A method of analyzing a vibration value

Abstract

This invention relates to the field of engine vibration characteristic analysis technology, and particularly to a method for analyzing vibration values. The method extracts historical vibration values ​​and corresponding historical engine characteristic values ​​at specific times, and converts them to a first characteristic domain for discretization to obtain a first vibration value group. A baseline vibration value feature band is generated based on the first vibration value group. The method then extracts the vibration values ​​to be analyzed and the corresponding engine characteristic values ​​at specific times, and converts them to a second characteristic domain for discretization to obtain a second vibration value group. A feature band for the second vibration value group is generated based on the second vibration value group. Anomalies and overall trend anomaly ranges are determined based on the comparison between the feature band for the second vibration value and the baseline vibration value feature band. This invention generates a feature band from historical data and compares it with the feature band generated from the data to be analyzed, achieving precise location of engine vibration anomalies and trend identification, providing a reliable basis for fault diagnosis.

Description

Technical Field

[0001] This invention relates to the field of engine vibration characteristic analysis technology, and in particular to a method for analyzing vibration values. Background Technology

[0002] As the core power unit of ships, vehicles, and industrial power generation, the reliability of engines directly affects the safety and economic benefits of the entire machine. Mechanical failures account for more than 60% of the total engine failure rate, and these failures significantly alter the vibration response characteristics of the system in their early stages. Therefore, vibration signal monitoring and analysis has become the most important technical approach for condition-based engine maintenance, the core of which lies in accurately extracting characteristic information that can characterize the structural health status from measured vibration values.

[0003] Existing vibration characteristic analysis methods mainly follow two technical routes: one is fault feature extraction based on signal processing, and the other is dynamic parameter identification based on operating condition correlation. Although signal processing-based methods have good identification capabilities for typical faults under steady-state operating conditions, the feature extraction process is disconnected from engine operating parameters, which can easily lead to poor feature stability and insufficient generalization ability under operating conditions. While dynamic parameter identification based on operating condition correlation can quantitatively evaluate the mechanical integrity and health degradation of the engine through continuous identification and trend analysis of engine structural dynamic parameters, it suffers from heavy online real-time computational burden and lacks a unified means to characterize the evolution of vibration characteristics under all operating conditions and throughout the entire life cycle.

[0004] Therefore, there is an urgent need for a new technical approach that can filter out the interference of operating condition fluctuations on vibration characteristics and achieve accurate tracking of individualized degradation trajectories, thereby providing an engineerable and deployable approach for engine structural condition assessment. Summary of the Invention

[0005] To address this issue, the present invention provides a vibration value analysis method to solve the problem in the prior art that the interference of operating condition fluctuations on vibration characteristics cannot be filtered out, thus making it impossible to accurately locate engine vibration anomalies and identify trends.

[0006] To achieve the above objectives, the present invention provides a method for analyzing vibration values, comprising: Extract historical vibration values ​​and corresponding historical engine characteristic values ​​from the original operational monitoring data of several historical batches of the engine; The historical vibration values ​​and the historical engine characteristic values ​​are transformed from the time domain to the first characteristic domain and the characteristic values ​​are discretized to obtain the first vibration value group corresponding to the first discrete working condition interval. Fitting is performed based on the first vibration value group to generate a reference vibration value feature baseband; Extract the vibration values ​​of the engine to be analyzed and the corresponding characteristic values ​​of the engine to be analyzed from the original working monitoring data of the engine to be analyzed. The vibration values ​​to be analyzed and the engine characteristic values ​​to be analyzed are transformed from the time domain to the second characteristic domain and the characteristic values ​​are discretized to obtain the second vibration value group corresponding to the second discrete working condition interval. Fitting is performed based on the second vibration value to generate the characteristic band of the vibration value to be analyzed; Based on the comparison results between the characteristic band of the vibration value to be analyzed and the characteristic base band of the benchmark vibration value, the abnormal points and the abnormal change range of the overall trend are determined.

[0007] Furthermore, the process of obtaining the first vibration value group includes: Using the historical vibration value as the vertical axis and the first characteristic quantity as the horizontal axis, a first mapping coordinate system is constructed to show the change of the historical vibration value with the corresponding first characteristic quantity. The first characteristic quantity includes the first rotational speed, the first converted rotational speed, and the first overload. The first mapping coordinate system includes coordinate systems corresponding to the first rotational speed domain, the first converted rotational speed domain, and the first overload domain, respectively. For each coordinate system, according to the preset first discretization step size or interval division strategy, the value range of the first feature quantity corresponding to the corresponding coordinate system is divided into several continuous first discrete working condition intervals. All historical vibration values ​​falling within the same first discrete operating condition range are aggregated to form the first vibration value group.

[0008] Furthermore, the process of generating the baseband characteristic of the reference vibration value includes: Statistical characteristic values ​​within the first discrete working condition interval are calculated corresponding to the first vibration value group to obtain the first maximum value, first minimum value, first median, first average value, first variance, first midpoint value, first lower quantile, first upper quantile, and first interquartile range; The reference vibration value characteristic baseband is obtained by fitting the first maximum value, the first minimum value, the first median, the first average value, the first variance, the first midpoint value, the first lower quantile, the first upper quantile, and the first interquartile range.

[0009] Furthermore, the process of obtaining the second vibration value group includes: Using the vibration value to be analyzed as the vertical axis and the second characteristic quantity as the horizontal axis, a second mapping coordinate system is constructed to show the change of the vibration value to be analyzed with the corresponding second characteristic quantity. The second characteristic quantity includes the second rotational speed, the second converted rotational speed, and the second overload. The second mapping coordinate system includes the coordinate systems corresponding to the second rotational speed domain, the second converted rotational speed domain, and the second overload domain, respectively. For each coordinate system, according to the preset second discretization step size or interval division strategy, the value range of the second feature quantity corresponding to the coordinate system is divided into several continuous second discrete working condition intervals. All the vibration values ​​to be analyzed that fall within the same second discrete operating condition range are aggregated to form the second vibration value group.

