific embodiments of the present invention are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily apparent from the following description. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0049] Figure 1 This is a schematic diagram of the steps of the wind turbine generator health status monitoring method based on sensor networks of the present invention;

[0050] Figure 2 This is a schematic diagram of the wind turbine generator health status monitoring system based on sensor networks according to the present invention. Detailed Implementation

[0051] This invention provides a sensor network-based method and system for monitoring the health status of wind turbine generators. It addresses the technical challenges of existing technologies, such as difficulty in distinguishing between dynamic disturbances and actual faults under varying operating conditions, insufficient utilization of the multi-physics coupling characteristics of vibration and temperature, lack of hierarchical health modeling based on the energy transfer chain, and susceptibility to transient noise interference leading to both false alarms and missed alarms. By constructing a hierarchical index set of blade, transmission chain, power generation, and support layers, and introducing operating condition scenario factors to distinguish between steady-state and dynamic disturbance scenarios, and establishing operating condition correction factors, the invention extracts vibration response intensity, temperature changes, and vibration-temperature coupling characteristics. A feature matrix and a baseline matrix based on pure steady-state data are constructed using a sliding time window. Local health deviations are calculated and weighted to form a comprehensive health index for the entire generator. This comprehensive index is then dynamically corrected based on real-time operating conditions and compared with a warning threshold. This achieves adaptive and robust health status monitoring under non-steady wind conditions, significantly improving early fault identification and warning accuracy, and effectively reducing false alarm and missed alarm rates.

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. It should be understood that the present invention is not limited to the exemplary embodiments described herein. 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. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, and not all of them.

[0053] Example 1, please refer to Figure 1 A method for monitoring the health status of wind turbine generators based on sensor networks is provided, the method comprising:

[0054] Step S1: Obtain the mechanical structure and energy transfer path of the wind turbine generator set, and construct a hierarchical index set; deploy sensor modules and construct historical data acquisition time periods.

[0055] Specifically, the mechanical structure and energy transfer path of the wind turbine generator are obtained, and based on the mechanical structure and energy transfer path, the wind turbine generator is divided into a blade layer, a transmission chain layer, a power generation layer, and a support layer. A hierarchical index set is constructed, denoted as . ,in, , , and These respectively represent the blade layer, transmission chain layer, power generation layer, and support layer of a wind turbine generator set;

[0056] In the blade layer Transmission chain layer Power generation layer and support layer An internal sensor module is installed, including a vibration sensor and a temperature sensor; historical data acquisition time periods are established, denoted as... ,in, Let t represent the t-th historical data collection time point, and T represent the total number of historical data collection time points.

[0057] Step S2: Construct a normalized dataset of the blade layer, transmission chain layer, power generation layer, and support layer at the historical data collection time points; collect wind speed data and rotational speed data, and analyze the operating conditions of the wind turbine generator set at the historical data collection time points.

[0058] Specifically, construct historical data collection time points. Lower blade layer Transmission chain layer Power generation layer and support layer The normalized dataset (normalized using the min-max method) is denoted as follows: , , and ,in, and These represent the historical data collection time points. Leaf layer collected from the bottom Vibration data and temperature data, and These represent the historical data collection time points. Lower-level transmission chain layer Vibration data and temperature data, and These represent the historical data collection time points. Lower collection of power generation layer Vibration data and temperature data, and These represent the historical data collection time points. Support layer for lower collection Vibration data and temperature data.

[0059] Furthermore, collect historical data at specific time points. The operating parameter data of the wind turbine generator set, including wind speed data and rotational speed data, are normalized and denoted as follows: and Based on historical data collection time points Wind speed data for wind turbine generators and speed data Analyze historical data collection time points The specific operating scenarios for wind turbine generators are as follows:

[0060] ;

[0061] in, Indicates the time point of historical data collection Operating scenario factors for wind turbine generators and These represent the historical data collection time points. Wind speed and rotational speed data for the wind turbine generator set. This represents a preset constant, typically with a value of 10. -6 Used to avoid division by zero. This represents a threshold value preset by experts based on historical experience. When t=1, it is defined as... ;

[0062] like This indicates the historical data collection time point. The operating scenario for the wind turbine generator set is a dynamic disturbance scenario; if This indicates the historical data collection time point. The operating scenario for the wind turbine generator set is a steady-state operation scenario.

