An on-line detection system for operation state of a coal-fired boiler

Abstract

This invention relates to the field of coal-fired boiler operation monitoring technology, specifically to an online monitoring system for the operating status of coal-fired boilers. By constructing a rotor shaft trajectory and calculating its ellipticity and irregularity, this invention achieves a quantitative description of the geometric characteristics of vibration faults, overcoming the limitations of relying solely on vibration amplitude or frequency components for diagnosis. Based on trajectory ellipticity, trajectory irregularity, and preset intervals, this invention initiates operating condition correlation analysis only when the rotor shaft trajectory exhibits typical fault characteristics. This reduces the computational load on the system while ensuring diagnostic accuracy, thereby improving the relevance and efficiency of the diagnosis.

Description

Technical Field

[0001] The present invention relates to the technical field of operation monitoring of coal-fired boilers, and particularly to an online detection system for the operation state of coal-fired boilers. Background Art

[0002] The steam turbine supporting the coal-fired boiler is the core power equipment, and the stable operation of its bearing rotor system is crucial. Oil film whirl and steam flow-induced vibration are two common and serious faults, and their early vibration characteristics may be similar in traditional spectrum analysis, both showing vibration components close to half-frequency or multiple frequencies, making it difficult to accurately distinguish.

[0003] Existing online monitoring systems mostly rely on vibration amplitude overrun alarms or simple spectrum analysis, lacking in-depth quantification of the geometric shape of the rotor center motion trajectory and failing to effectively combine the changes in the unit operation conditions for associated diagnosis. This is prone to false alarms or missed alarms and cannot provide specific fault type judgments, which is not conducive to accurate intervention by operators. Summary of the Invention

[0004] The purpose of the present invention is to overcome the defects of the existing technology and solve the problems of low fault recognition accuracy and high false alarm rate in the existing diagnostic system when distinguishing oil film whirl and steam flow-induced vibration due to the lack of quantitative characterization of the geometric shape of the rotor center trajectory and the dynamic correlation analysis of vibration response and operation conditions.

[0005] The technical solution adopted by the present invention to solve its technical problems is: an online detection system for the operation state of a coal-fired boiler, including: a signal acquisition module for synchronously acquiring vibration time series signals in two orthogonal directions of the X-axis and the Y-axis through vibration sensors installed at the bearings of the steam turbine supporting the boiler.

[0006] A trajectory construction module for constructing a rotor center trajectory curve based on the vibration time series signals and calculating the vibration main frequency amplitude sequence.

[0007] A feature quantification module for calculating the trajectory ellipticity and trajectory irregularity based on the rotor center trajectory curve.

[0008] A working condition correlation module for synchronously obtaining the real-time power data of the steam turbine when the trajectory ellipticity and trajectory irregularity jointly enter a preset interval, aligning the vibration main frequency amplitude sequence and the real-time power change sequence of the steam turbine in time; using the sliding time window method to calculate the time-varying Pearson correlation coefficient of the vibration main frequency amplitude sequence and the real-time power change sequence of the steam turbine in each window to obtain a correlation coefficient sequence; extracting the maximum value in the correlation coefficient sequence as the dominant correlation intensity between the vibration main frequency amplitude sequence and the power change sequence.

[0009] The fault diagnosis module is used to form a current feature vector by combining trajectory ellipticity, trajectory irregularity, and dominant correlation strength, compare it with a preset reference vector, and then determine whether it is an oil film whirl fault or a steam flow excitation fault based on the comparison result, and output the corresponding diagnosis result.

[0010] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention realizes the quantitative description of the geometric morphological characteristics of vibration faults by constructing the rotor shaft center trajectory and calculating the trajectory ellipticity and irregularity, thus overcoming the limitations of relying solely on vibration amplitude or frequency components for diagnosis.

[0011] 2. Based on trajectory ellipticity, trajectory irregularity and preset interval, this invention initiates working condition correlation analysis only when the rotor shaft trajectory shape exhibits typical fault characteristics, thereby reducing the computational load of the system and improving the pertinence and efficiency of diagnosis while ensuring diagnostic accuracy.

[0012] 3. This invention achieves automatic quantitative differentiation between oil film eddy and steam flow excitation faults by fusing trajectory ellipticity, trajectory irregularity, and vibration-power correlation intensity into a feature vector and comparing it with a preset reference vector, thereby improving the reliability of the diagnostic results. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a schematic diagram of the system module connections of the present invention.

[0015] Figure 2 This is a schematic diagram of the process for calculating the ellipticity of a trajectory according to the present invention.