[0010] Furthermore, the process of generating the characteristic band of the secondary vibration values ​​to be analyzed includes: Statistical characteristic values ​​within the second discrete working condition interval are calculated corresponding to the second vibration value group to obtain the second maximum value, second minimum value, second median, second average value, second variance, second midpoint value, second lower quantile, second upper quantile, and second interquartile range; The characteristic band of the vibration value to be analyzed is obtained by fitting the second maximum value, the second minimum value, the second median, the second average value, the second variance, the second midpoint value, the second lower quantile, the second upper quantile, and the second interquartile range.

[0011] Furthermore, the comparison process between the characteristic band of the secondary vibration value to be analyzed and the characteristic base band of the reference vibration value includes: The characteristic band of the vibration value to be analyzed and the characteristic base band of the reference vibration value are superimposed and aligned in the same characteristic quantity domain coordinate system; Based on the first discrete working condition interval and the second discrete working condition interval, the statistical characteristics of the vibration value feature band to be analyzed are compared with the baseline statistical characteristics of the baseline vibration value feature base band. If, within any discrete operating condition interval, the statistical characteristics of the vibration value feature band to be analyzed exceed the reference boundary value corresponding to the reference vibration value feature base band, then the interval is marked as an outlier. If, within multiple consecutive discrete operating condition intervals, the statistical characteristics of the vibration value characteristic band to be analyzed continuously exceed the reference boundary value corresponding to the reference vibration value characteristic baseband, and the magnitude of the exceedance shows a monotonically changing or cumulative offset trend, then the continuous interval is determined to be an interval with an abnormal overall trend.

[0012] Furthermore, vibration anomaly diagnostic information is generated based on the anomaly points and the overall trend anomaly change range. The vibration anomaly diagnostic information includes the location of the abnormal working condition range, the quantitative value of the anomaly degree, and the trend characteristics of the anomaly changing with the characteristic quantity.

[0013] Furthermore, the process of generating vibration anomaly diagnostic information also includes: The vibration distribution characteristics corresponding to the abnormal points and the abnormal change range of the overall trend are output as vibration abnormality alarm signals and abnormal operating condition range locations.

[0014] Furthermore, the process of cleaning the raw work monitoring data includes: Before obtaining the historical vibration values ​​and the historical engine characteristic values, data cleaning is performed on several historical batches of the original working monitoring data to remove abnormal record values. Before obtaining the vibration value and engine characteristic value of the analysis session, the original working monitoring data of the analysis session is cleaned to remove abnormal recorded values.

[0015] Furthermore, the data cleaning includes: Remove abnormal record values ​​from the original monitoring data caused by sensor malfunction or duplicate timestamps.

[0016] Compared with existing technologies, the vibration value analysis method of the present invention has the following advantages: It extracts historical vibration values ​​and corresponding historical engine characteristic values ​​from the original operational monitoring data of historical batches of engines, transforms both from the time domain to a first characteristic domain for discretization, and obtains a first vibration value group corresponding to a first discrete operating condition interval. Based on the first vibration value group, it performs fitting to generate a baseline vibration value feature band. Simultaneously, it extracts the vibration values ​​of the analysis session and the corresponding engine characteristic values ​​from the original operational monitoring data of the analysis session, and uses the same data cleaning rules, parameter thresholds, and execution order to transform the analysis session data from the time domain to a second characteristic domain for discretization, obtaining a second vibration value group corresponding to a second discrete operating condition interval. Based on the second vibration value group, it performs fitting to generate a vibration value feature band for the analysis session. Finally, based on the comparison results between the vibration value feature band for the analysis session and the baseline vibration value feature band, it determines the anomaly points and the overall trend anomaly change interval. With this setup, the present invention constructs a closed-loop analysis architecture encompassing historical benchmark construction, analysis session feature extraction, and source comparison diagnosis, achieving accurate localization and trend identification of engine vibration anomalies.

[0017] Furthermore, this invention constructs corresponding coordinate systems for a first rotational speed domain, a first converted rotational speed domain, and a first overload domain using a first rotational speed, a first converted rotational speed, and a first overload as the abscissa. These coordinate systems are then discretized at equal intervals according to a preset fixed step size, and historical vibration values ​​falling within the same interval are aggregated to form a first vibration value group. This setup pre-defines the boundary conditions of the vibration characteristics from three dimensions: excitation intensity, operating condition normalization, and transmission path correction, thus avoiding the impact of multiple feature coupling on the operating condition resolution.

[0018] Furthermore, this invention also calculates the first maximum value, first minimum value, first median, first average value, first variance, first midpoint value, first lower quantile, first upper quantile, and first interquartile range using the first vibration value group within each discrete operating condition interval. Based on the above statistical characteristic values, piecewise linear interpolation fitting is performed to obtain the baseline characteristic band of the benchmark vibration value. This setup predefines the vibration value distribution characteristics from multiple dimensions, including absolute boundary, geometric center, distribution center, overall level, fluctuation degree, main interval, and robust discreteness, so that the constructed baseline band possesses both robustness and sensitivity.

[0019] Furthermore, this invention ensures that the sub-feature band to be analyzed and the baseline baseband maintain strict homology in data source, processing flow, and mathematical form by employing the exact same feature selection, discretization step size, and fitting method as historical data. This setup eliminates false anomalies caused by differences in data preprocessing, ensuring that the comparison results accurately reflect changes in the engine's physical state.

