A method for online diagnosis and positioning of a generator rotor fault

Abstract

This invention discloses an online diagnosis and location method for generator rotor faults, belonging to the field of generator fault diagnosis technology. The method includes the following steps: S1, arranging a magnetic field sensor array to synchronously acquire the time-domain signal of the air gap magnetic field during operation; S2, performing spectrum analysis to extract the amplitude of the harmonic components at twice the electrical frequency; S3, online diagnosis of either a loose pole bolt fault or a mass imbalance fault; S4, when an online diagnosis of a loose pole bolt fault is made, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor located directly below the rotor and finding the point of maximum signal strength, the mechanical angle position of the loose pole is determined. This invention, based on a magnetic field sensor array arranged along the stator circumference and axis, extracts the spatial distribution characteristics of the harmonic amplitude at twice the electrical frequency and compares it with a threshold, enabling accurate online differentiation between loose pole bolt faults and mass imbalance faults, greatly improving the accuracy of online diagnosis.

Description

Technical Field

[0001] This invention relates to the field of generator fault diagnosis technology, and in particular to a method for online diagnosis and location of generator rotor faults. Background Technology

[0002] Semi-direct drive wind turbines have become the most common type of wind turbine due to their high power density and high efficiency. Their rotors are prone to two common faults during operation: mass imbalance and loose pole bolts. Both types of faults manifest as increased vibration on a macroscopic level, but their fault mechanisms, development processes, and maintenance methods are quite different. Therefore, accurately diagnosing the fault type and location is crucial for the reliable operation of wind turbines.

[0003] In addition, for semi-direct drive permanent magnet wind turbines with higher speeds, typically 300-600 RPM, the rotor is located inside the sealed motor, and the magnetic pole bolts are subjected to centrifugal force and alternating stress, making them highly susceptible to loosening. Online and timely fault detection is an important guarantee for the safe operation of the unit.

[0004] Currently, vibration or noise signals are mainly used to detect the operating status of the unit, followed by comprehensive maintenance. The disadvantages are: insufficient targeting, limited application scenarios, and difficulty in being applied to the internal structure of the unit.

[0005] The main bottlenecks of existing technologies are: 1. The "diagnostic ambiguity" dilemma of vibration monitoring: Vibration signals are the superposition of the global vibration response of the mechanical body. Mass imbalance and loose bolts can both cause a 2x frequency increase in vibration, and the root cause cannot be distinguished based on the vibration spectrum alone.

[0006] 2. The dilemma of "spatial information deficiency" in magnetic field monitoring: Although some studies have used sparse magnetic field sensors, with 1-3 sensors, to monitor the overall harmonics, they cannot distinguish the fault type and cannot achieve accurate positioning.

[0007] 3. Lack of effective early online monitoring methods for loose magnetic pole bolts: Bolt loosening is a gradual process. In the early stage of loosening, when the preload loss is less than 50%, the vibration signal changes slightly, but the magnetic field has already undergone measurable distortion. Traditional offline inspection methods miss the best time for maintenance.

[0008] In summary, despite continuous advancements in vibration monitoring and sparse magnetic field monitoring technologies, their reliance on time- or frequency-domain analysis limits them to fundamentally address the bottlenecks of "diagnostic ambiguity" and "lack of localization." Those skilled in the art generally agree that fundamentally distinguishing between mass imbalance and bolt loosening requires acquiring spatial characteristics of the fault. However, achieving this goal online and economically within a sealed rotating motor, and establishing effective diagnostic rules, has long remained a challenge without a feasible solution.

[0009] Chinese patent application CN118226349A, published on June 21, 2024, discloses a fault diagnosis method for the eccentric magnetic field of a steam turbine generator rotor. The method includes the following steps: Step 1: Installing a magnetic field detection sensor; Step 2: Electrically connecting the magnetic field detection sensor to an external display via a PLC microcontroller; Step 3: Establishing a simulation model of the steam turbine generator and solving for the magnetic field distribution under different degrees of rotor eccentricity; Step 4: Analyzing the results, using the finite element method to analyze the electromagnetic field and obtain the radial magnetic flux density and radial electromagnetic force in the air gap under different operating conditions.