[0063] It should be noted that the operational stability of wind turbines is directly determined by the fluctuations in input energy (wind speed) and turbine response (rotation speed). The smaller the relative rate of change between the two, the closer the turbine operation is to steady state. The larger the fluctuation, the more dynamic disturbance scenarios such as gusts, start-up, shutdown, and pitch control are represented. The calculation formula for the operating condition scenario factor uses a relative rate of change rather than an absolute change, eliminating the influence of differences in the absolute values ​​of wind speed / rotation speed under different wind conditions and turbine capacities. This achieves normalized judgment under all operating conditions and addresses the core pain point of health assessment under varying operating conditions being easily affected by disturbances from the data source. Traditional solutions do not distinguish between operating conditions and misjudge non-fault vibration / temperature fluctuations caused by gusts as health degradation, resulting in a large number of false alarms. This formula achieves accurate division between steady-state and dynamic disturbance scenarios, providing a core basis for subsequent full-process processing and defining a clean data range for the construction of the health baseline matrix. That is, this invention only uses steady-state scenario data to construct the health baseline, completely eliminating the contamination of the baseline by disturbance data and ensuring the accuracy of the health baseline.

[0064] Step S3: Analyze the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of each level at the historical data acquisition time points; construct the working condition correction factor based on the working condition scenario factor; obtain the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of all levels at the historical data acquisition time points, and construct the original feature sequence matrix; set the sliding time window length.

[0065] Specifically, based on historical data collection time points Lower blade layer Transmission chain layer Power generation layer and support layer Vibration and temperature data, and analysis of historical data acquisition time points. The vibration response intensity characteristics and temperature change characteristics of each level are as follows:

[0066] ;

[0067] in, Indicates the time point of historical data collection Vibration response intensity characteristics at the m-th level. Indicates the time point of historical data collection Temperature change characteristics at the m-th level and These represent the historical data collection time points. Vibration and temperature data at the m-th level. and These represent the historical data collection time points. Vibration and temperature data at the m-th level. This represents a preset constant, typically with a value of 10. -6 Used to avoid division by zero. When t=1, define , ;

[0068] It should be noted that the formula for calculating the vibration response intensity characteristics uses the absolute value of the relative rate of change of vibration data, with the core function of capturing the instantaneous fluctuation amplitude of the vibration signal. When wind turbine components experience wear, cracks, loosening, or other faults, the most direct manifestation is an abnormal sudden change in vibration amplitude; the relative rate of change design eliminates the differences in vibration reference values ​​under different levels and operating conditions, achieving cross-level characteristic normalization.

[0069] The temperature change characteristic calculation formula uses a signed relative temperature change rate to fully preserve the trend information of temperature rise / fall. Component failures (such as gearbox wear or generator winding short circuits) are accompanied by abnormal temperature rises, while cooling system failures cause abnormal temperature drops. The signed calculation can distinguish between different failure modes.

[0070] Features were extracted from the four levels of the energy transfer path, which aligns with the core design of the hierarchical modeling of this invention. This provides a foundation for subsequent fault location and weighted fusion, and completes the health feature extraction of the original sensor data. The high-dimensional, noisy vibration and temperature time series data are transformed into quantifiable and comparable dimensionless health features, solving the problem that the original data cannot be directly used for health assessment.

[0071] Calculate historical data collection time points The vibration-temperature coupling characteristics of the m-th level are calculated using the following formula:

[0072] ;

[0073] in, Indicates the time point of historical data collection The vibration-temperature coupling characteristics of the m-th level (positive values ​​indicate that vibration and temperature change in the same direction (which may indicate friction, increased load, etc.), and negative values ​​indicate that they change in opposite directions (which may indicate abnormal cooling, loosening, etc.)). This represents a preset constant;

[0074] It should be noted that in the evolution of wind turbine component failures, vibration and temperature are strongly physically related. Specifically, bearing wear will simultaneously lead to increased vibration (frictional impact) and increased temperature (frictional heat generation), with the two changing in the same direction. On the other hand, loose components will lead to increased vibration, reduced frictional heat generation, and decreased temperature, with the two changing in opposite directions.

