Fault early warning method for high-speed exhaust system

Abstract

The invention relates to the technical field of fault prediction and health management, and particularly discloses a fault early warning method for a high-speed exhaust system, and the method comprises the steps: S1, setting a fan speed raising stage as a power rising stage, collecting power and vibration data, and generating a corresponding trend chart; s2, fitting a curve for the trend chart, and counting historical data according to power feature points to obtain a comprehensive radial and axial reference curve; s3, generating upper and lower limit threshold points based on the statistical standard deviation, and fitting to obtain a dynamic upper and lower limit curve corresponding to the power; s4, comparing the real-time vibration with a dynamic threshold value, and if the two directions are normal, judging that no fault exists; and S5, judging the working condition through the vibration variance in the power stable period. According to the method, the dynamic power-vibration curve is constructed for the speed raising stage of the fan, the limitation of traditional steady state monitoring is broken through, and early hidden fault accurate recognition can be achieved; and radial and axial modeling and dynamic threshold value intervals are adopted, so that misjudgment and missed judgment are effectively reduced, and the fault early warning precision and the system operation reliability are improved.

Description

Technical Field

[0001] This invention relates to the field of fault prediction and health management technology, specifically to a fault early warning method for high-speed exhaust systems. Background Technology

[0002] High-speed exhaust systems are widely used in rail transit, large industrial plants, and underground buildings, serving as crucial equipment for ensuring ventilation and safe smoke extraction. During actual operation, the fan's operating conditions change drastically from startup to reaching rated power. Early, latent faults in the impeller, bearings, and transmission mechanisms are prone to occur. If these faults are not identified and warned of in time, they can easily escalate into excessive vibration, component damage, or even system shutdown, severely impacting the stability and safety of the ventilation system. Current fault monitoring for high-speed exhaust systems primarily focuses on the steady-state operation of the fan, paying insufficient attention to the dynamic characteristics during power increase. This makes effective early warning at the initial fault stage difficult, resulting in significant monitoring blind spots.

[0003] Existing wind turbine fault early warning methods mostly rely on fixed threshold judgments, failing to establish dynamic judgment benchmarks that match power changes. This makes them susceptible to load fluctuations and environmental interference, leading to insufficient early warning accuracy. Furthermore, most monitoring schemes lack differentiated analysis and statistical modeling for radial and axial vibrations, failing to distinguish the fault sensitivity characteristics of vibrations in different directions, easily resulting in misjudgments or missed diagnoses. In addition, existing technologies lack effective invalid record removal rules and sample statistical optimization methods during data processing, resulting in low utilization of historical operating data. The constructed judgment benchmarks fail to accurately reflect the normal operating range of the equipment, further reducing the reliability of fault identification. Overall, existing technologies are insufficient in terms of early warning timeliness, judgment accuracy, and system applicability, making it difficult to meet the long-term stable, safe, and reliable operation requirements of high-speed exhaust systems. Summary of the Invention

[0004] The purpose of this invention is to provide a fault early warning method for high-speed exhaust systems, thereby solving the above-mentioned technical problems.

[0005] The objective of this invention can be achieved through the following technical solutions: A fault early warning method for a high-speed exhaust system includes the following steps: The process of increasing the power of the wind turbine to the rated power is recorded as the power rise period; the operating power, radial vibration velocity and axial vibration velocity of the wind turbine during the power rise period are obtained, and a "power-radial" trend graph is generated with the operating power as the abscissa and the radial vibration velocity as the ordinate; a "power-axial" trend graph is generated with the operating power as the abscissa and the axial vibration velocity as the ordinate.

[0006] As a further aspect of the present invention: fitting the radial curve equation C1(x) corresponding to the "power-radial" trend graph, and fitting the axial curve equation C2(x) corresponding to the "power-axial" trend graph; A power interval p is preset, and a power characteristic point is set for each power interval p. Based on the "power-radial" trend graph, the following operations are performed: Obtain historical operating records of the wind turbine, and at the same power characteristic point, average the radial vibration velocity of N radial curve equations. Where N represents the number of radial curve equations, a represents the operating power corresponding to the power characteristic point, and C1 n (a) represents the radial curve equation corresponding to the nth operation record of the wind turbine; Generate radial average points (a, C1) avg (a) Repeat the above steps to generate radial average points corresponding to all power feature points, and generate the comprehensive radial curve equation C1_sta(x) by fitting the radial average points; similarly, generate the comprehensive axial curve equation C2_sta(x).