[0010] The fault diagnosis method for the eccentric magnetic field of a steam turbine generator rotor disclosed in this patent application can determine the degree of eccentricity and its position relative to the stator by analyzing the periodic changes in the air gap magnetic flux density signal. This information can be used to diagnose potential problems in the steam turbine generator and identify areas requiring maintenance or repair. However, the fault types diagnosed are relatively limited, and it cannot accurately distinguish between loose pole bolts and mass imbalance faults online, resulting in low online diagnostic accuracy. Summary of the Invention

[0011] To overcome the shortcomings of the prior art, this invention provides an online diagnosis and location method for generator rotor faults. Based on a magnetic field sensor array arranged along the stator circumference and axis, this invention extracts the spatial distribution characteristics of harmonic amplitudes at twice the electrical frequency and compares them with a threshold, which can accurately distinguish between loose magnetic pole bolts and mass imbalance faults online, greatly improving the accuracy of online diagnosis.

[0012] This invention is achieved through the following technical solution: A method for online diagnosis and location of generator rotor faults includes the following steps: S1. Arrange a magnetic field sensor array on the stator side of the generator to synchronously collect the time-domain signal of the air gap magnetic field under operating conditions; S2. Perform spectral analysis on the time-domain signal of each magnetic field sensor and extract the amplitude of the harmonic component at twice the electrical frequency. S3. Based on the amplitude of the harmonic component of 2 times the electrical frequency, analyze the spatial distribution characteristics, compare with the threshold, and diagnose online as magnetic pole bolt loosening fault or mass imbalance fault. S4. When an online diagnosis indicates a loose magnetic pole bolt, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor located directly below the rotor and finding the point of maximum signal strength, the mechanical angle position of the magnetic pole where the loose magnetic pole bolt occurred is determined.

[0013] In S1, arranging the magnetic field sensor array means arranging four magnetic field sensors along the circumference of the stator and correspondingly arranging four magnetic field sensors along the axial direction of the stator.

[0014] The magnetic field sensor array is fixed on the inner surface of the stator or the inner side of the slot wedge structure, with the sensing surface of the magnetic field sensor facing the air gap.

[0015] Four magnetic field sensors are arranged along the circumference of the stator to capture circumferential spatial distribution information, and four magnetic field sensors are arranged along the axial direction of the stator to capture axial consistency information.

[0016] The four magnetic field sensors arranged along the circumference of the stator are located at four orthogonal positions: directly above, directly below, to the left, and to the right of the rotor.

[0017] In S2, performing spectral analysis on the time-domain signal of each magnetic field sensor means performing spectral analysis through Fourier transform.

[0018] In S3, online diagnosis of a loose magnetic pole bolt fault refers to a situation where the amplitude measured by the magnetic field sensor located directly below the rotor is... The amplitude measured by the magnetic field sensor located directly above the rotor The ratio is greater than the first preset threshold. Furthermore, the relative difference in amplitude between the two axially corresponding magnetic field sensors is greater than the second preset threshold. If the online diagnosis is positive, it indicates a loose magnetic pole bolt fault.

[0019] The first preset threshold This refers to the circumferential distribution threshold, the first preset threshold. The value range is from 1.05 to 1.25.

[0020] The second preset threshold This refers to the axial difference threshold, used to detect magnetic pole tilt caused by bolt loosening; the second preset threshold. The value is 0.1.

[0021] The relative difference in amplitude between the two magnetic field sensors corresponding to the axis is calculated using Equation 1; Formula 1; in, For relative differences, The amplitude measured by the magnetic field sensor located directly above the rotor. The amplitude measured by the magnetic field sensor corresponding to the axial direction of the magnetic field sensor directly above the rotor.

[0022] In S3, online diagnosis of a mass imbalance fault refers to a situation where the amplitude measured by the magnetic field sensor located directly below the rotor is... The amplitude measured by the magnetic field sensor located directly above the rotor The ratio is less than the third preset threshold. Furthermore, the relative difference in amplitude between the two axially corresponding magnetic field sensors is less than the fourth preset threshold. ,in If so, it is diagnosed as a mass imbalance fault. S4 specifically includes: S41. Start the rotary car to control the rotor to rotate continuously at low speed; S42. During the turning process, monitor the signal strength of the magnetic field sensor located directly below the rotor; S43. When the signal strength of the magnetic field sensor reaches a preset local maximum, the current mechanical angle of the rotor is recorded by the angle encoder. ; S44, From a mechanical perspective Briefly stop the rotation and read the amplitude values ​​of the two corresponding magnetic field sensors along the axis to verify whether the relative difference is also at its maximum. If so, then adjust the mechanical angle. It directly corresponds to the faulty magnetic pole position where the magnetic pole bolt has become loose.