[0075] The numerator of the vibration-temperature coupling characteristic calculation formula is the product of the vibration and temperature changes. The sign directly reflects the consistency of the two changes (positive values ​​in the same direction, negative values ​​in opposite directions), and the absolute value reflects the degree of coordinated change. The denominator is the sum of the absolute values ​​of the two changes, which normalizes the coupling characteristics to the range of [-1, 1]. Regardless of the magnitude of the absolute values ​​of the vibration and temperature changes, only the coordinated change relationship between the two is considered. That is, if the vibration and temperature fluctuate synchronously due to gusts and the degree of coordination is normal, the coupling characteristics will be normal. Even if the abnormal coordination caused by a fault is extremely small, it will be accurately captured, thus completely avoiding false alarms of coupling characteristics caused by fluctuations in operating conditions.

[0076] It solves the problem of isolated processing of single physical quantities in the background technology, which is insufficient in the ability to identify complex faults and early latent faults. Early weak faults often do not cause single parameter to exceed the limit, but will first manifest as abnormal vibration-temperature coupling relationship. This feature can capture such early signals and greatly improve the advance warning of faults.

[0077] The essence of wind turbine failure is the evolutionary process of multi-physics coupling between the machine and the heat. Latent failures such as early bearing wear and microcracks in the blades only produce extremely small vibrations and temperature changes. Individually, any parameter may appear to be within the normal range, making them completely undetectable by traditional methods. This formula directly quantifies the synergy between vibration and temperature changes. Even if a single parameter is within its limits, it can accurately detect any abnormalities in the coupling relationship between the two. This is a core advantage that traditional methods of analyzing vibration and temperature in isolation cannot achieve.

[0078] Based on historical data collection time points Operating scenario factors of wind turbine generator sets Construct the operating condition correction factor, denoted as ,in, , and These represent the preset modulation coefficients (used to adjust the weight of the impact of dynamic disturbances on health indicators).

[0079] It should be noted that the core logic of the operating condition correction factor calculation formula is that the greater the fluctuation of the operating condition, the stronger the interference on the health indicators. The larger the value of the correction factor, the greater the subsequent reduction in weight of the health indicators. This realizes the adaptive quantitative correction of health indicators under dynamic disturbance scenarios. In the traditional solution, under gust scenarios, non-fault fluctuations will cause abnormal increases in health indicators and trigger false alarms. The operating condition correction factor calculation formula constructs a correction factor that is positively correlated with the intensity of the disturbance, which can inversely offset the fluctuations of non-fault indicators caused by operating condition disturbances.

[0080] Furthermore, based on historical data collection time points Vibration response intensity characteristics of the m-th level Temperature change characteristics and vibration-temperature coupling characteristics The vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the m-th level at all historical data acquisition time points are obtained, and the original feature sequence matrix is ​​constructed, denoted as... ,in, ;

[0081] Set the sliding time window length, denoted as L (if L=10, it means the sliding time window length includes 10 consecutive historical data collection time points), and denote the starting time point index of the k-th sliding time window as... Let the index of the end time of the k-th sliding time window be denoted as . ,in, , (To cover all time points without omission, the step size of adjacent windows is usually taken as 1, that is, the sliding step size is 1 time point).

[0082] Step S4: Construct the feature matrix of the level under the sliding time window; calculate the baseline feature matrix of the level; construct the feature matrix of the level under the current sliding time window; based on the feature matrix and the baseline feature matrix, calculate the local health deviation of the level and calculate the overall health index of the whole machine.