[0020] Furthermore, this invention also involves superimposing and aligning the secondary vibration characteristic band to be analyzed and the baseline vibration characteristic band in the same characteristic quantity domain coordinate system during the comparison process. The statistical characteristics of the secondary characteristic band to be analyzed are compared interval by interval with the baseline statistical characteristics of the baseline band. If the center value of the secondary characteristic band to be analyzed exceeds the baseline boundary in any interval, or if the characteristic band as a whole does not overlap with the baseline band, it is marked as an anomaly. If the statistical characteristics of the secondary characteristic band to be analyzed continuously exceed the baseline boundary in multiple consecutive intervals, and the magnitude of the exceedance shows a monotonically changing or cumulative offset trend, it is determined to be an interval with an overall abnormal trend. This setup achieves comprehensive identification from isolated anomalies to trend anomalies, providing multi-level diagnostic basis for engine health management.

[0021] Furthermore, this invention generates vibration anomaly diagnostic information based on anomaly points and overall trend anomaly change intervals. This information includes the location of the abnormal operating condition interval, the quantified value of the anomaly severity, and the trend characteristics of the anomaly changing with characteristic quantities. The corresponding vibration distribution characteristics are then output as vibration anomaly alarm signals and the location of the abnormal operating condition interval. This setup provides maintenance personnel with a basis for fault location and severity assessment.

[0022] Furthermore, this invention ensures data consistency by applying the exact same cleaning rules, parameter thresholds, and execution order to both historical data and the data to be analyzed; it further purifies the data source by removing abnormal records caused by sensor malfunctions and duplicate timestamps. This setup establishes a high-confidence data foundation, avoiding false alarms and missed alarms due to data quality issues. Attached Figure Description

[0023] Figure 1 This is a block diagram of the vibration value analysis method in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the vibration value analysis method in an embodiment of the present invention; Figure 3 This is a flowchart illustrating the logic for determining outliers based on the comparison between the center value and the baseline boundary in an embodiment of the present invention. Figure 4 This is a diagram illustrating the effect of comparing the characteristic band of the secondary vibration value with the characteristic base band of the reference vibration value in an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0025] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0026] It should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0027] Please see Figure 1 As shown, it is a block diagram of the vibration value analysis method in an embodiment of the present invention.

[0028] This embodiment includes a data acquisition module, a data cleaning module, a historical data processing module, a secondary data processing module, a comparison and analysis module, a diagnostic output module, and a control module.

[0029] The data acquisition module includes several vibration sensors, speed sensors, temperature sensors, pressure sensors, and overload sensors, which are used to collect vibration signals, speed signals, intake air temperature, intake air pressure, and flight overload signals in real time during engine operation to obtain raw operational monitoring data.

[0030] The data cleaning module is connected to the data acquisition module. It is used to clean the raw work monitoring data, remove abnormal record values ​​caused by sensor failure or duplicate timestamps, and use the same cleaning rules, parameter thresholds and execution order for historical batch data and the data to be analyzed to ensure data homogeneity.

[0031] The historical data processing module is connected to the data cleaning module and includes a historical discretization unit and a historical fitting unit. The historical discretization unit is used to transform the cleaned historical vibration values ​​and corresponding historical engine characteristic values ​​from the time domain to the first characteristic domain. Using the first speed, the first converted speed, and the first overload as characteristic values, it performs equal-interval discretization according to a preset fixed step size to obtain the first vibration value group corresponding to the first discrete operating condition interval. The historical fitting unit is used to calculate the first maximum value, the first minimum value, the first median, the first average value, the first variance, the first midpoint value, the first lower quantile, the first upper quantile, and the first interquartile range for the first vibration value group in each first discrete operating condition interval. Based on the above statistical characteristic values, it performs piecewise linear interpolation fitting to generate the reference vibration value characteristic baseband.

[0032] The data processing module for the analysis phase is connected to the data cleaning module, and includes a discretization unit and a fitting unit for the analysis phase. The discretization unit transforms the cleaned vibration values ​​and corresponding engine characteristic values ​​from the time domain to a second characteristic domain. Using the second rotational speed, second converted rotational speed, and second overload as characteristic quantities, it performs equal-interval discretization with the same fixed step size as the historical data processing module to obtain the second vibration value cluster corresponding to the second discrete operating condition interval. The fitting unit calculates the second maximum, second minimum, second median, second average, second variance, second midpoint, second lower quantile, second upper quantile, and second interquartile range for the second vibration value cluster within each second discrete operating condition interval. Based on these statistical characteristic values, it performs piecewise linear interpolation fitting to generate the vibration value feature band for the analysis phase.

[0033] The comparison and analysis module is connected to the historical data processing module and the data processing module for the sub-vibration values ​​to be analyzed, respectively. It is used to superimpose and align the characteristic band of the sub-vibration value to be analyzed with the base band of the reference vibration value in the same characteristic quantity domain coordinate system, and compare the statistical characteristics of the characteristic band of the sub-vibration value to the reference statistical characteristics of the base band in interval by interval. If the center value of the characteristic band of the sub-vibration value to be analyzed exceeds the reference boundary or the characteristic band as a whole does not overlap with the reference base band in any discrete working condition interval, the interval is marked as an outlier. If the statistical characteristics of the characteristic band of the sub-vibration value to be analyzed continuously exceed the reference boundary in multiple consecutive discrete working condition intervals and the exceedance shows a monotonically changing or cumulative offset trend, the consecutive interval is determined to be an interval with an abnormal overall trend.

[0034] The diagnostic output module is connected to the comparison and analysis module. It is used to generate vibration anomaly diagnostic information based on anomaly points and the overall trend anomaly change range, and output the vibration distribution characteristics corresponding to the anomaly points and the overall trend anomaly change range as vibration anomaly alarm signals and the location of the abnormal working condition range.