[0023] In S41, the rotor's low-speed continuous rotation refers to continuous rotation at a speed of 0.2 RPM-1 RPM.

[0024] It also includes the step of generating a diagnostic report, which refers to generating a report that includes the fault type, fault location, and confidence level.

[0025] The beneficial effects of this invention are mainly reflected in the following aspects: 1. Compared with the prior art, the present invention, based on a magnetic field sensor array arranged along the stator circumference and axis, extracts the spatial distribution characteristics of harmonic amplitudes of twice the electrical frequency and compares them with a threshold, which can accurately distinguish between loose magnetic pole bolts and mass imbalance faults online, greatly improving the accuracy of online diagnosis.

[0026] 2. This invention decouples and amplifies fault information that was originally homogenized by rotational motion by taking advantage of the spatial non-uniformity of the background field to form a reliable "spatial fingerprint," fundamentally solving the fundamental problem of online fault differentiation and location.

[0027] 3. The diagnostic rules of this invention are derived from clear physical rules, do not use black-box algorithms, and the results are easy to understand and verify, which helps to ensure diagnostic accuracy.

[0028] 4. This invention has low implementation cost. It only requires four magnetic field sensors to be arranged along the circumference of the stator and four magnetic field sensors to be arranged axially, for a total of eight magnetic field sensors, which reduces the installation complexity and data processing intensity.

[0029] 5. This invention enables early warning by online detection of generator rotor faults, avoiding sudden failures, and allows for planned shutdown and turning of the generator to locate faults, avoiding blind disassembly and repair.

[0030] 6. The present invention diagnoses faults by comparing data from multiple magnetic field sensors, combining the amplitude ratio of the magnetic field sensor directly below the rotor with that directly above the rotor, and the relative difference in amplitude between two corresponding magnetic field sensors along the axial direction, rather than the absolute threshold of a single signal. This approach can effectively resist electromagnetic noise interference and reduce false alarms.

[0031] 7. This invention can not only distinguish between loose magnetic pole bolts and mass imbalance faults online, but also, when a loose magnetic pole bolt fault is diagnosed, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor directly below the rotor, finding the point of maximum signal strength and combining it with the axial relative difference verification, the mechanical angle position of the faulty magnetic pole can be accurately located.

[0032] 8. In this invention, the rotor is continuously rotated at a speed of 0.2RPM-1RPM during the rotation process. This low-speed rotation operation can ensure stable signal acquisition.

[0033] 9. This invention uses an angle encoder to record the fault location. When the signal strength reaches a local maximum, the current mechanical angle of the rotor is recorded by the angle encoder. After a brief stop of the turning, the amplitude of the corresponding axial magnetic field sensor is read to verify the relative difference, which can ensure accurate fault location.

[0034] 10. This invention can generate a comprehensive diagnostic report including confidence levels, which facilitates maintenance personnel to fully understand the equipment status and make decisions.

[0035] 11. Compared to the patent document with publication number CN118226349A, which requires establishing a simulation model of the steam turbine generator and solving the magnetic field distribution under different degrees of eccentricity for offline analysis, this invention can achieve true online diagnosis and improve real-time performance by performing spectral analysis on the time-domain signal of the magnetic field sensor and extracting characteristic frequency band components.

[0036] 12. Compared with the patent document with publication number CN118226349A, which can only determine the degree of eccentricity and its position relative to the stator, but cannot accurately locate the mechanical angle position of the loose faulty magnetic pole, the present invention can achieve high-precision positioning of the faulty magnetic pole by stopping the machine and turning it, monitoring the maximum value point of the magnetic field sensor signal intensity and combining it with axial verification after diagnosing the loose magnetic pole bolt fault. Attached Figure Description

[0037] The present invention will now be further described in detail with reference to the accompanying drawings and specific embodiments, wherein: Figure 1 This is a flowchart of the online fault diagnosis process of the present invention. Detailed Implementation

[0038] Example 1 See Figure 1 A method for online diagnosis and location of generator rotor faults includes the following steps: S1. Arrange a magnetic field sensor array on the stator side of the generator to synchronously collect the time-domain signal of the air gap magnetic field under operating conditions; S2. Perform spectral analysis on the time-domain signal of each magnetic field sensor and extract the amplitude of the harmonic component at twice the electrical frequency. S3. Based on the amplitude of the harmonic component of 2 times the electrical frequency, analyze the spatial distribution characteristics, compare with the threshold, and diagnose online as magnetic pole bolt loosening fault or mass imbalance fault. S4. When an online diagnosis indicates a loose magnetic pole bolt, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor located directly below the rotor and finding the point of maximum signal strength, the mechanical angle position of the magnetic pole where the loose magnetic pole bolt occurred is determined.