[0083] Specifically, based on the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the m-th level under the k-th sliding time window, a feature matrix of the m-th level under the k-th sliding time window is constructed, denoted as... ,in, ;

[0084] We obtain all historical data acquisition time points for the wind turbine generator set under steady-state operation, and acquire the feature matrix corresponding to all sliding time windows (all steady-state scenarios) constructed with each historical data acquisition time point as the starting point. We then calculate the baseline feature matrix for the m-th level using the following formula:

[0085] ;

[0086] in, This represents the baseline feature matrix of the m-th level. When the operating scenario of a wind turbine generator is considered to be a steady-state operation scenario, it represents the set of all sliding time windows constructed with each historical data acquisition time point as the starting point. This represents the total number of sliding time windows when the operating scenario of the wind turbine generator is a steady-state operation scenario.

[0087] It should be noted that the benchmark feature matrix calculation formula only uses the feature matrix under steady-state operation. By eliminating the influence of random noise and minor operating condition fluctuations under steady-state conditions through arithmetic averaging, a standard feature matrix under the unit's health state is obtained, completely eliminating the contamination of the benchmark by dynamic disturbance data. Furthermore, by using a sliding time window, the one-dimensional time-series features are upgraded to a two-dimensional feature matrix, preserving the evolution trajectory of features over continuous time, rather than the single-point comparison of traditional schemes. This better characterizes the gradual process of the unit's health state and provides a quantitative reference standard for the unit's health state. It is the core benchmark for subsequent calculation of health deviation. The essence of health assessment is the comparison of the difference between real-time features and the health benchmark. The benchmark constructed by this benchmark feature matrix calculation formula is completely consistent with the unit's own health operation characteristics, rather than an industry-wide universal threshold, making it more adaptable.

[0088] Wind turbine operating data inevitably contains sensor noise and instantaneous random fluctuations caused by grid flicker. Traditional solutions use characteristics from a single point in time as a benchmark, which is easily contaminated by random noise. This benchmark feature matrix calculation formula uses the arithmetic mean of the feature matrices of all steady-state sliding windows. Through the averaging effect, the influence of random noise is completely offset. The resulting benchmark matrix is ​​the essential characteristic of the unit's health status, rather than an instantaneous value disturbed by noise. Individual abnormal data will not affect the stability of the overall benchmark.

[0089] During real-time monitoring, the current real-time data acquisition time point is used as the end point of the window. The current sliding time window is constructed by taking the previous L-1 consecutive historical data acquisition time points, and the feature matrix of the m-th level under the current sliding time window is constructed. ;

[0090] Based on the feature matrix of the m-th level under the current sliding time window Using the baseline feature matrix of the m-th level, calculate the local health deviation of the m-th level using the following formula:

[0091] ;

[0092] in, This represents the local health deviation at the m-th level. Denotes the Frobenius norm. This represents a preset constant (a very small constant to prevent division by zero errors).

[0093] It should be noted that the Frobenius norm (the square root of the sum of the squares of all elements in the matrix) comprehensively captures the overall difference between the real-time feature matrix and the health baseline matrix. This difference represents the degree of health degradation at the corresponding level; that is, the greater the difference, the higher the deviation, and the worse the health status. The Frobenius norm can cover changes in all feature dimensions and at all time points in the matrix, capturing both single-dimensional anomalies of vibration, temperature, and coupling features, as well as anomalies in the temporal evolution of features, without any omissions.

[0094] This formula for calculating local health deviation enables a quantitative and continuous assessment of the degree of health degradation at each level of the unit. It upgrades the binary judgment of health / failure to a continuous indicator that can characterize the entire lifecycle from health to minor degradation and then to severe failure, achieving early fault warning. It can completely capture anomalies in all dimensions and at all time points in the feature matrix. Whether it's a single-point abrupt change in vibration characteristics, a slow gradual change in temperature characteristics, or a long-term anomaly in coupling characteristics, all will be fully included in the norm difference, without missing any failure mode. Traditional Euclidean distance and cosine similarity either only focus on single-point differences or only on vector direction, and their ability to capture gradual anomalies in the time-series matrix is ​​extremely poor. The Frobenius norm is the optimal choice for quantifying differences in a two-dimensional feature matrix, perfectly suited to the matrix-based health assessment logic of this patent.