[0016] Figure 3 This is a schematic diagram of the process for calculating trajectory irregularity according to the present invention.

[0017] Figure 4 This is a schematic diagram of the process by which the present invention determines whether the fault is an oil film whirl or a steam flow excitation fault and outputs the corresponding diagnostic results. Detailed Implementation

[0018] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention. Furthermore, it should be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale.

[0019] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and apparatus should be considered part of the specification.

[0020] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0022] The following description, in conjunction with the accompanying drawings, details a specific scheme for an online monitoring system for the operating status of a coal-fired boiler provided by the present invention.

[0023] Please see Figure 1 The diagram shows a module connection diagram of an online monitoring system for the operating status of a coal-fired boiler provided by the present invention, which specifically includes: a signal acquisition module, a trajectory construction module, a feature quantization module, an operating condition association module, and a fault diagnosis module.

[0024] The output of the signal acquisition module is connected to the trajectory construction module, the output of the trajectory construction module is connected to the feature quantization module and the operating condition association module respectively, and the outputs of the feature quantization module and the operating condition association module are connected to the fault diagnosis module.

[0025] The operation process of the signal acquisition module is as follows: Step 1: Determine the azimuth angle of the maximum load zone based on the geometric parameters of the target bearing of the steam turbine matched with the boiler.

[0026] The geometric parameters include bearing type, bearing bush splitting surface position, oil wedge starting angle, bearing top clearance and left and right side clearance.

[0027] Among them, the bearing shell split surface refers to the assembly mating surface of the sliding bearing. The oil wedge initiation angle refers to the circumferential angle corresponding to the initial position where lubricating oil enters from the oil supply groove between the bearing shell and the journal to form a hydrodynamic oil film. The top clearance and the left and right side clearances directly reflect the degree of eccentricity of the journal in a static state.

[0028] Geometric parameters can usually be obtained from technical drawings provided by the equipment manufacturer or from on-site measurements.

[0029] The specific determination process is as follows: take the center line of the bearing bush split surface as the zero-degree reference line, and then define the positive angle along the rotor rotation direction, that is, specify the rotor clockwise or counterclockwise as the direction of angle increase, thereby establishing the bearing section coordinate system.

[0030] Next, by consulting the bearing design manual, the number of bearing-bearing oil wedges and their corresponding circumferential positions can be determined according to the bearing type.

[0031] For example, for elliptical bearings, there are usually two symmetrical oil wedges, located in the upper and lower halves respectively; for cylindrical bearings with two oil wedges, the bearing oil wedges are usually located in the range of about 70 to 110 degrees below the bearing shell split surface.

[0032] Next, calculate half the difference between the left and right clearances to obtain the horizontal component of the journal center relative to the bearing center. Then, calculate half the difference between the bearing inner diameter and the journal diameter, and subtract the top clearance to obtain the vertical component of the journal center relative to the bearing center.

[0033] Then, based on the components in the horizontal and vertical directions, the static eccentricity direction angle of the journal is calculated using the arctangent function in the four quadrants. The direction indicated by the static eccentricity direction angle is the static eccentricity direction.

[0034] The static eccentricity direction indicates the stable offset direction of the journal relative to the bearing center under the action of gravity and possible preload. The bearing oil wedge is usually located near this direction.

[0035] In order to focus on the main load-bearing area, with the static eccentric direction as the center, the rotor is extended 90 degrees to both the counterclockwise and clockwise sides to form an eccentric semicircle.

[0036] The eccentric semicircle is defined in this way because in sliding bearings, the area where an effective hydrodynamic oil film is formed and the main load-bearing capacity is generated is usually concentrated within a range of about 180 degrees near the static eccentric direction, rather than the entire circumference.

[0037] Furthermore, within the eccentric semicircle, several points are discretely sampled along the circumference, and the local gap at each point is calculated. The specific calculation formula is as follows: .

[0038] Where H represents the local gap. The average clearance is half the difference between the bearing inner diameter and the journal diameter, which can be obtained from the turbine design drawings. e is the eccentricity, which is the offset distance of the journal center relative to the bearing center. φ is the current sampling point; θ is the static eccentricity direction angle.

[0039] The adjacent circumferential angles are calculated sequentially along the rotor rotation direction. and Local gap at the location and ,like Less than This indicates that the radial clearance decreases within that angle range; identify the oil wedge whose radial clearance between the inner surface of the bearing bush and the journal gradually decreases along the rotor rotation direction, and use it as the main bearing oil wedge.

