Axial flow fan fault early warning system based on multi-source parameter analysis

Abstract

This invention relates to the field of axial flow fans, addressing the problems of low reliability and high false alarm rate in traditional axial flow fan early warning systems under dynamic operating conditions. Specifically, it is an axial flow fan fault early warning system based on multi-source parameter analysis, including a terminal equipment parameter docking module, a feature response analysis module, a feature fusion analysis module, a fault early warning module, and a dynamic threshold adjustment module. In this invention, vibration parameters, current parameters, acoustic emission parameters, and shell temperature parameters are each individually fused into parameters and high-risk parameters are independently judged. This allows for direct identification and early warning of high-risk events such as current overload, blade breakage, and overheating faults at the acquisition end. Simultaneously, the individual fused parameters are weighted and fused to generate a health indicator. Combined with the dynamic threshold adjustment module, the initial threshold is corrected in real time based on the fan speed and load parameters, outputting the current health threshold, thus achieving a complementary two-level early warning mechanism.

Description

Technical Field

[0001] This invention relates to the field of axial flow fans, specifically to an axial flow fan fault early warning system based on multi-source parameter analysis. Background Technology

[0002] Axial flow fans are widely used in key areas such as industrial ventilation, power cooling, and mine ventilation. Their operational reliability directly affects production safety and equipment lifespan. Currently, fault early warning technology for axial flow fans has mainly evolved from single-parameter threshold judgment to multi-parameter comprehensive monitoring.

[0003] Early fault warning systems mostly used single-parameter threshold judgment methods, such as monitoring only vibration amplitude or shell temperature. When the parameter exceeds a preset fixed threshold, an alarm is triggered. This type of method is simple to implement and responds directly, but it has obvious drawbacks: the amount of information from a single parameter is limited, it is difficult to distinguish fault types from interference signals, and it is prone to false alarms or missed alarms due to environmental noise or fluctuations.

[0004] With the development of sensing and signal processing technologies, monitoring systems based on multi-source parameter acquisition have emerged, which can simultaneously acquire multiple parameters such as vibration, current, and temperature, and achieve fault diagnosis through feature extraction and fusion analysis. However, the following drawbacks still exist: I. Multi-source parameter fusion methods generally adopt a full-parameter synchronous fusion strategy, which means that all collected parameters are directly input into the fusion model for processing. This method has high computational complexity and poor real-time performance. Moreover, when a parameter is disturbed, it will directly pollute the fusion result, causing the overall judgment to fail. Second, the warning thresholds of conventional systems are mostly fixed values ​​or simply segmented settings, which cannot be adaptively adjusted in real time according to changes in the operating status of the fan. The accuracy of the warnings decreases significantly under dynamic conditions such as variable speed and variable load.

[0005] To address the aforementioned technical problems, this invention proposes a solution. Summary of the Invention

[0006] In this invention, vibration parameters, current parameters, acoustic emission parameters, and shell temperature parameters are each fused individually to generate parameters and independently determine high-risk conditions. This allows for direct identification and early warning of high-risk events such as current overload, blade breakage, and overheating faults at the acquisition end. Simultaneously, the individual fused parameters are weighted and fused to generate a health indicator. Combined with a dynamic threshold adjustment module, the initial threshold is corrected in real time based on the fan speed and load parameters, and the current health threshold is output. This achieves a complementary two-level early warning mechanism, thus proposing an axial flow fan fault early warning system based on multi-source parameter analysis.