[0095] The local health deviations at all levels are weighted and fused to calculate the overall health index of the machine, denoted as . ,in, , This represents the preset influence factor for the m-th level (determined using the analytic hierarchy process or based on historical fault data statistics; default recommended value:). =0.35 (blade layer) =0.35 (transmission chain layer) =0.20 (power generation layer) =0.10 (support layer)), and, .

[0096] Step S5: Obtain the operating condition scenario factors and operating condition correction factors at the real-time data acquisition time point; dynamically correct the overall health indicators of the machine; preset the health warning indicator thresholds, analyze and issue health warnings.

[0097] Specifically, based on historical data collection time points Operating scenario factors of wind turbine generator sets and operating condition correction factor During real-time monitoring, wind speed and rotational speed data are acquired at the real-time data acquisition time point, and the operating condition scenario factor and operating condition correction factor of the wind turbine generator at the real-time data acquisition time point are recorded as follows: and ;

[0098] Overall health indicators of the machine Dynamic corrections will be made, as follows:

[0099] like This indicates that the operating scenario of the wind turbine generator at the real-time data acquisition time point is a steady-state operating scenario, let ,in, This indicates the overall health index of the entire machine after dynamic correction;

[0100] like If , it indicates that the operating scenario of the wind turbine generator at the data acquisition time point is a dynamic disturbance scenario, then let ,in, This indicates the overall health index of the entire machine after dynamic correction;

[0101] Preset health warning indicator thresholds (the health warning indicator thresholds are based on the overall health indicators of the whole machine under historical steady-state data) Distribution determination: Take the overall health index of the whole machine during the historical steady-state operation period. (95th percentile), if the overall health index of the machine is dynamically corrected If the health warning indicator threshold is greater than or equal to the threshold, a health warning will be issued to the relevant staff.

[0102] In this invention, the wind power health monitoring industry has long faced a dilemma: if the threshold is set too low, gusts of wind will trigger numerous false alarms, overwhelming maintenance personnel; if the threshold is set too high, early faults will be missed, leading to unplanned unit shutdowns. This formula perfectly solves this problem through a bi-branch design: no corrections are made in steady-state scenarios. It retains the highest fault sensitivity, enabling the identification of even minor early-stage faults; under dynamic disturbance scenarios, it utilizes comprehensive health indicators of the entire machine. Divided by operating condition correction factor It dynamically offsets the increase in non-fault indicators caused by operating condition disturbances, completely suppressing false alarms. It truly achieves absolute sensitivity when it needs to be sensitive and absolute anti-interference when it needs to be anti-interference, which is something that traditional fixed threshold and fixed correction schemes cannot achieve at all.

[0103] In dynamic disturbance scenarios, The increase comes from two sources: non-fault-related increases due to operating condition disturbances and actual increases due to unit health deterioration. This formula uses an operating condition correction factor. This completely offset the non-faulty increase, and the corrected By retaining only the true signals of unit health deterioration and comparing them with the early warning threshold, the accuracy and reliability of the early warning results are greatly improved.

[0104] Example 2, please refer to Figure 2 The system provides a health status monitoring system for wind turbine generators based on sensor networks. The system includes the following steps: a collection construction module, an operating condition scenario construction module, a feature calculation and matrix construction module, a comprehensive index calculation module, and an index correction and early warning module.

[0105] The collection construction module: acquires the mechanical structure and energy transfer path of the wind turbine generator set, constructs a hierarchical index set; deploys sensor modules and constructs historical data acquisition time periods;

[0106] The operating condition scenario construction module: constructs a normalized data set of the blade layer, transmission chain layer, power generation layer and support layer at the historical data collection time point; collects wind speed data and rotational speed data, and analyzes the operating condition scenario of the wind turbine generator set at the historical data collection time point.

[0107] The feature calculation and matrix construction module analyzes the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of each level at historical data acquisition time points; constructs a working condition correction factor based on the working condition scenario factor; obtains the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of all levels at all historical data acquisition time points, and constructs the original feature sequence matrix; and sets the sliding time window length.