[0040] If multiple main load-bearing wedges exist, the one with the smallest angle between its geometric centerline and the vertical centerline of the bearing housing bottom surface is selected as the final main load-bearing wedge. The area below the bearing's vertical centerline is typically the region of maximum static load. The main load-bearing wedge with the smallest angle between its geometric centerline and the vertical centerline is closest to the bearing's vertical centerline and bears a higher proportion of the static load.

[0041] Finally, in the bearing section coordinate system, with the center line of the bearing bush split surface as the zero-degree reference and the counterclockwise direction as positive, measure the angle corresponding to the geometric center line of the main bearing oil wedge, which is the azimuth angle of the maximum load area.

[0042] Step 2: Deploy vibration sensors based on the azimuth angle of the maximum load area and collect vibration time-series signals.

[0043] Specifically, since the maximum load area is the region where the bearing is most prone to oil film instability or abnormal structural response during steady-state operation, the direction indicated by the azimuth angle of the maximum load area is taken as the X-axis, and its normal direction is taken as the Y-axis, to establish the sensor installation coordinate system.

[0044] Then, vibration sensors are arranged on the radial projection area of ​​the outer surface of the bearing housing corresponding to the rotor shaft center. The radial projection area can effectively capture the relative displacement of the rotor shaft center within the bearing clearance.

[0045] Furthermore, vibration sensors are symmetrically arranged along the X and Y axes. This symmetrical and oriented arrangement accurately captures the vibration components of the rotor shaft in the maximum load zone and its perpendicular direction.

[0046] After the vibration sensors are installed, during the operation of the steam turbine, the timing signals output by all vibration sensors are collected synchronously and used as vibration timing signals in the X-axis and Y-axis directions, respectively.

[0047] The trajectory construction module is used to construct the rotor shaft center trajectory curve based on the vibration timing signal and calculate the vibration dominant frequency amplitude sequence.

[0048] Before constructing the rotor shaft trajectory curve based on the vibration timing signal, it is usually necessary to preprocess the original X-axis and Y-axis vibration timing signals. For example, bandpass filtering is performed to retain the frequency band related to the rotor operating frequency while suppressing high-frequency noise and low-frequency drift. The specific preprocessing process is existing technology and will not be described in detail in this invention.

[0049] During the formal construction, the vibration time sequence signals of the X-axis and Y-axis are first used as displacement components of the rotor shaft center on the corresponding coordinate axes of the sensor installation coordinate system; and the two displacement components are paired to form a two-dimensional coordinate point sequence using the same sampling time as an index.

[0050] To ensure the temporal continuity and reliability of the trajectory, the temporal continuity of the two-dimensional coordinate point sequence can be checked by determining whether the Euclidean distance between adjacent points exceeds the displacement increment estimated based on the velocity at the previous moment. If it exceeds, it is regarded as an abnormal jump point caused by signal interference or data packet loss, and the abnormal jump point is removed.

[0051] Next, based on the time required for one rotor rotation, obtained directly from the key phase signal (i.e., the rotor rotation period), linear interpolation is preferentially used to resample the two-dimensional coordinate point sequence with a fixed angular increment. This yields a trajectory point sequence with the rotor's circumferential rotation angle as the independent variable. The trajectory points are uniformly distributed according to the rotor's rotation angle, and each trajectory point corresponds to a specific rotation phase of the rotor. This eliminates the uneven distribution of trajectory point angles caused by slight fluctuations in rotational speed, making the trajectory shape analysis more stable.

[0052] Then, all trajectory points are connected sequentially according to the rotation phase from zero degrees to 360 degrees to form the rotor shaft trajectory curve.

[0053] Simultaneously, short-time Fourier transforms (SFTs) are performed on the X-axis and Y-axis vibration time-series signals. The SFT involves windowing the signal into frames and performing a Fourier transform on each frame to obtain two sets of vibration spectra. The Hanning window is a preferred window function, and its length needs to be set based on the rotor frequency and the required frequency resolution. The ratio of the frequency resolution to the window length is of interest.

[0054] For example, if the required frequency resolution is 2Hz, the theoretical window length should be at least 0.5 seconds. Meanwhile, to avoid introducing spurious frequency components due to truncating non-integer cycle signals, the actual window length should cover as many integer rotor rotation cycles as possible. If the rotor frequency is 25Hz, meaning one rotor cycle is 0.04 seconds, a theoretical window length of 0.5 seconds corresponds to approximately 12.5 rotor cycles. In this case, the window length should be adjusted to 13 complete cycles, or 0.52 seconds.

[0055] Then, for each identical analysis time window, the amplitudes of the same frequency points in the two sets of vibration spectra are vector-synthesized to obtain the comprehensive vibration spectrum.