[0007] The objective of this invention can be achieved through the following technical solution: an axial flow fan fault early warning system based on multi-source parameter analysis, including a terminal equipment parameter docking module, a feature response analysis module, a feature fusion analysis module, a fault early warning module, and a dynamic threshold adjustment module; The terminal device parameter interface module can collect the vibration parameters, current parameters, acoustic emission parameters and shell temperature parameters of the axial flow fan in real time. The feature response analysis module is used to analyze and process individual parameters separately, generate individual fusion parameters, determine whether there is a high risk of individual fusion parameters according to the set parameter allowable range, generate corresponding early warning signals, and send the corresponding early warning signals to the fault early warning module. The feature fusion analysis module obtains all individual fusion parameters through the feature response analysis module, and performs fusion analysis on each individual fusion parameter to obtain health indication values; The fault early warning module is used to directly trigger an early warning based on the early warning signal sent by the feature response analysis module, and is also used to compare the health indicator output by the feature fusion analysis module with the dynamic adaptive threshold to determine whether to trigger an early warning. The dynamic threshold adjustment module is used to adjust the preset initial threshold according to the wind turbine operation mode obtained by the terminal device parameter docking module, and output the adjusted current health threshold.

[0008] In a preferred embodiment of the present invention, the vibration parameters obtained by the terminal device parameter docking module include vibration amplitude and vibration frequency: The current parameters obtained by the terminal device parameter docking module include the effective value of the current and the instantaneous overshoot value of the current. The effective value of the current is obtained by calculating the root mean square of the instantaneous current, and the instantaneous overshoot value of the current is obtained by calculating the difference between the peak value of the current and the maximum value of the effective value of the current. The acoustic emission parameters obtained by the terminal device parameter docking module are high-frequency noise pulses with frequencies within a set range; The terminal device parameter docking module obtains the casing temperature parameters, namely the fan casing temperature and the casing temperature rise rate.

[0009] In a preferred embodiment of the present invention, the method for the feature response analysis module to generate single-item fusion parameters is as follows: The feature response analysis module integrates the vibration amplitude and vibration frequency to generate a vibration energy value. The feature response analysis module generates current load energy by weighting the effective value of the current and the instantaneous overshoot value of the current. The feature response analysis module calculates the pulse energy of the high-frequency noise pulses based on the acoustic emission parameters, retains the high-frequency noise with pulse energy greater than the filtering threshold, calculates the pulse frequency of the retained high-frequency noise, and obtains the acoustic emission fault features. The feature response analysis module compares the shell temperature and the upper limit of temperature at each moment to obtain the limit approach ratio, and obtains the heating rate at the same moment. The limit approach ratio is corrected by the heating rate to obtain the temperature risk ratio. Among them, vibration energy value, current load energy, acoustic emission fault characteristics, and temperature risk ratio are all individual fusion parameters.

[0010] In a preferred embodiment of the present invention, the feature response analysis module compares the vibration energy value with a set vibration energy threshold. If the vibration energy value exceeds the set vibration energy threshold, a vibration fault is determined. The feature response analysis module compares the current load energy with a set absolute current threshold. If the current load energy exceeds the set absolute current threshold, an overload fault is determined. The feature response analysis module compares the acoustic emission fault features with the set acoustic emission threshold. If the acoustic emission fault features exceed the set acoustic emission threshold, the blade structure is determined to be faulty. The feature response analysis module compares the temperature risk percentage with the set temperature risk threshold. If the temperature risk percentage exceeds the set temperature risk threshold, the fan is determined to be overheating.

[0011] In a preferred embodiment of the present invention, when the feature response analysis module does not trigger any warning signal, the feature fusion analysis module fuses each individual fusion parameter through a weighted algorithm to obtain a health indicator. The fault warning module compares the health indicator with a dynamic adaptive threshold. If the health indicator exceeds the dynamic adaptive threshold, the wind turbine is determined to be in a high-risk state, and a high-risk warning is generated.

[0012] In a preferred embodiment of the present invention, the specific fusion method of the feature fusion analysis module for individual fusion parameters is as follows: Obtain the contribution of each preset single fusion parameter to the axial flow fan failure, and generate weight coefficients based on the contribution, with the sum of the weight coefficients being 1; Obtain the allowable threshold corresponding to each individual fusion parameter, and calculate the ratio of each individual fusion parameter to the corresponding allowable threshold to obtain the individual proportion. Then, weight each individual proportion using a weighting coefficient to obtain the health indicator.