[0108] The comprehensive index calculation module: constructs the feature matrix of the level under the sliding time window; calculates the benchmark feature matrix of the level; constructs the feature matrix of the level under the current sliding time window; and calculates the local health deviation of the level based on the feature matrix and the benchmark feature matrix, and calculates the overall health index of the whole machine.

[0109] The indicator correction and early warning module: acquires the operating condition scenario factors and operating condition correction factors at the real-time data acquisition time point; dynamically corrects the overall health indicators of the machine; presets health early warning indicator thresholds, analyzes and issues health early warnings.

[0110] Furthermore, the feature calculation and matrix construction module includes a feature calculation unit and a matrix construction unit;

[0111] The feature calculation unit analyzes the vibration response intensity characteristics and temperature change characteristics of each layer at the historical data acquisition time point based on the vibration and temperature data of the blade layer, transmission chain layer, power generation layer, and support layer at the historical data acquisition time point; calculates the vibration-temperature coupling characteristics of each layer at the historical data acquisition time point; and constructs an operating condition correction factor based on the operating condition scenario factors of the wind turbine generator at the historical data acquisition time point.

[0112] The matrix construction unit: based on the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the lower level at all historical data acquisition time points, obtains the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the lower level at all historical data acquisition time points, and constructs the original feature sequence matrix; and sets the sliding time window length.

[0113] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. The sensor network-based wind turbine health status monitoring method and specific examples in Embodiment 1 are also applicable to the sensor network-based wind turbine health status monitoring system of this embodiment. Through the foregoing detailed description of the sensor network-based wind turbine health status monitoring method, those skilled in the art can clearly understand the sensor network-based wind turbine health status monitoring system of this embodiment. Therefore, for the sake of brevity, it will not be described in detail here. As for the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and relevant parts can be referred to in the method section.

[0114] In the several embodiments provided by this invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0115] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0116] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0117] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory, random access memory, magnetic disks, or optical disks.

[0118] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0119] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0120] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. If such modifications and variations fall within the scope of this invention and its equivalents, then this invention is also intended to include such modifications and variations.

Claims

1. A method for monitoring the health status of wind turbine generators based on sensor networks, characterized in that, The method includes the following steps: Step S1: Obtain the mechanical structure and energy transfer path of the wind turbine generator set, and construct a hierarchical index set; deploy sensor modules and construct historical data acquisition time periods; Step S2: Construct a normalized dataset of the blade layer, drive train layer, power generation layer, and support layer at the historical data collection time points; collect wind speed data and rotational speed data, and analyze the operating conditions of the wind turbine generator set at the historical data collection time points; Step S3: Analyze the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of each level at the historical data acquisition time points; construct the working condition correction factor based on the working condition scenario factor; obtain the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of all levels at the historical data acquisition time points, and construct the original feature sequence matrix; set the sliding time window length. Step S4: Construct the feature matrix of the level under the sliding time window; calculate the baseline feature matrix of the level; construct the feature matrix of the level under the current sliding time window; based on the feature matrix and the baseline feature matrix, calculate the local health deviation of the level and calculate the overall health index of the whole machine. Step S5: Obtain the operating condition scenario factors and operating condition correction factors at the real-time data acquisition time point; dynamically correct the overall health indicators of the machine; preset the health warning indicator thresholds, analyze and issue health warnings.

2. The method for monitoring the health status of wind turbine generator sets based on sensor networks according to claim 1, characterized in that, The specific implementation process of step S1 includes: The mechanical structure and energy transfer path of the wind turbine generator are obtained, and based on the mechanical structure and energy transfer path, the wind turbine generator is divided into a blade layer, a transmission chain layer, a power generation layer, and a support layer. A hierarchical index set is constructed, denoted as . ,in, , , and These respectively represent the blade layer, transmission chain layer, power generation layer, and support layer of a wind turbine generator set; In the blade layer Transmission chain layer Power generation layer and support layer An internal sensor module is installed, including a vibration sensor and a temperature sensor; historical data acquisition time periods are established, denoted as... ,in, Let t represent the t-th historical data collection time point, and T represent the total number of historical data collection time points.