[0039] This embodiment is the most basic implementation method. Compared with the prior art, based on the magnetic field sensor array arranged along the stator circumference and axis, the spatial distribution characteristics of the harmonic amplitude of twice the electrical frequency are extracted and compared with the threshold. This can accurately distinguish between loose magnetic pole bolts and mass imbalance faults online, greatly improving the accuracy of online diagnosis.

[0040] Example 2 See Figure 1 A method for online diagnosis and location of generator rotor faults includes the following steps: S1. Arrange a magnetic field sensor array on the stator side of the generator to synchronously collect the time-domain signal of the air gap magnetic field under operating conditions; S2. Perform spectral analysis on the time-domain signal of each magnetic field sensor and extract the amplitude of the harmonic component at twice the electrical frequency. S3. Based on the amplitude of the harmonic component of 2 times the electrical frequency, analyze the spatial distribution characteristics, compare with the threshold, and diagnose online as magnetic pole bolt loosening fault or mass imbalance fault. S4. When an online diagnosis indicates a loose magnetic pole bolt, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor located directly below the rotor and finding the point of maximum signal strength, the mechanical angle position of the magnetic pole where the loose magnetic pole bolt occurred is determined.

[0041] Preferably, in S1, arranging the magnetic field sensor array means arranging four magnetic field sensors along the circumferential direction of the stator and correspondingly arranging four magnetic field sensors along the axial direction of the stator.

[0042] The magnetic field sensor array is fixed on the inner surface of the stator, with the sensing surface of the magnetic field sensor facing the air gap.

[0043] Four magnetic field sensors are arranged along the circumference of the stator to capture circumferential spatial distribution information, and four magnetic field sensors are arranged along the axial direction of the stator to capture axial consistency information.

[0044] The four magnetic field sensors arranged along the circumference of the stator are located at four orthogonal positions: directly above, directly below, to the left, and to the right of the rotor.

[0045] This embodiment is a preferred implementation method. By decoupling and amplifying the fault information that was originally homogenized by rotational motion, the spatial non-uniformity of the background field is used to form a reliable "spatial fingerprint", which fundamentally solves the principle problem of online fault differentiation and location.

[0046] The diagnostic rules are derived from clear physical rules, do not employ black-box algorithms, and the results are easy to understand and verify, which helps ensure diagnostic accuracy.

[0047] Example 3 See Figure 1 A method for online diagnosis and location of generator rotor faults includes the following steps: S1. Arrange a magnetic field sensor array on the stator side of the generator to synchronously collect the time-domain signal of the air gap magnetic field under operating conditions; S2. Perform spectral analysis on the time-domain signal of each magnetic field sensor and extract the amplitude of the harmonic component at twice the electrical frequency. S3. Based on the amplitude of the harmonic component of 2 times the electrical frequency, analyze the spatial distribution characteristics, compare with the threshold, and diagnose online as magnetic pole bolt loosening fault or mass imbalance fault. S4. When an online diagnosis indicates a loose magnetic pole bolt, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor located directly below the rotor and finding the point of maximum signal strength, the mechanical angle position of the magnetic pole where the loose magnetic pole bolt occurred is determined.

[0048] In S1, arranging the magnetic field sensor array means arranging four magnetic field sensors along the circumference of the stator and correspondingly arranging four magnetic field sensors along the axial direction of the stator.

[0049] The magnetic field sensor array is fixed on the inner surface of the stator, with the sensing surface of the magnetic field sensor facing the air gap.

[0050] Four magnetic field sensors are arranged along the circumference of the stator to capture circumferential spatial distribution information, and four magnetic field sensors are arranged along the axial direction of the stator to capture axial consistency information.

[0051] The four magnetic field sensors arranged along the circumference of the stator are located at four orthogonal positions: directly above, directly below, to the left, and to the right of the rotor.

[0052] In S2, performing spectral analysis on the time-domain signal of each magnetic field sensor means performing spectral analysis through Fourier transform.