[0013] In a preferred embodiment of the present invention, the fan mode obtained by the dynamic threshold adjustment module includes fan speed and load parameters, and after obtaining the fan speed and load parameters, the speed and load parameters are divided into multiple mode intervals. The dynamic threshold adjustment module calculates the mean and standard deviation of the health indicators of the wind turbine health status within each mode interval. Based on the mean and standard deviation of the current pattern interval, a dynamic adaptive threshold is calculated according to a preset weighting relationship.

[0014] In a preferred embodiment of the present invention, when the fault early warning module obtains an early warning signal triggered by a single fusion project, it directly generates a first-level alarm signal and outputs a reminder; when it obtains a high-risk early warning, it generates a second-level early warning signal and outputs a reminder.

[0015] Compared with the prior art, the beneficial effects of the present invention are: In this invention, vibration parameters, current parameters, acoustic emission parameters, and shell temperature parameters are generated as individual fusion parameters and high-risk assessments are performed independently. This enables direct identification and early warning of high-risk events such as current overload, blade breakage, and overheating faults at the acquisition end. At the same time, it avoids contamination of subsequent multi-parameter fusion analysis by a single parameter and retains the advantage of fast response speed of single-parameter threshold judgment.

[0016] In this invention, when the feature response analysis module does not trigger any early warning signal, the individual fusion parameters such as vibration energy value, current load energy, acoustic emission fault characteristics, and temperature risk ratio are weighted and fused to generate a health indicator. Combined with the dynamic threshold adjustment module, the initial threshold is corrected in real time according to the fan speed and load parameters, and the current health threshold is output. This achieves mutual complementarity between the two-level early warning mechanisms, ensuring both rapid identification of obvious faults and sensitive capture of early micro-faults. This significantly improves the early warning cycle of the axial flow fan fault early warning system and reduces the false alarm rate and missed alarm rate.

[0017] This invention also automatically determines dynamic thresholds based on the fan's operating conditions, enabling the threshold of health indicators to be adaptively adjusted according to changes in operating conditions, effectively solving the problem of high false alarm rates under fixed threshold conditions with varying speeds and loads. Attached Figure Description

[0018] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0019] Figure 1 This is a system block diagram of the present invention; Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0020] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1: Please refer to Figures 1-2 As shown, the axial flow fan fault early warning system based on multi-source parameter analysis includes a terminal equipment parameter docking module, a feature response analysis module, a feature fusion analysis module, a fault early warning module, and a dynamic threshold adjustment module.

[0022] The terminal equipment parameter interface module is used to collect vibration parameters, current parameters, acoustic emission parameters and shell temperature parameters of the axial flow fan in real time. All parameters are collected using a unified time reference, and the sampling frequency is set according to the characteristics of each parameter.