[0056] Subsequently, all frequency points in the comprehensive vibration spectrum whose frequency values ​​fall within the preset octave band of the rotor's operating frequency are selected as candidate points.

[0057] The preset octave band of the rotor operating frequency is usually set to cover the frequency range where oil film whirl and steam flow excitation may occur. The oil film whirl frequency is usually 0.4 to 0.5 times the rotor operating frequency, while the steam flow excitation frequency is mostly 0.5 to 1.0 times or 1.5 to 2.0 times the rotor operating frequency.

[0058] In this invention, 0.3 to 2.5 times the rotor operating frequency is exemplarily used as the preset octave band. If the system is mainly used to monitor the low-speed heavy-load start-up process, the lower limit can be expanded to 0.2 times; if high-frequency whirl is of concern, the upper limit can be expanded to 3.0 times.

[0059] Next, the largest amplitude is selected from the candidate points as the dominant vibration frequency, and its frequency and amplitude are recorded; and the amplitudes are arranged in the order of the analysis time window to form a dominant vibration frequency amplitude sequence.

[0060] In cases where the spectrum is too flat or the signal-to-noise ratio is extremely low, candidate points may not be found. In such cases, the amplitude of the dominant vibration frequency can be set to zero or a preset baseline value.

[0061] The baseline value can be determined statistically based on the background noise level of the comprehensive vibration spectrum in historical normal operation data, for example, by taking the average of the lowest amplitude level in multiple measurements.

[0062] The feature quantization module is used to calculate the trajectory ellipticity and trajectory irregularity based on the rotor shaft center trajectory curve.

[0063] After completing the construction of the rotor shaft trajectory curve and the extraction of the vibration dominant frequency sequence, in order to further quantify the degree of abnormality in the rotor shaft trajectory morphology, this invention introduces two key indicators: trajectory ellipticity and trajectory irregularity.

[0064] Please see Figure 2 Step 1: Calculate the ellipticity of the trajectory. Specifically, perform least-squares ellipse fitting on all trajectory points on the rotor shaft center trajectory curve to obtain the fitted ellipse.

[0065] If the fitting fails, the calculation of trajectory ellipticity and trajectory irregularity for this period is abandoned, and feature quantization for this period ends. Otherwise, the center coordinates, semi-major axis length, and major axis direction of the fitted ellipse are extracted.

[0066] To eliminate the influence of installation angle deviation on ellipticity calculation, the coordinates of all trajectory points are rotated around the center of the ellipse. The rotation angle is the angle between the major axis of the fitted ellipse and the horizontal axis of the sensor installation coordinate system. This aligns the direction of the major axis of the fitted ellipse with the direction of the horizontal axis of the sensor installation coordinate system.

[0067] In the rotated coordinate system, calculate the Euclidean distance from each trajectory point to the center coordinate of the fitted ellipse, and normalize it based on the length of the major semi-axis to obtain the normalized radial value; then calculate the variance of all normalized radial values ​​over a complete rotation cycle, add one to the variance, and the result is defined as the trajectory ellipticity; the trajectory ellipticity is dimensionless.

[0068] The complete rotation cycle refers to all the trajectory points corresponding to one rotation of the rotor.

[0069] The first addition is to prevent the ellipticity from being zero when the variance is zero, ensuring that the ellipticity is always greater than or equal to 1. The closer the ellipticity is to 1, the closer the rotor shaft center trajectory is to the ideal ellipse.

[0070] Please see Figure 3 Step 2: Calculate the trajectory irregularity, specifically: take each trajectory point on the rotor shaft trajectory curve as the center and construct a local neighborhood based on a preset fixed neighborhood radius.

[0071] The determination of the preset fixed neighborhood radius needs to balance local detail capture and noise resistance. In this invention, 3 to 5 times the average spacing between trajectory points can be used as the fixed neighborhood radius. If the trajectory points are very dense, a smaller multiple can be used to capture details, such as 3 times; if the trajectory noise is large, a larger multiple can be used to smooth local fluctuations, such as 4.5 times.

[0072] Next, the curvature k of all trajectory points (x, y) within each local neighborhood is calculated using the central difference method. The specific calculation process is as follows: .

[0073] in, , It is the first derivative; , It is the second derivative.

[0074] To eliminate the influence of the absolute size of the trajectory on the curvature value, each curvature value is multiplied by the length of the major semi-axis of the fitted ellipse corresponding to the current rotor shaft center trajectory to obtain the normalized curvature.