[0053] In S3, online diagnosis of a loose magnetic pole bolt fault refers to a situation where the amplitude measured by the magnetic field sensor located directly below the rotor is... The amplitude measured by the magnetic field sensor located directly above the rotor The ratio is greater than the first preset threshold. Furthermore, the relative difference in amplitude between the two axially corresponding magnetic field sensors is greater than the second preset threshold. If the online diagnosis is positive, it indicates a loose magnetic pole bolt fault.

[0054] This embodiment is a preferred implementation method with low implementation cost. It only requires four magnetic field sensors to be arranged along the circumference of the stator and four magnetic field sensors to be arranged axially, for a total of eight magnetic field sensors, which reduces the installation complexity and data processing intensity.

[0055] Online detection of generator rotor faults enables early warning, avoids sudden failures, and allows for planned shutdowns and manual rotation to locate faults, avoiding blind disassembly and repair.

[0056] Example 4 See Figure 1 A method for online diagnosis and location of generator rotor faults includes the following steps: S1. Arrange a magnetic field sensor array on the stator side of the generator to synchronously collect the time-domain signal of the air gap magnetic field under operating conditions; S2. Perform spectral analysis on the time-domain signal of each magnetic field sensor and extract the amplitude of the harmonic component at twice the electrical frequency. S3. Based on the amplitude of the harmonic component of 2 times the electrical frequency, analyze the spatial distribution characteristics, compare with the threshold, and diagnose online as magnetic pole bolt loosening fault or mass imbalance fault. S4. When an online diagnosis indicates a loose magnetic pole bolt, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor located directly below the rotor and finding the point of maximum signal strength, the mechanical angle position of the magnetic pole where the loose magnetic pole bolt occurred is determined.

[0057] In S1, arranging the magnetic field sensor array means arranging four magnetic field sensors along the circumference of the stator and correspondingly arranging four magnetic field sensors along the axial direction of the stator.

[0058] The magnetic field sensor array is fixed on the inner side of the slot wedge structure, with the sensing surface of the magnetic field sensor facing the air gap.

[0059] Four magnetic field sensors are arranged along the circumference of the stator to capture circumferential spatial distribution information, and four magnetic field sensors are arranged along the axial direction of the stator to capture axial consistency information.

[0060] The four magnetic field sensors arranged along the circumference of the stator are located at four orthogonal positions: directly above, directly below, to the left, and to the right of the rotor.

[0061] In S2, performing spectral analysis on the time-domain signal of each magnetic field sensor means performing spectral analysis through Fourier transform.

[0062] More preferably, in step S3, online diagnosis of a loose magnetic pole bolt fault refers to the situation where the amplitude measured by the magnetic field sensor located directly below the rotor is... The amplitude measured by the magnetic field sensor located directly above the rotor The ratio is greater than the first preset threshold. Furthermore, the relative difference in amplitude between the two axially corresponding magnetic field sensors is greater than the second preset threshold. If the online diagnosis is positive, it indicates a loose magnetic pole bolt fault.

[0063] The first preset threshold This refers to the circumferential distribution threshold, the first preset threshold. The value range is from 1.05 to 1.25.

[0064] The second preset threshold This refers to the axial difference threshold, used to detect magnetic pole tilt caused by bolt loosening; the second preset threshold. The value is 0.1.

[0065] The relative difference in amplitude between the two magnetic field sensors corresponding to the axis is calculated using Equation 1; Formula 1; in, For relative differences, The amplitude measured by the magnetic field sensor located directly above the rotor. The amplitude measured by the magnetic field sensor corresponding to the axial direction of the magnetic field sensor directly above the rotor.

[0066] In S3, online diagnosis of a mass imbalance fault refers to a situation where the amplitude measured by the magnetic field sensor located directly below the rotor is... The amplitude measured by the magnetic field sensor located directly above the rotor The ratio is less than the third preset threshold. Furthermore, the relative difference in amplitude between the two axially corresponding magnetic field sensors is less than the fourth preset threshold. ,in If so, it is diagnosed as a mass imbalance fault. This embodiment is a preferred implementation. Fault diagnosis is based on the comparison of data from multiple magnetic field sensors, combining the amplitude ratio of the magnetic field sensor directly below the rotor with that directly above the rotor, and the relative difference in amplitude between two corresponding magnetic field sensors along the axial direction, rather than the absolute threshold of a single signal. This effectively resists electromagnetic noise interference and reduces false alarms. It can not only distinguish between loose magnetic pole bolts and mass imbalance faults online, but also, when a loose magnetic pole bolt fault is diagnosed, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor directly below the rotor, finding the point of maximum signal strength, and verifying it in conjunction with the relative difference along the axial direction, the mechanical angular position of the faulty magnetic pole can be accurately located.