[0023] Vibration parameters include vibration amplitude and vibration frequency. Sensors can be installed at key locations on the bearing housing or casing of the axial flow fan. The sensor output is converted into a voltage signal by a charge amplifier and acquired by an analog-to-digital converter to extract the vibration amplitude and the dominant vibration frequency. The vibration amplitude reflects the intensity of the vibration, and the vibration frequency reflects the periodicity of the vibration. The current parameters include the effective value of the current and the instantaneous overshoot value of the current. The effective value of the current is obtained by using a Hall current sensor to collect the instantaneous current waveform of a certain phase line of the motor. Multiple discrete points are collected within one power frequency cycle. Then, the instantaneous current values ​​of each sampling point are squared, summed, divided by the number of sampling points, and then the square root is taken to obtain the effective value of the current. This value is equivalent to the DC current value corresponding to the heat effect generated by the AC current on the resistor, reflecting the average load level of the motor during continuous operation. The method for obtaining the instantaneous overshoot value of the current is as follows: the current waveform is acquired using a high-speed sampling frequency, the peak current value is extracted in each sampling window, and the theoretical sine wave peak value corresponding to the effective value of the current in that window is calculated. The measured peak current value is subtracted from the theoretical sine wave peak value to obtain the instantaneous overshoot value of the current. This value reflects the transient impact component in the current waveform that exceeds the normal sine wave peak value, and its unit is consistent with the current unit. The acoustic emission parameters are high-frequency noise pulses within a set frequency range. They can be collected by attaching a piezoelectric acoustic emission sensor to the surface of the fan casing. Transient waveforms exceeding a preset trigger threshold are detected in real time, and each waveform exceeding the threshold is recorded as a high-frequency noise pulse. The casing temperature parameters include the fan casing temperature and the casing temperature rise rate. Temperature sensors can be installed on key parts of the fan casing. The casing temperature is obtained directly from the sensor readings. The temperature rise rate is obtained by calculating the difference between the temperature values ​​at two consecutive moments. That is, the temperature at the current moment is subtracted from the temperature at the previous moment, and then divided by the sampling interval time to obtain the temperature change per unit time. The feature response analysis module is used to analyze and process individual parameters separately, generate individual fusion parameters, and determine whether there is a high risk based on the set parameter allowable range. If there is, it generates a corresponding early warning signal and sends it to the fault early warning module. The feature response analysis module comprehensively calculates vibration amplitude and vibration frequency to generate vibration energy values. Specifically, it multiplies the vibration amplitude and vibration frequency at each sampling moment to obtain the instantaneous vibration intensity at that moment. Then, it accumulates and averages these instantaneous intensities within a set time window. Vibrations with high frequency and large amplitude will obtain higher energy values, avoiding the underestimation of two different risk modes: low-frequency large amplitude and high-frequency small amplitude. The feature response analysis module compares the vibration energy value with a preset vibration energy threshold. If the vibration energy value exceeds the threshold, it is determined to be a vibration fault and a vibration warning signal is generated. The feature response analysis module generates current load energy by weighted averaging the effective current value and the instantaneous current overshoot value. Specifically, the effective current value is multiplied by a set first weighting coefficient, the instantaneous current overshoot value is multiplied by a set second weighting coefficient, and then the two products are added together. The sum of the two weighting coefficients is 1. The effective current value has a slightly higher weight to reflect the dominant role of continuous load, while retaining the contribution of the instantaneous current overshoot value to transient impact. The feature response analysis module compares the current load energy with a preset absolute current threshold. If the current load energy exceeds the threshold, it is determined to be a current overload fault and a current overload warning signal is generated. The feature response analysis module calculates the pulse energy of high-frequency noise pulses based on acoustic emission parameters: for each high-frequency noise pulse, it calculates the cumulative amount of its waveform intensity over the entire pulse duration to obtain the single pulse energy. Then, it sets a filtering threshold. This threshold is greater than the maximum value of normal background noise pulse energy. Only high-frequency noise pulses with pulse energy greater than this filtering threshold are retained, thereby filtering out weak pulse interference caused by environmental noise and electrical noise. For the retained effective pulse sequence, the number of pulses per unit time is counted to obtain the acoustic emission fault characteristics. These characteristics directly reflect the frequency of effective acoustic emission events caused by faults such as blade crack propagation and shell abrasion. The feature response analysis module compares the acoustic emission fault features with the preset acoustic emission threshold. If the acoustic emission fault features exceed the threshold, it is determined to be a blade structure fault and a blade fault warning signal is generated. The feature response analysis module compares the shell temperature and the upper temperature limit at each moment to obtain the limit approach ratio and acquires the heating rate at the same moment. The heating rate is used to correct the limit approach ratio to obtain the temperature risk ratio. Specifically, the current shell temperature is first divided by the maximum allowable shell temperature of the fan to obtain a ratio between 0 and 1. The closer the ratio is to 1, the closer the temperature is to the limit. Then, the ratio is corrected according to the heating rate: when the heating rate is large, it means that the temperature is rising rapidly. Even if the current temperature has not yet reached the upper limit, its potential risk is higher than that of the steady-state high temperature state. Therefore, the limit approach ratio is amplified by a certain proportion. When the heating rate is small, the correction magnitude is also reduced accordingly. When the temperature risk ratio obtained after correction is greater than 1, it means that the temperature risk has exceeded the normal controllable range. The feature response analysis module compares the temperature risk ratio with the preset temperature risk threshold. If the temperature risk ratio exceeds the threshold, it is determined to be a fan overheating fault and an overheating warning signal is generated. Once any of the above warning signals is generated, the feature response analysis module immediately sends it to the fault warning module.