[0007] As a further aspect of the present invention: at the same power characteristic point, calculate the standard deviation s of the radial vibration velocity of N radial curve equations, and generate radial upper threshold points (a, C1). avg (a) + s) and radial lower threshold point (a, C1) avg (a)-s), repeat the above steps for all power characteristic points; The upper radial threshold point is used to fit the upper radial threshold point to generate the upper radial curve equation C1_up(x), and the lower radial threshold point is used to fit the lower radial threshold point to generate the lower radial curve equation C1_down(x). Similarly, the upper axial limit curve equation C2_up(x) and the lower axial limit curve equation C2_down(x) are generated.

[0008] As a further aspect of the present invention: when the fan starts and is in the power rise period, the real-time operating power b of the fan and the radial vibration velocity C1_b corresponding to the operating power b are obtained. If C1_down(b)≤C1_b≤C1_up(b), it is recorded as normal radial vibration; otherwise, it is recorded as abnormal radial vibration. The above operation is repeated to complete the judgment of axial vibration. When both radial and axial vibrations are normal, it is recorded as normal during the power rise period; otherwise, it is recorded as abnormal during the power rise period, indicating to staff that there is a malfunction in the ventilation system.

[0009] As a further aspect of the present invention: when the power rise period is normal, the period after the fan reaches the target power and operates stably for a preset time is recorded as the power stabilization period; When the power is stable, the vibration sensor is installed at a preset radial position on the drive end of the fan to collect the axial vibration amplitude. The preset collection frequency is 1kHz and the collection duration is 1kV. Calculate the variance of the collected axial vibration amplitude. If the variance is greater than or equal to the preset vibration limit value, it indicates that there is a fault in the exhaust system, prompting the staff to shut down the fan and carry out maintenance; if it is less than the preset vibration limit, it indicates that the exhaust system is normal.

[0010] As a further aspect of the present invention, the power rise period corresponding to 0% to 0.5% of the rated power of the wind turbine is excluded and not included in subsequent calculations.

[0011] As a further aspect of the present invention, the sampling frequency during the power rise period is not less than 10Hz.

[0012] As a further aspect of the present invention: invalid records of shutdown, power outage, and manual power adjustment are removed.

[0013] The beneficial effects of the present invention are as follows: (1) By constructing trend diagrams of "power-radial vibration" and "power-axial vibration" and fitting dynamic curves during the power rise phase of the wind turbine, the present invention breaks through the limitation of traditional fault monitoring that only targets the steady-state operation phase. It can capture early hidden fault characteristics during the critical period of drastic changes in the operating conditions of wind turbine startup and speed-up, and achieve accurate identification of fault budding stage. Compared with the fixed threshold judgment method, the present invention establishes a dynamic judgment benchmark that changes with power as a variable. It can effectively reduce the influence of load fluctuations, environmental interference and other factors on the monitoring results, significantly improve the timeliness and stability of fault early warning, avoid the gradual development of serious problems such as component damage and system shutdown due to the failure to detect early faults in time, and improve the operational safety and reliability of high-speed exhaust system.

[0014] (2) Radial and axial vibration velocities are collected separately, their curves are independently fitted, and differentiated upper and lower thresholds are constructed to achieve refined statistical modeling of multi-directional vibration characteristics. This can fully distinguish the sensitivity characteristics of vibration in different directions to faults and effectively reduce the risk of misjudgment and missed judgment caused by single threshold judgment. At the same time, by calculating the average and standard deviation of multiple historical operating curves, a comprehensive benchmark curve and dynamic threshold range are formed, making the judgment benchmark more consistent with the actual normal operating range of the equipment and greatly improving the accuracy of fault identification. This method can make full use of historical operating data, improve data utilization and model applicability, and provide a stable and reliable full-cycle fault early warning capability for high-speed exhaust systems. Attached Figure Description

[0015] The invention will now be further described with reference to the accompanying drawings.

[0016] Figure 1This is a schematic diagram of the structure of a fault early warning method for a high-speed exhaust system according to the present invention; Figure 2 This is a flowchart illustrating a fault early warning method for a high-speed exhaust system according to the present invention. Detailed Implementation

[0017] 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. 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.

[0018] Please see Figure 1 As shown, the present invention is a fault early warning method for a high-speed exhaust system, comprising the following steps: The process of increasing the power of the wind turbine to the rated power is recorded as the power rise period; the operating power, radial vibration velocity and axial vibration velocity of the wind turbine during the power rise period are obtained, and a "power-radial" trend graph is generated with the operating power as the abscissa and the radial vibration velocity as the ordinate; a "power-axial" trend graph is generated with the operating power as the abscissa and the axial vibration velocity as the ordinate.