[0035] The control module is connected to the diagnostic output module and the engine control system respectively. It is used to generate corresponding control commands based on the vibration abnormality alarm signal and the location of the abnormal operating condition range, so as to adjust the engine operating parameters or trigger maintenance warnings.

[0036] Please see Figure 2 The diagram shown is a flowchart illustrating the vibration value analysis method in an embodiment of the present invention. The process in this embodiment includes at least the following steps: S1: Extract historical vibration values ​​and corresponding historical engine characteristic values ​​from the original working monitoring data of several historical batches of the engine. Specifically, vibration characteristic parameters are extracted from the original monitoring data of the engine's history or the analysis batch, and the data is cleaned to remove abnormal recorded values ​​to obtain historical characteristic values ​​and corresponding historical engine characteristic values, as well as vibration values ​​and corresponding engine characteristic values ​​of the analysis batch. Here, the analysis batch refers to the specific flight mission or work batch that requires vibration status assessment. The original historical work monitoring data and the original work monitoring data of the analysis batch must use the exact same cleaning rules, parameter thresholds, and execution order to ensure the data homogeneity of the baseband of the reference vibration characteristic and the characteristic band of the vibration value of the analysis batch.

[0037] In this embodiment, the system checks whether key fields such as timestamp, vibration value, and engine speed contain empty values ​​or invalid placeholders. If a field has no data, a non-numerical invalid flag is returned, or a sensor-specific fault code (such as...) is returned. Records deemed invalid are discarded entirely. For multiple records from the same acquisition channel and under the same timestamp, if all measurement point values ​​are identical, only the first record is retained; if the measurement point values ​​are different, the arithmetic mean of the vibration value and engine characteristic value under that timestamp is taken and merged into a single record. Simultaneously, if network caching causes later acquired data to be assigned an earlier timestamp, resulting in a timeline inversion, the timestamp order is recalibrated based on the acquisition system record sequence number; if this cannot be corrected, the out-of-order record is discarded. The technology for determining whether data is unrepairable is existing technology and will not be elaborated upon here.

[0038] Furthermore, to facilitate the subsequent construction of a reference vibration characteristic baseband, steady-state identification and cleaning can be performed. The standard deviation of the rotational speed is calculated with a sliding window of 1 second. If the rotational speed fluctuation within the window is less than 0.5% of the rated speed and the continuous steady-state duration exceeds 2 seconds, it is determined to be a steady-state condition and retained, while other transient condition data are discarded.

[0039] S2: Transform the historical vibration values ​​and historical engine characteristic values ​​from the time domain to the first characteristic domain and discretize the characteristic values ​​to obtain the first vibration value group corresponding to the first discrete working condition interval; In this embodiment, the first rotational speed, the first converted rotational speed, and the first overload are obtained as the first characteristic quantities. The first rotational speed determines the frequency and amplitude of synchronous excitation such as rotor imbalance and misalignment, and is the most significant influencing factor on vibration energy. The first converted rotational speed, by correcting for intake air temperature and pressure, eliminates the deviation of environmental conditions on the rotational speed characterization condition, ensuring that historical data collected at different times and spaces have comparable operating condition benchmarks with the data to be analyzed. The first overload reflects the inertial load borne by the machine body, directly affecting casing deformation, bearing clearance, and the force transmission path of the installation joint. Vibration characteristics corresponding to different overload states at the same rotational speed exhibit significant differences. These three parameters comprehensively characterize the boundary conditions of vibration characteristics from three independent dimensions: excitation intensity, operating condition normalization, and transmission path correction.

[0040] Using historical vibration values ​​as the ordinate, and the first rotational speed, first converted rotational speed, and first overload as the abscissas, corresponding coordinate systems are constructed: the first rotational speed domain coordinate system, the first converted rotational speed domain coordinate system, and the first overload domain coordinate system. Each coordinate system represents the distribution law of vibration values ​​with the change of a single characteristic quantity, avoiding the impact of multiple characteristic quantity coupling on the working condition resolution.

[0041] For each coordinate system, the range of characteristic values ​​is divided at equal intervals according to a preset fixed step size to obtain a continuous and non-overlapping first discrete working condition interval. The step size selection follows the following engineering constraint principles.

[0042] The first speed domain uses a fixed step size of 100 rpm. Under steady-state cruise conditions, the speed fluctuation of a turbofan engine typically does not exceed ±10 rpm, and a 100 rpm interval is sufficient to distinguish adjacent operating points with engineering significance. Simultaneously, at this step size, the number of vibration sampling points within each discrete interval of the historical dataset is no less than 30 points to meet the robustness requirements of sample size for subsequent statistical calculations. If the step size is too large, it will lead to a decrease in operating condition resolution and mask local vibration characteristics; if the step size is too small, it will result in sparse samples within the interval, loss of statistical significance, and inability to construct a reliable baseline vibration characteristic. If there are fewer than 30 vibration data records within a certain first discrete operating condition interval, smoothing methods such as merging adjacent intervals or interpolation based on neighboring intervals are used to ensure the robustness of the statistics.

[0043] The first converted speed domain uses a fixed step size of 0.01, where the first converted speed is a dimensionless normalized parameter, and 0.01 corresponds to 1% of the relative speed change, which matches the physical resolution of 100 rpm in the speed domain, ensuring that the two characteristic domains are comparable in terms of operating condition precision.

[0044] The first overload domain uses a fixed step size of 0.1g, where 0.1g is the typical resolution for aircraft maneuver overload records. The overload range covered is -1g to +4g, and dividing it into 0.1g steps yields 50 discrete intervals, sufficient to distinguish the vibration characteristics differences under typical maneuvering conditions such as level flight, hovering, pull-up, and dive, thus maintaining the statistical validity of the sample size in each interval.