[0024] Example 2: Please refer to Figures 1-2 As shown, the feature fusion analysis module obtains all individual fusion parameters through the feature response analysis module, namely vibration energy value, current load energy, acoustic emission fault characteristics and temperature risk ratio. It is activated when the feature response analysis module does not trigger any warning signal, and performs fusion analysis on each individual fusion parameter to obtain health indication. The specific fusion method for individual fusion parameters in the feature fusion analysis module is as follows.

[0025] S1: Obtain the contribution of each preset single fusion parameter to the axial flow fan failure, and generate weight coefficients based on the contribution. For example, based on engineering experience or historical failure data analysis, the acoustic emission failure feature is set to have the highest contribution to the overall failure; the contribution of vibration energy value, current load energy and temperature risk ratio are relatively balanced, and the sum of each weight coefficient is 1. S2: Obtain the allowable threshold corresponding to each individual fusion parameter; S3: Calculate the ratio of each individual fusion parameter to its corresponding allowable threshold to obtain the individual ratio. This ratio is a dimensionless number. When the parameter reaches the threshold, the ratio is 1, and when it is below the threshold, the ratio is less than 1. S4: Weighted allocation of each individual proportion is performed using weighting coefficients. This involves multiplying each individual proportion by its corresponding weighting coefficient and then summing all the products to obtain the health indicator. The health indicator is a dimensionless number in the range of 0 to 1. When all parameters are far below the threshold, the health indicator is close to 0; when all parameters are close to the threshold, the health indicator is close to 1. This value comprehensively reflects the overall health status of the axial flow fan at the current moment. The larger the value, the closer it is to a fault state. The dynamic threshold adjustment module obtains the fan mode parameters, including fan speed and load parameters, from the terminal equipment parameter docking module. The load parameters can be characterized by the effective value of the current or the output power. After obtaining the speed and load parameters, they are divided into multiple mode intervals. For example, the speed is divided into low speed, medium speed and high speed, and the load is divided into light load, medium load and heavy load. The speed intervals and load intervals are combined to obtain nine mode partitions. For each mode partition, the system maintains an independent health indicator statistical baseline to adapt to the differences in the normal range of health indicators under different operating conditions. Within each mode interval, the mean and standard deviation of the health indicators of the wind turbine in a healthy state are statistically analyzed. Sufficient mode-stable samples can be collected for statistical calculation during the initial stage of system commissioning or the healthy operation phase after each maintenance. The formula for calculating the dynamic adaptive threshold is: the mean of the health indicators within the current mode interval, plus the preset threshold coefficient multiplied by the standard deviation of the health indicators within that interval. The physical meaning of this calculation method is: the health indicators in a healthy state fluctuate around the mean within a certain range, and those exceeding this range can be regarded as an over-limit state, triggering an alarm. The dynamic threshold adjustment module continuously updates the mean and standard deviation of the health indicator using a sliding time window. Within each time step, it extracts historical data of the health indicator within the window based on the current mode interval and recalculates the mean and standard deviation, thereby achieving real-time correction of the threshold. When the wind turbine has been running for a long time without failure, the mean of the health indicator may drift slowly due to normal wear and tear. The dynamic update mechanism can automatically adapt to this drift and avoid false alarms caused by fixed thresholds. When the fault early warning module receives any early warning signal sent by the feature response analysis module, it directly generates a first-level alarm signal and outputs a reminder. The first-level alarm corresponds to a clear and high-risk fault type, including vibration fault, current overload fault, blade fault and overheating fault, prompting the operator to stop the machine immediately for inspection. When the feature response analysis module does not trigger any early warning signal, but the health indicator output by the feature fusion analysis module exceeds the current health threshold output by the dynamic threshold adjustment module, the fault early warning module determines that the wind turbine is in a high-risk state, generates a level-two early warning signal, and outputs a reminder. The level-two early warning corresponds to a risk state that reflects potential early faults or multiple minor risks in synergy, prompting operators to strengthen monitoring or arrange maintenance and inspection in the near future.