[0019] It should be noted that the entire dynamic process from the start-up of the wind turbine to the point where the power rises to the rated power is defined as the power rise period. This stage is the critical stage where the wind turbine's operating conditions change most drastically and early latent faults are most easily exposed.

[0020] After the wind turbine starts up and enters the power increase phase, three operational data points—operating power, radial vibration velocity, and axial vibration velocity—are simultaneously acquired through a real-time acquisition module. Operating power is collected by a power sensor, while radial and axial vibration velocities are collected by vibration sensors positioned radially and axially on the wind turbine bearing housing, respectively, ensuring a one-to-one correspondence between the data in terms of time and power. Using the acquired operating power as the x-axis and the corresponding radial vibration velocity at the same moment as the y-axis, a "power-radial" vibration trend chart is generated, visually reflecting the distribution and trend of radial vibration as power changes during the turbine's speed-up process. Simultaneously, using the same operating power as the x-axis and the corresponding axial vibration velocity at the same moment as the y-axis, a "power-axial" vibration trend chart is generated, allowing for independent display and comparative analysis of radial and axial vibration characteristics during the power increase period. Both trend charts are based on continuous power ranges, fully covering the entire process of the wind turbine increasing from low power to rated power. This provides an intuitive and reliable data foundation for subsequent curve fitting, benchmark construction, and anomaly detection, thereby enabling accurate characterization and quantitative analysis of the dynamic vibration characteristics of the wind turbine during the speed-up phase.

[0021] In another preferred embodiment of the present invention, the radial curve equation C1(x) corresponding to the "power-radial" trend graph is fitted, and the axial curve equation C2(x) corresponding to the "power-axial" trend graph is fitted. A power interval p is preset, and a power characteristic point is set for each power interval p. Based on the "power-radial" trend graph, the following operations are performed: Obtain historical operating records of the wind turbine, and at the same power characteristic point, average the radial vibration velocity of N radial curve equations. Where N represents the number of radial curve equations, a represents the operating power corresponding to the power characteristic point, and C1 n (a) represents the radial curve equation corresponding to the nth operation record of the wind turbine; Generate radial average points (a, C1) avg (a) Repeat the above steps to generate radial average points corresponding to all power feature points, and generate the comprehensive radial curve equation C1_sta(x) by fitting the radial average points; similarly, generate the comprehensive axial curve equation C2_sta(x).

[0022] It is worth noting that, in another preferred embodiment of the present invention, the "power-radial" trend graph and the "power-axial" trend graph are first fitted to obtain the corresponding radial curve equation C1(x) and axial curve equation C2(x). Subsequently, a power interval p is preset, which can be adaptively adjusted according to the rated power and operating characteristics of the fan. For example, for a high-speed exhaust fan with a rated power of 100kW, the power interval p can be set to 1kW. A power feature point is set for each power interval p, thereby forming a sequence of equally spaced power feature points in the range from 0 to the rated power, ensuring the granularity and accuracy of subsequent statistical analysis. Based on the "power-radial" trend chart, the following operations are performed: historical operating records of the fan are obtained. These historical operating records need to undergo invalid data removal preprocessing, that is, records under abnormal operating conditions such as shutdown, power outage, and manual power adjustment are removed. At the same time, low power data in the range of 0 to 0.5% of the rated power are removed to ensure the validity and representativeness of the sample. At the same power feature point, the radial vibration velocity of N radial curve equations is averaged. The radial average point (a, C1) is generated. avg (a) Repeat the above steps to generate radial average points corresponding to all power feature points, and then perform quadratic curve fitting through the radial average points to generate the comprehensive radial curve equation C1_sta(x). This comprehensive curve can reflect the typical statistical law of the radial vibration velocity of the wind turbine changing with power during the power rise period. Similarly, perform the same statistical and fitting operation based on the "power-axial" trend graph to generate the comprehensive axial curve equation C2_sta(x). The significance of this approach lies in the fact that by statistically averaging multiple historical operating curves at equally spaced power characteristic points, the influence of random factors such as load fluctuations and environmental interference during a single operation can be effectively eliminated. This allows the comprehensive curve to more accurately reflect the vibration baseline characteristics of the wind turbine under normal operating conditions. At the same time, the setting of equally spaced power characteristic points ensures the uniformity and comparability of the statistical analysis, providing a reliable statistical basis and quantitative evidence for the subsequent construction of dynamic threshold intervals and the realization of accurate fault early warning, thereby improving the accuracy of fault identification.