[0045] Finally, all historical vibration values ​​falling within the same first discrete operating condition interval are aggregated to form the first vibration value cluster corresponding to that interval. Each first discrete operating condition interval independently corresponds to a vibration value cluster, and the data within the cluster all come from multiple observation records at different time points within the state boundary of the same characteristic quantity. The first vibration value cluster obtained so far will serve as the basic dataset for subsequently constructing the benchmark vibration characteristic baseband, used to calculate the statistical distribution characteristics of vibration values ​​within each operating condition interval, thereby providing a reference benchmark under the same operating condition for the anomaly comparison of the vibration values ​​to be analyzed.

[0046] S3: Fit based on the first vibration value group to generate the reference vibration value characteristic baseband; In this embodiment, for each first vibration value group within a first discrete operating condition interval, the first maximum value, first minimum value, first median, first average value, first variance, first midpoint value, first lower quantile, first upper quantile, and first interquartile range are calculated respectively. The first maximum value is the maximum vibration value within the first discrete operating condition interval, reflecting the upper boundary extreme case of the vibration response under the first discrete operating condition; the first minimum value is the minimum vibration value within the first discrete operating condition interval, reflecting the lower boundary extreme case; the first median is the value located in the middle position after sorting the vibration values ​​within the first discrete operating condition interval (if the sample size is even, the average of the two middle values ​​is taken), reflecting the central position of the distribution; the first average value... The first variance is the arithmetic mean of the vibration values ​​within the first discrete operating condition interval, reflecting the overall average level of the distribution; the first variance is the squared average of the deviations of the vibration values ​​from their mean within the first discrete operating condition interval, reflecting the degree of data fluctuation; the first midpoint is the arithmetic mean of the first maximum and the first minimum values, reflecting the geometric center of the amplitude range of the first vibration values; the first lower quantile Q1 is the value at the 25th percentile after sorting the vibration values ​​within the first discrete operating condition interval, describing the lower tail of the distribution; the first upper quantile Q3 is the value at the 75th percentile after sorting the vibration values ​​within the first discrete operating condition interval, describing the upper tail of the distribution; the first interquartile range is the difference between the first upper quantile and the first lower quantile, measuring the robustness of the data's dispersion.

[0047] The aforementioned nine statistical characteristics comprehensively depict the distribution characteristics of the first vibration value group from different dimensions. Specifically, the first maximum and minimum values ​​define the absolute boundaries of the vibration response; the first midpoint reflects the geometric center of the amplitude range; the first median and first mean characterize the central tendency of the distribution from different perspectives (the median is resistant to outlier interference, while the mean reflects the overall level); the first variance and first interquartile range jointly measure the dispersion of the data (variance is sensitive to local fluctuations, while the interquartile range robustly resists outliers); and the first lower quantile and first upper quantile describe the main range of the distribution. Through the combination of these multidimensional statistics, a comprehensive and robust statistical foundation can be provided for the subsequent construction of the baseline vibration value characteristic zone.

[0048] Furthermore, for the first discrete operating condition intervals corresponding to the first speed domain coordinate system, the first converted speed domain coordinate system, and the first overload domain coordinate system, the midpoint of the interval within each interval is taken as the abscissa. Find the first upper quantile of each interval. Using the vertical axis, we obtain the discrete points of the first upper boundary. ; Calculate the first lower quantile of each interval Using the vertical axis as the ordinate, we obtain the discrete points of the first lower boundary. ; the first median of each interval As the ordinate, the first central reference point is obtained. .

[0049] The piecewise linear interpolation method is used to discretize the first upper boundary point set. according to Connect the smaller and larger segments with straight line segments to generate a continuous first upper boundary fitting curve. Discretize the first lower boundary point set. Perform the same processing to generate a continuous first lower boundary fitting curve. ; set of first central reference points Connect to generate the first center reference line. Among them, the piecewise linear interpolation method is used for two adjacent discrete points. and In the interval any position inside corresponding function pass To obtain a continuous piecewise linear curve.

[0050] Fitted curve from the first upper boundary Fitted curve with the first lower boundary The defined continuous band region forms the baseline characteristic zone of the reference vibration value, wherein the baseline characteristic zone of the reference vibration value is located on the horizontal axis. Each value corresponds to a vertical interval. , representing the normal fluctuation range corresponding to the historical vibration values ​​under the first characteristic quantity state.

[0051] S4: Extract the vibration value of the engine to be analyzed and the corresponding characteristic value of the engine to be analyzed from the original working monitoring data of the engine to be analyzed. S5: Transform the vibration value to be analyzed and the characteristic value of the engine to be analyzed from the time domain to the second characteristic domain and discretize the characteristic values ​​to obtain the second vibration value group corresponding to the second discrete working condition interval. In this embodiment, the second rotational speed, the second converted rotational speed, and the second overload are obtained as the second characteristic quantities. The definitions of the second rotational speed, the second converted rotational speed, and the second overload are consistent with the definitions of the first rotational speed, the first converted rotational speed, and the first overload, respectively, to ensure the uniformity of the physical meaning of the characteristic quantity domain and to lay a common origin basis for the comparison between the characteristic band of the secondary vibration value to be analyzed and the characteristic base band of the reference vibration value.

[0052] Using the vibration value to be analyzed as the ordinate, and the second rotational speed, second converted rotational speed, and second overload as the abscissas, corresponding coordinate systems are constructed: the second rotational speed domain coordinate system, the second converted rotational speed domain coordinate system, and the second overload domain coordinate system. Each coordinate system represents the distribution law of the vibration value to be analyzed as a function of a single characteristic quantity.

[0053] For each coordinate system, the range of values ​​for the second characteristic quantity is divided at equal intervals according to a preset fixed step size that is exactly the same as that for the first discrete operating condition interval, thus obtaining a continuous and non-overlapping second discrete operating condition interval. The step size selection follows the same engineering constraint principle as that for the first discrete operating condition interval.