[0067] Example 5 See Figure 1 A method for online diagnosis and location of generator rotor faults includes the following steps: S1. Arrange a magnetic field sensor array on the stator side of the generator to synchronously collect the time-domain signal of the air gap magnetic field under operating conditions; S2. Perform spectral analysis on the time-domain signal of each magnetic field sensor and extract the amplitude of the harmonic component at twice the electrical frequency. S3. Based on the amplitude of the harmonic component of 2 times the electrical frequency, analyze the spatial distribution characteristics, compare with the threshold, and diagnose online as magnetic pole bolt loosening fault or mass imbalance fault. S4. When an online diagnosis indicates a loose magnetic pole bolt, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor located directly below the rotor and finding the point of maximum signal strength, the mechanical angle position of the magnetic pole where the loose magnetic pole bolt occurred is determined.

[0068] In S1, arranging the magnetic field sensor array means arranging four magnetic field sensors along the circumference of the stator and correspondingly arranging four magnetic field sensors along the axial direction of the stator.

[0069] The magnetic field sensor array is fixed on the inner surface of the stator, with the sensing surface of the magnetic field sensor facing the air gap.

[0070] Four magnetic field sensors are arranged along the circumference of the stator to capture circumferential spatial distribution information, and four magnetic field sensors are arranged along the axial direction of the stator to capture axial consistency information.

[0071] The four magnetic field sensors arranged along the circumference of the stator are located at four orthogonal positions: directly above, directly below, to the left, and to the right of the rotor.

[0072] In S2, performing spectral analysis on the time-domain signal of each magnetic field sensor means performing spectral analysis through Fourier transform.

[0073] In S3, online diagnosis of a loose magnetic pole bolt fault refers to a situation where the amplitude measured by the magnetic field sensor located directly below the rotor is... The amplitude measured by the magnetic field sensor located directly above the rotor The ratio is greater than the first preset threshold. Furthermore, the relative difference in amplitude between the two axially corresponding magnetic field sensors is greater than the second preset threshold. If the online diagnosis is positive, it indicates a loose magnetic pole bolt fault.

[0074] The first preset threshold This refers to the circumferential distribution threshold, the first preset threshold. The value range is from 1.05 to 1.25.

[0075] The second preset threshold This refers to the axial difference threshold, used to detect magnetic pole tilt caused by bolt loosening; the second preset threshold. The value is 0.1.

[0076] The relative difference in amplitude between the two magnetic field sensors corresponding to the axis is calculated using Equation 1; Formula 1; in, For relative differences, The amplitude measured by the magnetic field sensor located directly above the rotor. The amplitude measured by the magnetic field sensor corresponding to the axial direction of the magnetic field sensor directly above the rotor.

[0077] In S3, online diagnosis of a mass imbalance fault refers to a situation where the amplitude measured by the magnetic field sensor located directly below the rotor is... The amplitude measured by the magnetic field sensor located directly above the rotor The ratio is less than the third preset threshold. Furthermore, the relative difference in amplitude between the two axially corresponding magnetic field sensors is less than the fourth preset threshold. ,in If so, it is diagnosed as a mass imbalance fault. S4 specifically includes: S41. Start the rotary car to control the rotor to rotate continuously at low speed; S42. During the turning process, monitor the signal strength of the magnetic field sensor located directly below the rotor; S43. When the signal strength of the magnetic field sensor reaches a preset local maximum, the current mechanical angle of the rotor is recorded by the angle encoder. ; S44, From a mechanical perspective Briefly stop the rotation and read the amplitude values ​​of the two corresponding magnetic field sensors along the axis to verify whether the relative difference is also at its maximum. If so, then adjust the mechanical angle. It directly corresponds to the faulty magnetic pole position where the magnetic pole bolt has become loose.

[0078] In S41, the rotor's low-speed continuous rotation refers to continuous rotation at a speed of 0.2 RPM.

[0079] It also includes the step of generating a diagnostic report, which refers to generating a report that includes the fault type, fault location, and confidence level.

[0080] This embodiment is the best implementation method. It uses an angle encoder to record the fault location. When the signal strength reaches a local maximum, the current mechanical angle of the rotor is recorded by the angle encoder. After a brief stop of the turning, the amplitude of the corresponding axial magnetic field sensor is read to verify the relative difference, which can ensure accurate fault location.