[0023] In a preferred embodiment, at the same power characteristic point, the standard deviation s of the radial vibration velocity of N radial curve equations is calculated to generate the radial upper threshold point (a, C1). avg (a) + s) and radial lower threshold point (a, C1) avg (a)-s), repeat the above steps for all power characteristic points; The upper radial threshold point is used to fit the upper radial threshold point to generate the upper radial curve equation C1_up(x), and the lower radial threshold point is used to fit the lower radial threshold point to generate the lower radial curve equation C1_down(x). Similarly, the upper axial limit curve equation C2_up(x) and the lower axial limit curve equation C2_down(x) are generated.

[0024] Understandably, at the same power characteristic point, based on the obtained average radial vibration velocity, the standard deviation s of the radial vibration velocity at the power characteristic point a is further calculated for N radial curve equations. This standard deviation can quantitatively reflect the dispersion of the radial vibration velocity at the power point and reflect the reasonable range of vibration fluctuation under normal working conditions. Subsequently, radial upper threshold points (a, C1_avg(a)+s) and radial lower threshold points (a, C1_avg(a)-s) are generated based on the mean and standard deviation. The above standard deviation calculation and threshold point generation operations are performed sequentially on all power feature points to obtain radial upper threshold point sequences and radial lower threshold point sequences covering the entire power rise interval. Next, by performing curve fitting on the radial upper threshold point sequence (using cubic spline interpolation or polynomial fitting), the radial upper limit curve equation C1_up(x) is generated, which represents the reasonable upper limit of radial vibration velocity at the corresponding power. Similarly, by performing the same curve fitting operation on the radial lower threshold point sequence, the radial lower limit curve equation C1_down(x) is generated, which represents the reasonable lower limit of radial vibration velocity at the corresponding power. Likewise, based on statistical data of axial vibration, the standard deviation of axial vibration velocity is calculated at each power characteristic point to generate axial upper and lower threshold points. Then, by curve fitting, the axial upper limit curve equation C2_up(x) and axial lower limit curve equation C2_down(x) are generated respectively. This constructs threshold intervals that dynamically change with power for radial and axial vibrations. These intervals accurately reflect the reasonable fluctuation range of vibration velocity at each power point under normal operating conditions, providing a precise dynamic benchmark for subsequent real-time monitoring and anomaly judgment. This effectively avoids misjudgments and missed judgments caused by fixed thresholds, improving the accuracy of fault early warning.

[0025] In a preferred embodiment, when the fan starts and is in the power rise period, the real-time operating power b of the fan and the radial vibration velocity C1_b corresponding to the operating power b are acquired. If C1_down(b)≤C1_b≤C1_up(b), the radial vibration is recorded as normal; otherwise, it is recorded as abnormal. The above operation is repeated to complete the judgment of axial vibration. When both radial and axial vibrations are normal, it is recorded as normal during the power rise period; otherwise, it is recorded as abnormal during the power rise period, indicating to staff that there is a malfunction in the ventilation system.

[0026] It is important to note that once the wind turbine receives the start command and enters the power ramp-up phase, the real-time monitoring module immediately enters the working state. It synchronously acquires the real-time operating power b of the wind turbine and the corresponding radial vibration velocity C1_b at a sampling frequency of not less than 10Hz, ensuring the real-time and continuous data acquisition and fully covering every operating condition node in the ramp-up process.

[0027] After acquiring a set of real-time data, the system will automatically call the pre-constructed radial lower limit curve equation C1_down(x) and radial upper limit curve equation C1_up(x), substitute them into the real-time operating power b, and calculate to obtain the reasonable threshold range of radial vibration velocity at that power point. Then, the real-time radial vibration velocity C1_b is precisely compared with this threshold range. If C1_down(b) ≤ C1_b ≤ C1_up(b), the current radial vibration state is determined to be normal; if the real-time value exceeds this range, whether it is higher than the upper limit or lower than the lower limit, it is immediately recorded as radial vibration abnormality, and the power, vibration value, and timestamp at that moment are retained as evidence of the abnormality.