[0054] The second speed domain uses a fixed step size of 100 rpm; the second converted speed domain uses a fixed step size of 0.01 g; and the second overload domain uses a fixed step size of 0.1 g.

[0055] The step size must be completely consistent with the step size of the first discrete working condition interval to ensure that the data to be analyzed and the historical data have the same resolution at the working condition discretization level, so that each discrete interval can correspond one by one when comparing the feature bands in the subsequent process.

[0056] Finally, all vibration values ​​to be analyzed that fall within the same second discrete operating condition interval are aggregated to form the second vibration value group corresponding to that interval. Each second discrete operating condition interval independently corresponds to a vibration value group, and the data within the group all come from multiple observation records within the state boundary of the same characteristic quantity. The obtained second vibration value group will serve as the basis dataset for subsequently generating the characteristic band of the vibration values ​​to be analyzed, and will be used to calculate the statistical distribution characteristics of the vibration values ​​to be analyzed within each operating condition interval, thereby providing a data foundation for anomaly comparison with the baseline vibration value characteristic band.

[0057] S6: Fit based on the second vibration value to generate the characteristic band of the vibration value to be analyzed; In this embodiment, for each second vibration value group within a second discrete operating condition interval, the second maximum value, second minimum value, second median, second average value, second variance, second midpoint value, second lower quantile, second upper quantile, and second interquartile range are calculated respectively. The second maximum value is the maximum vibration value within the second discrete operating condition interval, reflecting the upper boundary extreme case of the vibration response to be analyzed under that operating condition; the second minimum value is the minimum vibration value within the second discrete operating condition interval, reflecting the lower boundary extreme case; the second median is the value located in the middle position after sorting the vibration values ​​within the second discrete operating condition interval (if the sample size is even, the average of the two middle values ​​is taken), reflecting the central position of the distribution; the second average value... The first value is the arithmetic mean of the vibration values ​​within the second discrete operating condition interval, reflecting the overall average level of the distribution; the second variance is the squared average of the deviations of the vibration values ​​within the second discrete operating condition interval from their mean, reflecting the degree of data fluctuation; the second midpoint is the arithmetic mean of the second maximum and the second minimum values, reflecting the geometric center of the amplitude range of the second vibration values; the second lower quantile P1 is the value at the 25th percentile after sorting the vibration values ​​within the second discrete operating condition interval, describing the lower tail of the distribution; the second upper quantile P3 is the value at the 75th percentile after sorting the vibration values ​​within the second discrete operating condition interval, describing the upper tail of the distribution; the second interquartile range is the difference between the second upper quantile and the second lower quantile, measuring the robustness of the data's dispersion.

[0058] The aforementioned nine statistical characteristics comprehensively depict the distribution characteristics of the second vibration value cluster from different dimensions. Specifically, the second maximum and second minimum values ​​define the absolute boundaries of the vibration response; the second midpoint reflects the geometric center of the amplitude range; the second median and second mean characterize the central tendency of the distribution from different perspectives (the median is resistant to outlier interference, while the mean reflects the overall level); the second variance and second interquartile range jointly measure the dispersion of the data (variance is sensitive to local fluctuations, while the interquartile range robustly resists outliers); and the second lower quantile and second upper quantile describe the main range of the distribution. The combination of these multidimensional statistics provides a comprehensive data foundation for subsequently generating the characteristic bands of the vibration values ​​to be analyzed.

[0059] Furthermore, for the second discrete operating condition intervals corresponding to the second speed domain coordinate system, the second converted speed domain coordinate system, and the second overload domain coordinate system, the midpoint of the interval within each interval is taken as the abscissa. Find the second upper quantile of each interval. Using the ordinate as the vertical axis, we obtain the discrete points of the second upper boundary. ; Calculate the second lower quantile of each interval Using the ordinate as the vertical axis, we obtain the discrete points of the second lower boundary. ; the second median of each interval As the ordinate, the second central reference point is obtained. .

[0060] The piecewise linear interpolation method is used to discretize the set of points on the second upper boundary. according to Connect the smaller and larger lines sequentially with straight line segments to generate a continuous second upper boundary fitting curve. ;Discrete the set of points on the second lower boundary Perform the same process to generate a continuous second lower boundary fitting curve. ; Set the second central reference point Connect to generate a second center reference line. .

[0061] Fitted curve from the second upper boundary Second lower boundary fitting curve The defined continuous band region forms the characteristic band of the secondary vibration values ​​to be analyzed, wherein the characteristic band of the secondary vibration values ​​to be analyzed is located on the horizontal axis. Each value corresponds to a vertical interval. , indicating the normal fluctuation range corresponding to the vibration values ​​to be analyzed under the second characteristic quantity state.

[0062] S7: Based on the comparison results between the characteristic band of the vibration value to be analyzed and the characteristic base band of the benchmark vibration value, determine the abnormal points and the abnormal change range of the overall trend.

[0063] In this embodiment, since historical data and data to be analyzed respectively construct the characteristic bands of vibration values ​​to be analyzed and the characteristic base bands of reference vibration values ​​in three characteristic domains, the comparison needs to be performed in the same characteristic domain coordinate system.

[0064] Establish a speed domain comparison coordinate system, a converted speed domain comparison coordinate system, and an overload domain comparison coordinate system, with speed, converted speed, and overload as the abscissa and vibration value as the ordinate.

[0065] The fitting curve of the characteristic baseband of the reference vibration value in this domain is the first upper boundary fitting curve. First lower boundary fitting curve and the first center reference line The fitting curve of the characteristic band of the secondary vibration value to be analyzed in this domain is the second upper boundary fitting curve. The second lower boundary fitting curve and the second center reference line The coordinates are superimposed in the speed domain comparison coordinate system, the converted speed domain comparison coordinate system, and the overload domain comparison coordinate system, respectively.