[0081] It can generate comprehensive diagnostic reports including confidence levels, making it easier for maintenance personnel to fully understand the equipment status and make decisions.

[0082] The basic principle of this invention is as follows: By utilizing the specific modulation effect of rotor faults in permanent magnet wind turbines on the air gap magnetic field, and by using a magnetic field sensor array arranged in a specific spatial location, the amplitude distribution characteristics of the harmonic components at twice the electrical frequency are extracted, thereby enabling online diagnosis, classification, and location of the fault.

[0083] First, theoretically, the rotor's inherent static eccentricity and the rotor's elastic bending caused by gravity result in a non-uniform distribution of the air gap in the circumferential direction, with the lower air gap being smaller than the upper one. When the magnetic pole bolts loosen, the loose magnetic poles are mainly subjected to centrifugal force, producing an approximately constant radial outward displacement. Since the air gap magnetic induction intensity is approximately inversely proportional to the air gap length, this constant displacement applied to the non-uniform background air gap leads to different rates of change of the magnetic field at different circumferential positions: below the smaller air gap, i.e., directly below the rotor, the relative change of the magnetic field is greater than that above the rotor, thus forming a circumferentially asymmetrical magnetic field distribution.

[0084] Secondly, both mass imbalance and loose pole bolts cause the air gap length to change periodically over time, thus modulating the air gap permeability. The product of the permanent magnet magnetomotive force and the air gap permeability determines the air gap magnetic flux density. When the modulation function includes the mechanical rotation frequency, a sideband frequency is generated in the magnetic flux density through trigonometric function product-sum-difference. For permanent magnet motors with a large number of pole pairs, this sideband frequency is very close to twice the electrical frequency in the spectrum. Combined with resolution limitations and energy leakage in spectrum analysis, this significantly enhances the harmonic components of the twice electrical frequency. Therefore, the twice electrical frequency was selected as the characteristic frequency for monitoring rotor dynamic eccentricity faults.

[0085] In terms of specific working principles, the two failure modes exhibit different spatial distribution characteristics: Mass imbalance is a synchronous dynamic eccentricity of the rotor as a whole. The amplitude of the harmonic component of the second electrical frequency caused by it is stronger at the top and weaker at the bottom in the circumferential direction. That is, the amplitude of the magnetic field sensor directly above the rotor is greater than the amplitude of the magnetic field sensor directly below the rotor in a fixed spatial distribution pattern, and the signal heights at both ends of the axis are consistent.

[0086] Loose magnetic pole bolts can cause a decrease in local connection stiffness, which can easily lead to axial tilting of the magnetic pole or uneven force at both ends, resulting in a significant difference in the magnetic field signal at both ends of the axis. At the same time, due to the inherent distribution of the background air gap being weaker at the top and stronger at the bottom, the magnetic field change caused by the loose magnetic pole rotating to the bottom is greater, making the amplitude of the 2nd electrical frequency harmonic component of the lower magnetic field sensor more prominent than that of the upper one.

[0087] When the online diagnosis indicates that the magnetic pole bolt is loose, a shutdown and turning operation is performed. The rotor is continuously rotated at a speed of 0.2 RPM-1 RPM. During the turning operation, the signal strength of the magnetic field sensor located directly below the rotor is continuously monitored. When the signal strength reaches a preset local maximum value, the current mechanical angle of the rotor is recorded by the angle encoder. At this time, the turning is briefly stopped, and the amplitude values ​​of the two corresponding magnetic field sensors in the axial direction are read and their relative difference is verified to be at its maximum. This is because the air gap magnetic field distortion caused by the loose bolt is most significant when the faulty magnetic pole passes directly below it. Thus, the mechanical angle position of the magnetic pole where the bolt is loose can be directly determined through this maximum value point.

Claims

1. A method for online diagnosis and location of generator rotor faults, characterized in that, Includes the following steps: S1. Arrange a magnetic field sensor array on the stator side of the generator to synchronously collect the time-domain signal of the air gap magnetic field under operating conditions; S2. Perform spectral analysis on the time-domain signal of each magnetic field sensor and extract the amplitude of the harmonic component at twice the electrical frequency. S3. Based on the amplitude of the harmonic component of 2 times the electrical frequency, analyze the spatial distribution characteristics, compare with the threshold, and diagnose online as magnetic pole bolt loosening fault or mass imbalance fault. S4. When an online diagnosis indicates a loose magnetic pole bolt, a shutdown and turning operation is performed. By monitoring the signal strength of the magnetic field sensor located directly below the rotor and finding the point of maximum signal strength, the mechanical angle position of the magnetic pole where the loose magnetic pole bolt occurred is determined.