[0028] After completing the radial vibration assessment, the system synchronously analyzes the real-time acquired axial vibration velocity according to the same logic and process, comparing it with the threshold range calculated from the upper and lower limit curve equations of the axial vibration to determine whether the axial vibration is normal or abnormal. Only when both the radial and axial vibration assessment results are normal will the current power rise phase operation status of the fan be recorded as overall normal; if the vibration in any direction is determined to be abnormal, the system will immediately mark the power rise phase as abnormal and issue a warning signal through both on-site audible and visual alarms and the back-end monitoring terminal, alerting staff to the potential early-stage faults in the high-speed exhaust system.

[0029] The significance of this approach lies in the fact that it enables refined monitoring of the power rise period through real-time comparison of dynamic thresholds, breaking through the limitations of traditional fixed thresholds and accurately capturing early latent faults in components such as impellers and bearings. At the same time, the adoption of radial and axial dual judgment logic avoids the risk of missed judgments caused by monitoring in a single direction, ensuring the comprehensiveness and accuracy of fault early warning. This provides key support for staff to intervene in troubleshooting in a timely manner and prevent the fault from escalating, effectively ensuring the safe operation of the high-speed exhaust system.

[0030] In a preferred embodiment, when the power rise period is normal and the fan reaches the target power and operates stably for a preset period of time, the power stabilization period is recorded. When the power is stable, the vibration sensor is installed at a preset radial position on the drive end of the fan to collect the axial vibration amplitude. The preset collection frequency is 1kHz and the collection duration is 1kV. Calculate the variance of the collected axial vibration amplitude. If the variance is greater than or equal to the preset vibration limit value, it indicates that there is a fault in the exhaust system, prompting the staff to shut down the fan and carry out maintenance; if it is less than the preset vibration limit, it indicates that the exhaust system is normal.

[0031] It should be noted that once the power rise period is determined to be normal, and the wind turbine reaches the preset target power and continues to operate stably for a preset duration (e.g., 30 seconds), the system automatically marks the current operating phase as the power stabilization period. This stage is the core operating condition for the long-term stable operation of the wind turbine, and also a key verification step to confirm the health status of the equipment after verifying the absence of hidden faults during the speed-up phase. When the wind turbine enters the power stabilization period, vibration sensors pre-installed at preset radial positions on the wind turbine drive end will start data acquisition at a preset acquisition frequency of 1kHz. The acquisition duration can be set from 10 to 60 seconds depending on the equipment operating conditions and monitoring requirements to ensure that the acquired vibration data has sufficient statistical representativeness. After the acquisition is completed, the system calculates the variance of the acquired axial vibration amplitude sequence. This variance can quantify the degree of fluctuation and dispersion of the vibration signal, effectively characterizing the vibration stability of the wind turbine under steady-state operation.

[0032] If the calculated variance is greater than or equal to the preset vibration limit value, it is determined that there is a potential fault in the exhaust system. The system will immediately issue an early warning through the back-end terminal, prompting staff to shut down the fan in time and carry out targeted maintenance. If the variance is less than the preset vibration limit value, it is determined that the exhaust system is operating normally during the power stability period and can continue to be put into use.

[0033] In another preferred embodiment of the present invention, the power rise period corresponding to 0% to 0.5% of the rated power of the wind turbine is excluded and is not included in subsequent calculations.

[0034] It is understandable that the operating conditions of a wind turbine are extremely unstable in the very low power range at startup. It is susceptible to factors such as startup shock, electrical transient fluctuations, and initial mechanical friction. The vibration and power data show obvious instantaneous disturbances and cannot reflect the true normal operating characteristics of the equipment.

[0035] Including this portion of data in the statistics would cause the baseline curve to shift and the threshold range to become distorted, thereby reducing the accuracy of fault identification. By removing this invalid range, the interference data at the moment of startup can be effectively eliminated, ensuring that the samples used for subsequent modeling are more representative and reliable. This makes the constructed dynamic vibration baseline more closely match the normal speed-up pattern of the wind turbine, improving the accuracy and stability of subsequent early warning judgments.

[0036] In another preferred embodiment of the present invention, the sampling frequency during the power rise period is not less than 10Hz.

[0037] It is worth noting that the sampling frequency during the power rise period is no less than 10Hz, which can ensure high density of operating power and radial and axial vibration velocity during the rapid change of the wind turbine speed-up condition, avoid the loss of key data points, and fully capture the subtle characteristics of vibration changes with power.

[0038] Sufficient data acquisition frequency can improve the accuracy of trend chart plotting and curve fitting, providing reliable data support for subsequent dynamic threshold construction and real-time anomaly judgment, and effectively improving the accuracy and timeliness of fault early warning.