[0066] For each discrete operating condition interval (i.e., each characteristic value) The corresponding interval will be used to compare the statistical characteristics of the vibration value characteristic band to be analyzed with the baseline statistical characteristics of the baseline vibration value characteristic band. The baseline vibration value characteristic band will be used in the [interval]. The normal fluctuation range provided by each interval is The baseline center value is The characteristic band of the secondary vibration value needs to be analyzed in the first... The normal fluctuation range provided by each interval is The central value is .

[0067] By comparison With reference boundary The relationship, and and The relative positions of the vibrations to the reference boundary are used to quantify the degree of deviation of the vibrations from the historical reference.

[0068] Please see Figure 3 As shown, the logical flowchart for determining outliers based on the comparison results between the center value and the benchmark boundary is shown in this embodiment of the invention.

[0069] If within any discrete operating condition range Less than or Greater than At this point, it is necessary to analyze the center value of the characteristic band of the secondary vibration. Exceeding the reference boundary value corresponding to the characteristic baseband of the reference vibration value, or Less than or Greater than In this case, if the overall fluctuation range of the secondary characteristic band does not overlap with the baseline band, then the intervals in these two situations are marked as outliers. Outliers indicate that under this specific operating condition, the secondary vibration characteristics to be analyzed have significantly deviated from the historical normal level, which may indicate a local fault or change in condition.

[0070] If, within multiple consecutive discrete operating condition intervals, the statistical characteristics (such as the center value) of the secondary vibration value characteristic band to be analyzed continuously exceed the benchmark boundary value corresponding to the benchmark vibration value characteristic band, and the magnitude of the exceedance shows a monotonic change or a cumulative offset trend, then the continuous interval is determined to be an interval with an abnormal overall trend.

[0071] Based on the marked anomaly points and the overall trend of abnormal changes, vibration anomaly diagnostic information is generated. This diagnostic information includes the location of the abnormal working condition interval, the quantitative value of the anomaly degree, and the trend characteristics of the anomaly as a function of the characteristic quantity. The quantitative value of the anomaly degree can be characterized by the absolute magnitude of the center value exceeding the boundary.

[0072] Please see Figure 4 As shown in the figure, in this embodiment of the invention, the effect of comparing the secondary vibration value characteristic band with the reference vibration value characteristic base band needs to be analyzed.

[0073] Specifically, the visualization of statistical distribution is used to achieve anomaly quantification and identification. By comparing the distribution characteristics of the vibration data of this operation with the historical vibration data of the entire life cycle, the current working status of the engine can be evaluated in real time.

[0074] In this embodiment, Figure 4 The horizontal axis represents the discretized relative engine speed, indicating the percentage of the current operating point relative to the maximum speed (e.g., 90 means 90% of the engine's maximum speed). Since the speed is in a continuous state of change, it is discretized into several operating intervals for statistical purposes. The vertical axis represents the vibration value, reflecting the vibration intensity of the engine at that speed. Figure 4 The blue elements represent the vibration value changes of this engine throughout all its historical operations, while the red elements represent the vibration value changes of this engine during the current operation. The blue elements include blue cuboids and blue stars, and the red elements include red cuboids and red dots.

[0075] Furthermore, the blue cuboid shape represents the distribution of the middle 50% of all historical vibration values ​​of this engine, and the horizontal line in the cuboid shape is the median of n vibration values. In this embodiment, if there are z historical vibration values, they are sorted from smallest to largest, and the distribution of the middle 50% of vibration values ​​refers to the distribution of all vibration values ​​between the z / 4th vibration value and the 3z / 4th vibration value.

[0076] The blue star graphic represents the vibration values ​​at each of the two extremes out of all historical vibration values ​​of this engine. In this embodiment, if there are w historical vibration values, they are sorted from smallest to largest. The vibration values ​​at each extreme refer to the points from the minimum vibration value to 5%*w and the points from 95%*w to the maximum vibration value.

[0077] The red cuboid represents the vibration distribution of the engine in the middle 50% of the operation, while the red dot represents the vibration distribution of the engine at each of the two ends of the operation.

[0078] The proportions of blue and red elements can be flexibly adjusted according to the needs of actual applications to ensure adaptability to different engines and operating conditions. For example, to improve the sensitivity of anomaly detection, the middle 50% of all historical vibration values ​​and the middle 50% of vibration values ​​in this operation can be reduced to 30%, and the 5% at each end of all historical vibration values ​​and the 5% at each end of vibration values ​​in this operation can be reduced to 1%. To allow for a more lenient tolerance, the middle 50% of all historical vibration values ​​and the middle 50% of vibration values ​​in this operation can be increased to 70%, and the 5% at each end of all historical vibration values ​​and the 5% at each end of vibration values ​​in this operation can be adjusted to 6%.

[0079] All technologies not mentioned in the above embodiments are existing technologies. It is understood that no specific limitation is made to any preset parameter or critical parameter in the embodiments of the present invention, and the above values ​​are not limited thereto. Those skilled in the art can adjust the preset parameters or critical parameters accordingly based on actual needs, analysis of historical data, or equipment usage.

[0080] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for analyzing vibration values, characterized in that, include: Extract historical vibration values ​​and corresponding historical engine characteristic values ​​from the original operational monitoring data of several historical batches of the engine; The historical vibration values ​​and the historical engine characteristic values ​​are transformed from the time domain to the first characteristic domain and the characteristic values ​​are discretized to obtain the first vibration value group corresponding to the first discrete working condition interval. Fitting is performed based on the first vibration value group to generate a reference vibration value feature baseband; Extract the vibration values ​​of the engine to be analyzed and the corresponding characteristic values ​​of the engine to be analyzed from the original working monitoring data of the engine to be analyzed. The vibration values ​​to be analyzed and the engine characteristic values ​​to be analyzed are transformed from the time domain to the second characteristic domain and the characteristic values ​​are discretized to obtain the second vibration value group corresponding to the second discrete working condition interval. Fitting is performed based on the second vibration value to generate the characteristic band of the vibration value to be analyzed; Based on the comparison results between the characteristic band of the vibration value to be analyzed and the characteristic base band of the benchmark vibration value, the abnormal points and the abnormal change range of the overall trend are determined.