[0075] Then, the variance of all normalized curvatures within each local neighborhood is calculated first, serving as a local curvature fluctuation index. Next, the standard deviation of all local curvature fluctuation indices is calculated, serving as the first irregularity component. The first irregularity component reflects the degree of drastic change in the local curvature of the trajectory; a larger value indicates a more drastic change in the local curvature of the rotor shaft trajectory, while a smaller value indicates a smoother local rotor shaft trajectory.

[0076] Simultaneously, the total arc length of the rotor shaft center trajectory curve is first obtained by accumulating the Euclidean distances between adjacent trajectory points, and the circumference of the ellipse is calculated using the Ramanujan approximation formula. Then, the absolute value of the difference between the total arc length of the rotor shaft center trajectory curve and the circumference of the fitted ellipse is calculated, and normalization is performed based on the circumference of the fitted ellipse to obtain the second irregularity component.

[0077] The second irregularity component measures the overall difference between the rotor shaft center trajectory profile length and the ideal elliptical profile. The larger the value, the greater the difference between the rotor shaft center trajectory profile and the ideal ellipse.

[0078] Finally, the first irregularity component and the second irregularity component are weighted and summed to obtain the trajectory irregularity.

[0079] The weights corresponding to the first and second irregularity components can be determined based on historical fault data. For example, at least 100 samples of oil film whirl and steam flow excitation are collected, and their respective first and second irregularity components are calculated. The Fisher discriminant ratio is used for optimization to maximize the inter-class distance between the two fault types: oil film whirl and steam flow excitation. Finally, the optimal ratio is normalized and used as the weight.

[0080] The operating condition correlation module is used to indicate that the rotor shaft trajectory exhibits a typical fault pattern when both trajectory ellipticity and trajectory irregularity enter a preset range. Oil film whirl, due to the imbalance caused by the superposition of half-speed whirl, typically presents a concave banana shape or teardrop shape with low trajectory ellipticity; steam flow excitation, due to multi-frequency resonance, easily forms an approximately symmetrical petal shape or elliptical shape with high trajectory ellipticity.

[0081] At this point, real-time power data of the steam turbine is acquired simultaneously, and the dominant correlation strength between the vibration dominant frequency amplitude sequence and the power change sequence is analyzed.

[0082] The preset interval can be determined based on a large amount of historical data. For example, the mean and standard deviation of trajectory ellipticity and irregularity in the historical data of oil film eddy and steam flow excitation can be calculated respectively, and the range of mean ± N times the standard deviation can be used as the preset interval.

[0083] In a preferred embodiment of the present invention, N can be 3, thereby covering most of the typical data points corresponding to the fault mode.

[0084] The specific analysis process for the dominant correlation strength is as follows: the vibration dominant frequency amplitude sequence and the real-time power change sequence of the steam turbine are time-aligned to ensure that the timestamps of the two are consistent.

[0085] Then, a sliding time window method is used, where the length of the sliding window needs to cover a sufficient number of data points, such as 10 to 20 rotor rotation cycles; in this invention, 15 can be used as an example. Within each window, the time-varying Pearson correlation coefficient between the vibration dominant frequency amplitude sequence and the real-time power change sequence of the turbine is calculated to obtain the correlation coefficient sequence.

[0086] .

[0087] Here, the numerator represents the covariance of the vibration dominant frequency amplitude sequence and the real-time power change sequence of the steam turbine, and the denominator is the product of their standard deviations. i is the data point index, with a value range of 1, 2, ..., N, where N is the number of data points in the window. and These represent the vibration dominant frequency amplitude and power of the i-th data point, respectively; and This represents the mean value within the window.

[0088] Statistical analysis was performed on the correlation coefficient sequence to extract its maximum value, mean, and standard deviation. The maximum value reflects the strongest vibration response to power changes within the analysis period. The maximum value was used as the dominant correlation strength between the vibration dominant frequency amplitude sequence and the power change sequence, and was subsequently used to construct feature vectors for the fault diagnosis module.

[0089] The mean reflects the overall trend of association, while the standard deviation reflects the stability of association. The mean and standard deviation are used as auxiliary association indicators for subsequent assessment of association stability.

[0090] It should be added that if effective real-time power data cannot be obtained when correlation analysis is required, the system will pause the current correlation analysis and issue a data source anomaly alarm. At the same time, it can issue a preliminary morphological anomaly alarm based only on the current trajectory ellipticity and trajectory irregularity.

[0091] Please see Figure 4 The fault diagnosis module is used to form a current feature vector by combining trajectory ellipticity, trajectory irregularity and dominant correlation strength, compare it with a preset reference vector, and then determine whether it is an oil film whirl fault or a steam flow excitation fault based on the comparison result, and output the corresponding diagnosis result.