[0039] In another preferred embodiment of the present invention, invalid records of shutdown, power outage, and manual power adjustment are removed.

[0040] It is worth noting that abnormal data interference from non-normal operating conditions can be eliminated, ensuring that all historical operating samples are valid data of the wind turbine's autonomous and normal speed-up. Using pure and valid samples to construct the baseline curve and threshold range can more realistically reflect the normal operating patterns of the equipment, avoid model distortion, and significantly improve the reliability of fault diagnosis and the accuracy of early warning.

[0041] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A fault early warning method for a high-speed exhaust system, characterized in that, Includes the following steps: The process of increasing the power of the wind turbine to the rated power is recorded as the power rise period; the operating power, radial vibration velocity and axial vibration velocity of the wind turbine during the power rise period are obtained, and a "power-radial" trend graph is generated with the operating power as the abscissa and the radial vibration velocity as the ordinate; a "power-axial" trend graph is generated with the operating power as the abscissa and the axial vibration velocity as the ordinate.

2. The fault early warning method for a high-speed exhaust system according to claim 1, characterized in that, Fit the radial curve equation C1(x) corresponding to the "power-radial" trend graph, and fit the axial curve equation C2(x) corresponding to the "power-axial" trend graph; A power interval p is preset, and a power characteristic point is set for each power interval p. Based on the "power-radial" trend graph, the following operations are performed: Obtain historical operating records of the wind turbine, and at the same power characteristic point, average the radial vibration velocity of N radial curve equations. Where N represents the number of radial curve equations, a represents the operating power corresponding to the power characteristic point, and C1 n (a) represents the radial curve equation corresponding to the nth operation record of the wind turbine; Generate radial average points (a, C1) avg (a) Repeat the above steps to generate radial average points corresponding to all power feature points, and generate the comprehensive radial curve equation C1_sta(x) by fitting the radial average points; similarly, generate the comprehensive axial curve equation C2_sta(x).

3. The fault early warning method for a high-speed exhaust system according to claim 2, characterized in that, At the same power characteristic point, calculate the standard deviation s of the radial vibration velocity of N radial curve equations, and generate the radial upper threshold point (a, C1). avg (a) + s) and radial lower threshold point (a, C1) avg (a)-s), repeat the above steps for all power characteristic points; The upper radial threshold point is used to fit the upper radial threshold point to generate the upper radial curve equation C1_up(x), and the lower radial threshold point is used to fit the lower radial threshold point to generate the lower radial curve equation C1_down(x). Similarly, the upper axial limit curve equation C2_up(x) and the lower axial limit curve equation C2_down(x) are generated.

4. The fault early warning method for a high-speed exhaust system according to claim 1, characterized in that, When the fan starts and is in the power rise period, the real-time operating power b of the fan and the radial vibration velocity C1_b corresponding to the operating power b are obtained. If C1_down(b)≤C1_b≤C1_up(b), it is recorded as normal radial vibration; otherwise, it is recorded as abnormal radial vibration. Repeat the above operation to complete the judgment of axial vibration. When both radial and axial vibrations are normal, it is recorded as normal during the power rise period; otherwise, it is recorded as abnormal during the power rise period, indicating to staff that there is a malfunction in the ventilation system.

5. A fault early warning method for a high-speed exhaust system according to claim 4, characterized in that, When the power rise period is normal, and the fan reaches the target power and operates stably for a preset period of time, the power stabilization period is recorded. When the power is stable, the vibration sensor is installed at a preset radial position on the drive end of the fan to collect the axial vibration amplitude. The preset collection frequency is 1kHz and the collection duration is 1kV. Calculate the variance of the collected axial vibration amplitude. If the variance is greater than or equal to the preset vibration limit value, it indicates that there is a fault in the exhaust system, prompting the staff to shut down the fan and carry out maintenance. If the vibration is less than the preset vibration limit, it means the exhaust system is normal.

6. A fault early warning method for a high-speed exhaust system according to claim 1, characterized in that, The power rise period corresponding to 0% to 0.5% of the rated power of the wind turbine is excluded and not included in subsequent calculations.

7. A fault early warning method for a high-speed exhaust system according to claim 1, characterized in that, The sampling frequency during the power rise period is no less than 10Hz.

8. A fault early warning method for a high-speed exhaust system according to claim 1, characterized in that, Invalid records include those of machine shutdown, power outage, and manual power adjustment.

Images