2. The method for analyzing vibration values ​​according to claim 1, characterized in that, The process of obtaining the first vibration value group includes: Using the historical vibration value as the vertical axis and the first characteristic quantity as the horizontal axis, a first mapping coordinate system is constructed to show the change of the historical vibration value with the corresponding first characteristic quantity. The first characteristic quantity includes the first rotational speed, the first converted rotational speed, and the first overload. The first mapping coordinate system includes coordinate systems corresponding to the first rotational speed domain, the first converted rotational speed domain, and the first overload domain, respectively. For each coordinate system, according to the preset first discretization step size or interval division strategy, the value range of the first feature quantity corresponding to the corresponding coordinate system is divided into several continuous first discrete working condition intervals. All historical vibration values ​​falling within the same first discrete operating condition range are aggregated to form the first vibration value group.

3. The method for analyzing vibration values ​​according to claim 2, characterized in that, The process of generating the reference vibration value characteristic baseband includes: Statistical characteristic values ​​within the first discrete working condition interval are calculated corresponding to the first vibration value group to obtain the first maximum value, first minimum value, first median, first average value, first variance, first midpoint value, first lower quantile, first upper quantile, and first interquartile range; The reference vibration value characteristic baseband is obtained by fitting the first maximum value, the first minimum value, the first median, the first average value, the first variance, the first midpoint value, the first lower quantile, the first upper quantile, and the first interquartile range.

4. The method for analyzing vibration values ​​according to claim 1, characterized in that, The process of obtaining the second vibration value group includes: Using the vibration value to be analyzed as the vertical axis and the second characteristic quantity as the horizontal axis, a second mapping coordinate system is constructed to show the change of the vibration value to be analyzed with the corresponding second characteristic quantity. The second characteristic quantity includes the second rotational speed, the second converted rotational speed, and the second overload. The second mapping coordinate system includes the coordinate systems corresponding to the second rotational speed domain, the second converted rotational speed domain, and the second overload domain, respectively. For each coordinate system, according to the preset second discretization step size or interval division strategy, the value range of the second feature quantity corresponding to the corresponding coordinate system is divided into several continuous second discrete working condition intervals. All the vibration values ​​to be analyzed that fall within the same second discrete operating condition range are aggregated to form the second vibration value group.

5. The method for analyzing vibration values ​​according to claim 4, characterized in that, The process of generating the characteristic band of the vibration values ​​to be analyzed includes: Statistical characteristic values ​​within the second discrete working condition interval are calculated corresponding to the second vibration value group to obtain the second maximum value, second minimum value, second median, second average value, second variance, second midpoint value, second lower quantile, second upper quantile, and second interquartile range; The characteristic band of the vibration value to be analyzed is obtained by fitting the second maximum value, the second minimum value, the second median, the second average value, the second variance, the second midpoint value, the second lower quantile, the second upper quantile, and the second interquartile range.

6. The method for analyzing vibration values ​​according to claim 1, characterized in that, The comparison process between the characteristic band of the secondary vibration value to be analyzed and the characteristic base band of the reference vibration value includes: The characteristic band of the vibration value to be analyzed and the characteristic base band of the reference vibration value are superimposed and aligned in the same characteristic quantity domain coordinate system; Based on the first discrete working condition interval and the second discrete working condition interval, the statistical characteristics of the vibration value feature band to be analyzed are compared with the baseline statistical characteristics of the baseline vibration value feature base band. If, within any discrete operating condition interval, the statistical characteristics of the vibration value feature band to be analyzed exceed the reference boundary value corresponding to the reference vibration value feature base band, then the corresponding interval will be marked as an outlier. If, within multiple consecutive discrete operating condition intervals, the statistical characteristics of the vibration value characteristic band to be analyzed continuously exceed the benchmark boundary value corresponding to the benchmark vibration value characteristic baseband, and the magnitude of the exceedance shows a monotonically changing or cumulative offset trend, then the corresponding consecutive intervals are determined to be intervals with abnormal overall trends.

7. The method for analyzing vibration values ​​according to claim 6, characterized in that, Vibration anomaly diagnostic information is generated based on the anomaly points and the overall trend anomaly change range. The vibration anomaly diagnostic information includes the location of the abnormal working condition range, the quantitative value of the anomaly degree, and the trend characteristics of the anomaly changing with the characteristic quantity.

8. The method for analyzing vibration values ​​according to claim 7, characterized in that, The process of generating vibration anomaly diagnostic information also includes: The vibration distribution characteristics corresponding to the abnormal points and the abnormal change range of the overall trend are output as vibration abnormality alarm signals and abnormal operating condition range locations.

9. The method for analyzing vibration values ​​according to claim 1, characterized in that, The process of cleaning the raw work monitoring data includes: Before obtaining the historical vibration values ​​and the historical engine characteristic values, data cleaning is performed on several historical batches of the original working monitoring data to remove abnormal record values. Before obtaining the vibration value and engine characteristic value of the analysis session, the original working monitoring data of the analysis session is cleaned to remove abnormal recorded values.

10. The method for analyzing vibration values ​​according to claim 9, characterized in that, The data cleaning includes: Remove abnormal record values ​​from the original work monitoring data caused by sensor failure or duplicate timestamps.

Images