2. The method for online diagnosis and location of generator rotor faults according to claim 1, characterized in that: In S1, arranging the magnetic field sensor array means arranging four magnetic field sensors along the circumference of the stator and correspondingly arranging four magnetic field sensors along the axial direction of the stator.

3. The method for online diagnosis and location of generator rotor faults according to claim 1, characterized in that: The magnetic field sensor array is fixed on the inner surface of the stator or the inner side of the slot wedge structure, with the sensing surface of the magnetic field sensor facing the air gap.

4. The method for online diagnosis and location of generator rotor faults according to claim 2, characterized in that: Four magnetic field sensors are arranged along the circumference of the stator to capture circumferential spatial distribution information, and four magnetic field sensors are arranged along the axial direction of the stator to capture axial consistency information.

5. The method for online diagnosis and location of generator rotor faults according to claim 2, characterized in that: The four magnetic field sensors arranged along the circumference of the stator are located at four orthogonal positions: directly above, directly below, to the left, and to the right of the rotor.

6. The method for online diagnosis and location of generator rotor faults according to claim 1, characterized in that: In S2, performing spectral analysis on the time-domain signal of each magnetic field sensor means performing spectral analysis through Fourier transform.

7. The method for online diagnosis and location of generator rotor faults according to claim 1, characterized in that: In S3, online diagnosis of a loose magnetic pole bolt fault refers to a situation where the amplitude measured by the magnetic field sensor located directly below the rotor is... The amplitude measured by the magnetic field sensor located directly above the rotor The ratio is greater than the first preset threshold. Furthermore, the relative difference in amplitude between the two axially corresponding magnetic field sensors is greater than the second preset threshold. If the online diagnosis is positive, it indicates a loose magnetic pole bolt fault.

8. The method for online diagnosis and location of generator rotor faults according to claim 7, characterized in that: The first preset threshold This refers to the circumferential distribution threshold, the first preset threshold. The value range is from 1.05 to 1.

25.

9. The method for online diagnosis and location of generator rotor faults according to claim 7, characterized in that: The second preset threshold This refers to the axial difference threshold, used to detect magnetic pole tilt caused by bolt loosening; the second preset threshold. The value is 0.

1.

10. The method for online diagnosis and location of generator rotor faults according to claim 7, characterized in that: The relative difference in amplitude between the two magnetic field sensors corresponding to the axis is calculated using Equation 1; Formula 1; in, For relative differences, The amplitude measured by the magnetic field sensor located directly above the rotor. The amplitude measured by the magnetic field sensor corresponding to the axial direction of the magnetic field sensor directly above the rotor.

11. The method for online diagnosis and location of generator rotor faults according to claim 7, characterized in that: In S3, online diagnosis of a mass imbalance fault refers to a situation where the amplitude measured by the magnetic field sensor located directly below the rotor is... The amplitude measured by the magnetic field sensor located directly above the rotor The ratio is less than the third preset threshold. Furthermore, the relative difference in amplitude between the two axially corresponding magnetic field sensors is less than the fourth preset threshold. ,in If so, it is diagnosed as a mass imbalance fault.

12. The method for online diagnosis and location of generator rotor faults according to claim 1, characterized in that: S4 specifically includes: S41. Start the rotary car to control the rotor to rotate continuously at low speed; S42. During the turning process, monitor the signal strength of the magnetic field sensor located directly below the rotor; S43. When the signal strength of the magnetic field sensor reaches a preset local maximum, the current mechanical angle of the rotor is recorded by the angle encoder. ; S44, From a mechanical perspective Briefly stop the rotation and read the amplitude values ​​of the two corresponding magnetic field sensors along the axis to verify whether the relative difference is also at its maximum. If so, then adjust the mechanical angle. It directly corresponds to the faulty magnetic pole position where the magnetic pole bolt has become loose.

13. The method for online diagnosis and location of generator rotor faults according to claim 12, characterized in that: In S41, the rotor's low-speed continuous rotation refers to continuous rotation at a speed of 0.2 RPM-1 RPM.

14. The method for online diagnosis and location of generator rotor faults according to claim 1, characterized in that: It also includes the step of generating a diagnostic report, which refers to generating a report that includes the fault type, fault location, and confidence level.

Images