3. The method for monitoring the health status of wind turbine generator sets based on sensor networks according to claim 2, characterized in that, The specific implementation process of step S2 includes: Constructing historical data collection time points Lower blade layer Transmission chain layer Power generation layer and support layer The normalized datasets are denoted as follows: , , and ,in, and These represent the historical data collection time points. Leaf layer collected from the bottom Vibration data and temperature data, and These represent the historical data collection time points. Lower-level transmission chain layer Vibration data and temperature data, and These represent the historical data collection time points. Lower collection of power generation layer Vibration data and temperature data, and These represent the historical data collection time points. Support layer for lower collection Vibration data and temperature data.

4. The method for monitoring the health status of wind turbine generator sets based on sensor networks according to claim 3, characterized in that, The specific implementation process of step S2 also includes: Historical data collection time points The operating parameter data of the wind turbine generator set, including wind speed data and rotational speed data, are normalized and denoted as follows: and Based on historical data collection time points Wind speed data for wind turbine generators and speed data Analyze historical data collection time points The specific operating scenarios for wind turbine generators are as follows: ； in, Indicates the time point of historical data collection Operating scenario factors for wind turbine generators and These represent the historical data collection time points. Wind speed and rotational speed data for the wind turbine generator set. This represents a preset constant. This represents the preset threshold value for change. When t=1, it is defined as follows: ; like This indicates the historical data collection time point. The operating scenario for the wind turbine generator set is a dynamic disturbance scenario; if This indicates the historical data collection time point. The operating scenario for the wind turbine generator set is a steady-state operation scenario.

5. The method for monitoring the health status of wind turbine generator sets based on sensor networks according to claim 4, characterized in that, The specific implementation process of step S3 includes: Based on historical data collection time points Lower blade layer Transmission chain layer Power generation layer and support layer Vibration and temperature data, and analysis of historical data acquisition time points. The vibration response intensity characteristics and temperature change characteristics of each level are as follows: ； in, Indicates the time point of historical data collection Vibration response intensity characteristics at the m-th level. Indicates the time point of historical data collection Temperature change characteristics at the m-th level and These represent the historical data collection time points. Vibration and temperature data at the m-th level. and These represent the historical data collection time points. Vibration and temperature data at the m-th level. This represents a preset constant. When t=1, define , ; Calculate historical data collection time points The vibration-temperature coupling characteristics of the m-th level are calculated using the following formula: ； in, Indicates the time point of historical data collection The vibration-temperature coupling characteristics of the m-th level. This represents a preset constant; Based on historical data collection time points Operating Scenario Factors of Wind Turbine Generators Construct the operating condition correction factor, denoted as ,in, , and These represent the preset modulation coefficients.

6. The method for monitoring the health status of wind turbine generator sets based on sensor networks according to claim 5, characterized in that, The specific implementation process of step S3 also includes: Based on historical data collection time points Vibration response intensity characteristics of the m-th level Temperature change characteristics and vibration-temperature coupling characteristics The vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the m-th level at all historical data acquisition time points are obtained, and the original feature sequence matrix is ​​constructed, denoted as... ,in, ; Set the length of the sliding time window, denoted as L, and denote the starting time index of the k-th sliding time window as... Let the index of the end time of the k-th sliding time window be denoted as . ,in, , .

7. The method for monitoring the health status of wind turbine generator sets based on sensor networks according to claim 6, characterized in that, The specific implementation process of step S4 includes: Based on the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the m-th level under the k-th sliding time window, the feature matrix of the m-th level under the k-th sliding time window is constructed, denoted as... ,in, ; We obtain all historical data acquisition time points for the wind turbine generator set under steady-state operation, and acquire the feature matrix corresponding to all sliding time windows constructed with each historical data acquisition time point as the starting point. We then calculate the baseline feature matrix for the m-th level using the following formula: ； in, This represents the baseline feature matrix of the m-th level. When the operating scenario of a wind turbine generator is considered to be a steady-state operation scenario, it represents the set of all sliding time windows constructed with each historical data acquisition time point as the starting point. This represents the total number of sliding time windows when the operating scenario of the wind turbine generator is a steady-state operation scenario. During real-time monitoring, the current real-time data acquisition time point is used as the end point of the window. The current sliding time window is constructed by taking the previous L-1 consecutive historical data acquisition time points, and the feature matrix of the m-th level under the current sliding time window is constructed. ; Based on the feature matrix of the m-th level under the current sliding time window Using the baseline feature matrix of the m-th level, calculate the local health deviation of the m-th level using the following formula: ； in, This represents the local health deviation at the m-th level. Describing the Frobenius norm, This represents a preset constant; The local health deviations at all levels are weighted and fused to calculate the overall health index of the machine, denoted as . ,in, , Let represent the preset influence factor for the m-th level, and . .