[0092] The preset reference vectors include the oil film whirl fault reference vector and the steam flow excitation fault reference vector.

[0093] These reference vectors can be determined by: screening single typical fault case data confirmed by on-site vibration testing and disassembly inspection; calculating the trajectory ellipticity, trajectory irregularity, and dominant correlation strength of each case; calculating the median of each component of the eigenvector of the oil film whirl and steam flow excitation case sets respectively; and using the obtained three-dimensional median point as the reference vector of the corresponding fault.

[0094] If there is no historical fault data for the current steam turbine at the initial deployment of the system, the reference vector can be determined based on the typical fault case data corresponding to the same or similar models.

[0095] During diagnosis, the current trajectory ellipticity, trajectory irregularity, and dominant correlation strength are combined to form the current feature vector; the Euclidean distance between the current feature vector and the oil film whirl fault reference vector and the steam flow excitation fault reference vector are calculated to obtain the first deviation and the second deviation, respectively.

[0096] If the smaller of the first deviation and the second deviation is greater than or equal to the fault determination threshold, it indicates that although the current state may be abnormal, it does not meet the typical characteristics of the two known types of faults. It may be other faults or a composite state. In this case, it is determined to be a fault without a clear target and manual intervention is guided.

[0097] When both the first deviation and the second deviation are less than the fault determination threshold, if the ratio of the first deviation to the second deviation is within a preset range, it is determined to be feature fuzzy, and the fault type cannot be clearly distinguished, and a feature fuzzy alarm is output.

[0098] The proximity interval can be statistically set based on the distribution of the first and second deviation ratios of oil film eddy and steam flow excitation in historical data. For example, the 10% to 90% quantile of the ratio distribution can be taken. In this invention, the proximity interval is preferably 0.9 to 1.1.

[0099] Otherwise, if the first deviation is less than the second deviation and less than the fault determination threshold, it indicates that the current state is closer to an oil film whirl fault. The typical characteristics of an oil film whirl fault are a banana-shaped trajectory, low trajectory ellipticity, moderate trajectory irregularity, and a weak correlation between the amplitude of the dominant vibration frequency and the power; then it is determined to be an oil film whirl fault.

[0100] If the second deviation is less than the first deviation and less than the fault determination threshold, it indicates that the current state is closer to a steam flow excitation fault. The typical characteristics of a steam flow excitation fault are a petal-shaped trajectory, high trajectory ellipticity, large trajectory irregularity, and a strong correlation between the amplitude of the dominant vibration frequency and the power; then it is determined to be a steam flow excitation fault.

[0101] Finally, the output includes the determined fault type, the first deviation, and the second deviation as diagnostic results.

[0102] The specific process for determining the fault judgment threshold can be as follows: calculate the Euclidean distance from the feature vector to the oil film whirl fault reference vector and the steam flow excitation fault reference vector under at least 100 sets of normal operating conditions, and take the 90th percentile of the minimum value among all distances as the initial fault judgment threshold to ensure that the probability of misjudging as a fault under normal conditions is low.

[0103] To further enhance the depth of diagnosis, after determining the above-mentioned fault types, auxiliary correlation indicators can be used to verify feature consistency.

[0104] The specific process is as follows: First, calculate the difference between the dominant association strength and the mean, and use the ratio of this difference to the standard deviation as the association significance coefficient.

[0105] Steam flow excitation is essentially aerodynamic excitation. An increase in turbine power will increase steam flow, which in turn increases the excitation force and thus increases vibration. The correlation should be positive.

[0106] If the fault type is steam flow excitation fault and the correlation significance coefficient is negative, it indicates that the vibration and power do not show a typical positive correlation, which may be affected by other factors or is not typical steam flow excitation. In this case, the assessment conclusion is that the correlation strength between vibration and power does not reach the typical level of steam flow excitation.

[0107] Oil film whirl faults are caused by oil film instability and are mainly affected by speed, oil temperature and load. They are not directly and strongly correlated with power, and the correlation significance coefficient should usually be close to 0 or slightly negative.

[0108] If the fault type is oil film whirl fault and the correlation significance coefficient is positive, it indicates that there may be steam flow excitation coupling or other external excitations, and compound faults should be noted; at this time, the assessment conclusion is that the correlation strength between vibration and power is abnormally high.

[0109] Otherwise, the assessment concludes that the correlation strength between vibration and power matches the typical characteristics of this fault type. Finally, the assessment conclusion and diagnostic results are correlated and output.