[0026] Example 3: Please refer to Figures 1-2 As shown, the overall workflow of the system is as follows: Step 1: The terminal equipment parameter docking module synchronously collects the vibration, current, acoustic emission and shell temperature parameters of the axial flow fan in real time; Step 2: The feature response analysis module processes each individual parameter separately, generates individual fusion parameters, and determines whether there is a high risk. If there is, it directly sends an early warning signal to the fault early warning module. If none of the individual fusion parameters trigger an early warning, the feature fusion analysis module obtains each individual fusion parameter and generates a health indicator through weighted fusion. Step 3: The dynamic threshold adjustment module corrects the initial threshold according to the current wind turbine operating mode and outputs the current health threshold; Step 4: The fault warning module compares the health indicator with the current health threshold to determine whether to trigger a warning.

[0027] Thresholds, preset values, or preset ranges are set for result comparison and analysis to determine whether they are good or bad. The magnitude of these values ​​is determined by a combination of large-scale model analysis of the sample data and human experience. They can also be adjusted appropriately based on seasonal or common-sense influence conditions. Similarly, the weighting ratio coefficients and influence factors are set based on the magnitude of each parameter's influence on the results. These values ​​are assigned to reflect the overall impact on the results. They are also determined by a combination of large-scale model analysis of the sample data and human experience. They can also be adjusted appropriately based on seasonal or common-sense influence conditions.

[0028] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A fault early warning system for axial flow fans based on multi-source parameter analysis, characterized in that, It includes a terminal device parameter docking module, a feature response analysis module, a feature fusion analysis module, a fault early warning module, and a dynamic threshold adjustment module; The terminal device parameter interface module can collect the vibration parameters, current parameters, acoustic emission parameters and shell temperature parameters of the axial flow fan in real time. The feature response analysis module is used to analyze and process individual parameters separately, generate individual fusion parameters, determine whether there is a high risk of individual fusion parameters according to the set parameter allowable range, generate corresponding early warning signals, and send the corresponding early warning signals to the fault early warning module. The feature fusion analysis module obtains all individual fusion parameters through the feature response analysis module, and performs fusion analysis on each individual fusion parameter to obtain health indication values; The fault early warning module is used to directly trigger an early warning based on the early warning signal sent by the feature response analysis module, and is also used to compare the health indicator output by the feature fusion analysis module with the dynamic adaptive threshold to determine whether to trigger an early warning. The dynamic threshold adjustment module is used to adjust the preset initial threshold according to the wind turbine operation mode obtained by the terminal device parameter docking module, and output the adjusted current health threshold.

2. The axial flow fan fault early warning system based on multi-source parameter analysis according to claim 1, characterized in that, The vibration parameters obtained by the terminal device parameter docking module include vibration amplitude and vibration frequency: The current parameters obtained by the terminal device parameter docking module include the effective value of the current and the instantaneous overshoot value of the current. The effective value of the current is obtained by calculating the root mean square of the instantaneous current, and the instantaneous overshoot value of the current is obtained by calculating the difference between the peak value of the current and the maximum value of the effective value of the current. The acoustic emission parameters obtained by the terminal device parameter docking module are high-frequency noise pulses with frequencies within a set range; The terminal device parameter docking module obtains the casing temperature parameters, namely the fan casing temperature and the casing temperature rise rate.