8. The method for monitoring the health status of wind turbine generator sets based on sensor networks according to claim 7, characterized in that, The specific implementation process of step S5 includes: Based on historical data collection time points Operating Scenario Factors of Wind Turbine Generators and operating condition correction factor During real-time monitoring, wind speed and rotational speed data are acquired at the real-time data acquisition time point, and the operating condition scenario factor and operating condition correction factor of the wind turbine generator at the real-time data acquisition time point are recorded as follows: and ; Overall health indicators of the machine Dynamic corrections will be made, as follows: like This indicates that the operating scenario of the wind turbine generator at the real-time data acquisition time point is a steady-state operating scenario, let ,in, This indicates the overall health index of the entire machine after dynamic correction; like If , it indicates that the operating scenario of the wind turbine generator at the data acquisition time point is a dynamic disturbance scenario, then let ,in, This indicates the overall health index of the entire machine after dynamic correction; The preset health warning indicator threshold, if the overall health index of the whole machine is dynamically adjusted... If the health warning indicator threshold is greater than or equal to the threshold, a health warning will be issued to the relevant staff.

9. A wind turbine generator health status monitoring system based on sensor networks, characterized in that, The system includes: a set construction module, a working condition scenario construction module, a feature calculation and matrix construction module, a comprehensive index calculation module, and an index correction and early warning module; The collection construction module: acquires the mechanical structure and energy transfer path of the wind turbine generator set, constructs a hierarchical index set; deploys sensor modules and constructs historical data acquisition time periods; The operating condition scenario construction module: constructs a normalized data set of the blade layer, transmission chain layer, power generation layer and support layer at the historical data collection time point; collects wind speed data and rotational speed data, and analyzes the operating condition scenario of the wind turbine generator set at the historical data collection time point. The feature calculation and matrix construction module analyzes the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of each level at historical data acquisition time points; constructs a working condition correction factor based on the working condition scenario factor; obtains the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of all levels at all historical data acquisition time points, and constructs the original feature sequence matrix; and sets the sliding time window length. The comprehensive index calculation module: constructs the feature matrix of the level under the sliding time window; calculates the benchmark feature matrix of the level; constructs the feature matrix of the level under the current sliding time window; and calculates the local health deviation of the level based on the feature matrix and the benchmark feature matrix, and calculates the overall health index of the whole machine. The indicator correction and early warning module: acquires the operating condition scenario factors and operating condition correction factors at the real-time data acquisition time point; dynamically corrects the overall health indicators of the machine; presets health early warning indicator thresholds, analyzes and issues health early warnings.

10. The wind turbine generator health status monitoring system based on sensor networks according to claim 9, characterized in that, The feature calculation and matrix construction module includes a feature calculation unit and a matrix construction unit; The feature calculation unit analyzes the vibration response intensity characteristics and temperature change characteristics of each layer at the historical data acquisition time point, based on the vibration and temperature data of the blade layer, transmission chain layer, power generation layer and support layer at the historical data acquisition time point. Calculate the vibration-temperature coupling characteristics at the lower level of historical data acquisition time points; construct operating condition correction factors based on the operating condition scenario factors of wind turbine generators at the historical data acquisition time points; The matrix construction unit: Based on the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the lower level at all historical data acquisition time points, it acquires the vibration response intensity characteristics, temperature change characteristics, and vibration-temperature coupling characteristics of the lower level at all historical data acquisition time points, and constructs the original feature sequence matrix. Set the length of the sliding time window.

Images