[0110] It is important to note that if the vibration dominant frequency amplitude sequence remains at zero or the baseline value, the system can be considered as having a signal abnormality. The fault type discrimination process based on the dominant correlation strength will be suspended, and a vibration feature missing alarm will be directly output to guide manual intervention.

[0111] If valid real-time power data cannot be obtained when correlation analysis is required, the system will pause the current correlation analysis.

[0112] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0113] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0114] In addition, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0115] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0116] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An online monitoring system for the operating status of a coal-fired boiler, characterized in that, include: The signal acquisition module is used to simultaneously acquire vibration timing signals in two orthogonal directions, the X-axis and the Y-axis, through vibration sensors installed at the turbine bearings of the boiler. The trajectory construction module is used to construct the rotor shaft center trajectory curve based on the vibration time sequence signal and calculate the vibration dominant frequency amplitude sequence. The feature quantization module is used to calculate the ellipticity and irregularity of the trajectory based on the rotor shaft center trajectory curve. The operating condition correlation module is used to synchronously acquire the real-time power data of the steam turbine when the trajectory ellipticity and trajectory irregularity both enter the preset range, and to time-align the vibration main frequency amplitude sequence with the real-time power change sequence of the steam turbine. The sliding time window method is used to calculate the time-varying Pearson correlation coefficient between the vibration dominant frequency amplitude sequence and the real-time power change sequence of the steam turbine within each window, thus obtaining the correlation coefficient sequence. The maximum value in the correlation coefficient sequence is extracted as the dominant correlation strength between the vibration dominant frequency amplitude sequence and the power change sequence. The fault diagnosis module is used to form a current feature vector by combining trajectory ellipticity, trajectory irregularity, and dominant correlation strength, compare it with a preset reference vector, and then determine whether it is an oil film whirl fault or a steam flow excitation fault based on the comparison result, and output the corresponding diagnosis result.

2. The online monitoring system for the operating status of a coal-fired boiler according to claim 1, characterized in that, The deployment of the vibration sensor and the acquisition process of the vibration timing signal are as follows: Determine the azimuth angle of the maximum load zone based on the geometric parameters of the target turbine bearing; A sensor installation coordinate system is established with the direction indicated by the azimuth of the maximum load area as the X-axis and its normal direction as the Y-axis. Vibration sensors are symmetrically arranged along the X and Y axes in the radial projection area corresponding to the rotor shaft center on the outer surface of the bearing housing. During the operation of the steam turbine, the timing signals output by all vibration sensors are collected synchronously and used as vibration timing signals in the X-axis and Y-axis directions, respectively.

3. The online monitoring system for the operating status of a coal-fired boiler according to claim 2, characterized in that, The process for determining the azimuth angle of the maximum load region is as follows: The geometric parameters include bearing type, bearing bush splitting surface position, oil wedge starting angle, bearing top clearance and left and right side clearances; Using the centerline of the bearing bush split surface as the zero-degree reference line, the positive angle is defined along the rotor rotation direction to establish the bearing section coordinate system; The number and circumferential position of the bearing-bearing oil wedges are determined according to the bearing type. Calculate the static eccentricity direction of the journal based on the top clearance, left and right side clearances, bearing inner diameter and journal diameter; With the static eccentric direction as the center, extend 90 degrees to both sides along the circumference to form an eccentric side semicircle; Within the eccentric semicircle, identify the oil wedge where the radial clearance between the inner surface of the bearing bush and the journal gradually decreases along the rotor rotation direction, and use it as the main bearing oil wedge. If there are multiple main bearing wedges, the one with the smallest angle between its geometric center line and the vertical center line of the bearing housing bottom surface shall be selected as the final main bearing wedge. The angle corresponding to the final geometric center line of the main load-bearing wedge in the bearing section coordinate system is taken as the azimuth angle of the maximum load zone.

4. The online monitoring system for the operating status of a coal-fired boiler according to claim 2, characterized in that, The process of constructing the rotor shaft center trajectory curve is as follows: The vibration timing signals of the X-axis and Y-axis are respectively used as the displacement components of the rotor shaft center on the corresponding coordinate axes of the sensor mounting coordinate system; Using the same sampling time as an index, the two displacement components are paired to form a two-dimensional coordinate point sequence; Based on the rotor rotation period, the two-dimensional coordinate point sequence is resampled at equal angles to generate a trajectory point sequence with the rotor rotation circumferential angle as the independent variable. Each trajectory point corresponds to a specific rotation phase of the rotor. Connect all trajectory points sequentially according to their rotational phase from zero degrees to 360 degrees to form the rotor shaft centerline trajectory curve.