3. The axial flow fan fault early warning system based on multi-source parameter analysis according to claim 1, characterized in that, The method by which the feature response analysis module generates individual fusion parameters is as follows: The feature response analysis module integrates the vibration amplitude and vibration frequency to generate a vibration energy value. The feature response analysis module generates current load energy by weighting the effective value of the current and the instantaneous overshoot value of the current. The feature response analysis module calculates the pulse energy of the high-frequency noise pulses based on the acoustic emission parameters, retains the high-frequency noise with pulse energy greater than the filtering threshold, calculates the pulse frequency of the retained high-frequency noise, and obtains the acoustic emission fault features. The feature response analysis module compares the shell temperature and the upper limit of temperature at each moment to obtain the limit approach ratio, and obtains the heating rate at the same moment. The limit approach ratio is corrected by the heating rate to obtain the temperature risk ratio. Among them, vibration energy value, current load energy, acoustic emission fault characteristics, and temperature risk ratio are all individual fusion parameters.

4. The axial flow fan fault early warning system based on multi-source parameter analysis according to claim 1, characterized in that, The feature response analysis module compares the vibration energy value with the set vibration energy threshold. If the vibration energy value exceeds the set vibration energy threshold, a vibration fault is determined. The feature response analysis module compares the current load energy with a set absolute current threshold. If the current load energy exceeds the set absolute current threshold, a current overload fault is determined. The feature response analysis module compares the acoustic emission fault features with the set acoustic emission threshold. If the acoustic emission fault features exceed the set acoustic emission threshold, the blade structure is determined to be faulty. The feature response analysis module compares the temperature risk percentage with the set temperature risk threshold. If the temperature risk percentage exceeds the set temperature risk threshold, the fan is determined to be overheating.

5. The axial flow fan fault early warning system based on multi-source parameter analysis according to claim 1, characterized in that, When the feature response analysis module does not trigger any warning signal, the feature fusion analysis module fuses each individual fusion parameter using a weighted algorithm to obtain a health indicator. The fault warning module compares the health indicator with a dynamic adaptive threshold. If the health indicator exceeds the dynamic adaptive threshold, the wind turbine is determined to be in a high-risk state, and a high-risk warning is generated.

6. The axial flow fan fault early warning system based on multi-source parameter analysis according to claim 1, characterized in that, The specific fusion method for individual fusion parameters in the feature fusion analysis module is as follows: Obtain the contribution of each preset single fusion parameter to the axial flow fan failure, and generate weight coefficients based on the contribution, with the sum of the weight coefficients being 1; Obtain the allowable threshold corresponding to each individual fusion parameter, and calculate the ratio of each individual fusion parameter to the corresponding allowable threshold to obtain the individual proportion. Then, weight each individual proportion using a weighting coefficient to obtain the health indicator.

7. The axial flow fan fault early warning system based on multi-source parameter analysis according to claim 1, characterized in that, The dynamic threshold adjustment module obtains the fan mode including fan speed and load parameters, and after obtaining the fan speed and load parameters, divides the speed and load parameters into multiple mode ranges. The dynamic threshold adjustment module calculates the mean and standard deviation of the health indicators of the wind turbine health status within each mode interval. Based on the mean and standard deviation of the current pattern interval, a dynamic adaptive threshold is calculated according to a preset weighting relationship.

8. The axial flow fan fault early warning system based on multi-source parameter analysis according to claim 1, characterized in that, When the fault early warning module receives an early warning signal triggered by a single fusion project, it directly generates a level-one alarm signal and outputs a reminder. When it receives a high-risk early warning, it generates a level-two early warning signal and outputs a reminder.

Images