5. The online monitoring system for the operating status of a coal-fired boiler according to claim 1, characterized in that, The calculation process for the vibration dominant frequency amplitude sequence is as follows: Short-time Fourier transforms were performed on the vibration time-series signals of the X and Y axes to obtain two sets of vibration spectra; For each identical analysis time window, the amplitudes at the same frequency points in the two sets of vibration spectra are vector-synthesized to obtain the comprehensive vibration spectrum; All frequency points in the comprehensive vibration spectrum whose frequency values ​​are within a preset octave band of the rotor's operating frequency are selected as candidate points; Select the candidate point with the largest amplitude as the dominant vibration frequency, and record its frequency and amplitude. The amplitude values ​​are arranged in the order of the analysis time window to form the vibration dominant frequency amplitude sequence.

6. The online monitoring system for the operating status of a coal-fired boiler according to claim 2, characterized in that, The calculation process for the ellipticity of the trajectory is as follows: Perform least-squares ellipse fitting on all trajectory points on the rotor shaft center trajectory curve to obtain the fitted ellipse; Extract the center coordinates, semi-major axis length, and major axis direction of the fitted ellipse; All trajectory points are rotated synchronously to align the major axis of the fitted ellipse with the horizontal axis of the sensor mounting coordinate system. In the rotated coordinate system, calculate the Euclidean distance from each trajectory point to the center coordinate of the fitted ellipse, and normalize it based on the length of the major semi-axis to obtain the normalized radial value. Calculate the variance of all normalized radial values ​​over a complete rotation cycle, add one to this variance, and define the result as the trajectory ellipticity.

7. The online monitoring system for the operating status of a coal-fired boiler according to claim 6, characterized in that, The calculation process for the trajectory irregularity is as follows: Using each trajectory point on the rotor shaft trajectory curve as the center, a local neighborhood is constructed based on a preset fixed neighborhood radius; The curvature of all trajectory points in each local neighborhood is calculated using the central difference method, and its variance is calculated as an index of local curvature fluctuation. Calculate the standard deviation of all local curvature fluctuation indices and use it as the first irregularity component; Calculate the absolute value of the difference between the total arc length of the rotor shaft center trajectory curve and the circumference of the fitted ellipse, and normalize it based on the circumference of the fitted ellipse to obtain the second irregularity component. The trajectory irregularity is obtained by weighted summation of the first and second irregularity components.

8. The online monitoring system for the operating status of a coal-fired boiler according to claim 1, characterized in that, Statistical analysis was performed on the correlation coefficient series to extract its mean and standard deviation, which were then used as auxiliary correlation indicators.

9. The online monitoring system for the operating status of a coal-fired boiler according to claim 8, characterized in that, Determine whether the fault is oil film whirl or steam flow vibration and output the corresponding diagnostic results, specifically: The preset reference vectors include an oil film whirl fault reference vector and a steam flow excitation fault reference vector. The current trajectory ellipticity, trajectory irregularity, and dominant correlation strength are combined to form the current feature vector; Calculate the Euclidean distance between the current feature vector and the reference vectors for oil film whirl fault and steam flow excitation fault, and obtain the first deviation and the second deviation respectively; If the smaller of the first deviation and the second deviation is greater than or equal to the fault determination threshold, it is determined to be a fault without a clear target. When the smaller of the first deviation and the second deviation is less than the fault determination threshold, if the ratio of the first deviation to the second deviation is within a preset range, it is determined to be feature fuzzy. Otherwise, if the first deviation is less than the second deviation, it is determined to be an oil film whirl fault; if the second deviation is less than the first deviation, it is determined to be a steam flow excitation fault. The output includes the determined fault type, the first deviation, and the second deviation as diagnostic results.

10. The online monitoring system for the operating status of a coal-fired boiler according to claim 9, characterized in that, Specifically, the mean and standard deviation are used as auxiliary correlation indicators: Calculate the difference between the dominant association strength and the mean, and use the ratio of this difference to the standard deviation as the association significance coefficient; If the fault type is steam flow excitation fault and the correlation significance coefficient is negative, the assessment conclusion is that the correlation strength between vibration and power does not reach the typical level of steam flow excitation. If the fault type is oil film whirl fault and the correlation significance coefficient is positive, the assessment conclusion is that the correlation strength between vibration and power is abnormally high. Otherwise, the assessment conclusion is that the correlation strength between vibration and power conforms to the typical characteristics of this type of fault; the assessment conclusion is then output in conjunction with the diagnostic results.

Images