Devices for equipment monitoring
The apparatus and method utilize statistical process algorithms to analyze spectral sensor data, addressing the challenge of variable detection thresholds in rotating equipment monitoring by detecting and identifying the cause of abnormal vibrations, thereby enhancing maintenance efficiency.
Patent Information
- Application Number
- CN202080023115.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2019-03-22
- Filing Date
- 2020-03-18
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2040-03-18
AI Technical Summary
Current methods for monitoring rotating equipment in the process industry face challenges in setting detection thresholds that vary with load and rotational speed, making them cumbersome and time-consuming.
An apparatus and method using statistical process algorithms, such as Hotelling's T2 statistic and squared prediction error (Q-statistic), to analyze spectral sensor data and identify frequency ranges associated with equipment degradation, enabling real-time monitoring and fault detection.
Provides enhanced awareness of equipment degradation by detecting abnormal vibrations and identifying the primary cause, allowing for timely maintenance and reducing downtime.
Smart Images

Figure CN113557414B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to an apparatus for monitoring a device, a system for monitoring a device, a method for monitoring a device, and to a computer program element and a computer-readable medium. Background Art
[0002] Currently in the process industry, the inspection and maintenance of equipment with moving parts (e.g., rotating parts) contributes primarily to operating expenses. For example, condition monitoring of rotating equipment typically relies on the collection and analysis of various types of sensors (e.g., acceleration, velocity, and displacement sensors, microphones, acoustic arrays, torque meters, encoders, tachometers, image cameras, fiber optic sensors, thermal sensors, and stress sensors). Signal processing-based algorithms are then used to calculate health indicators for various failure modes of various equipment types (e.g., pumps, gearboxes, and bearings). These health indicators are then tracked to detect or extrapolated to predict specific equipment failures.
[0003] Several solutions for the monitoring of rotating equipment are known. These typically rely on monitoring a simple index, such as ISO 10186-1,7, which reflects the overall vibration level and compares it to predefined thresholds that are usually set conservatively.
[0004] Indicators associated with signal processing methods that are related to specific faults of specific components (e.g., bearings, gearboxes, pumps) have also been implemented. However, industrial use of such advanced indicators still suffers from the necessity to define specific detection thresholds for each rotating equipment that vary with load, rotational speed, and other operating parameters. The implementation of these methods is cumbersome and time-consuming due to the lack of the ability to systematically select such thresholds.
[0005] There is a need to solve this problem. Summary of the invention
[0006] It would therefore be advantageous to have an improved ability to monitor equipment having moving parts in order to detect whether the equipment is degraded or about to degrade or become damaged.
[0007] The objects of the invention are solved by the subject-matter of the independent claims, wherein further embodiments are incorporated in the dependent claims.
[0008] In a first aspect, there is provided an apparatus for equipment monitoring, the apparatus comprising:
[0009] - Input unit;
[0010] - Processing unit; and
[0011] - Output unit.
[0012] The input unit is configured to provide a plurality of batches of time sensor data for an item of the device to the processing unit. Each batch of the time sensor data includes a plurality of time sensor values as a function of time. The processing unit is configured to process the plurality of batches of the time sensor data to determine a plurality of batches of spectral sensor data. Each batch of the spectral sensor data includes a plurality of spectral sensor values as a function of frequency. The processing unit is configured to implement at least one statistical process algorithm to process a plurality of spectral sensor values for a plurality of batches of spectral sensor data in order to determine a plurality of indicator values. For each batch of spectral sensor data, there is an indicator value determined by each statistical process algorithm. Each statistical process algorithm has an associated threshold, and the processing unit is configured to determine a batch of spectral sensor data of interest, which has an indicator value greater than the threshold for the associated statistical process algorithm, using at least one threshold and a plurality of indicator values. The processing unit is configured to determine a frequency range of interest based on a plurality of spectral sensor values for a batch of spectral sensor data of interest.
[0013] In this manner, an unsupervised technique for monitoring industrial equipment (e.g., rotating equipment) is achieved. An abnormal increase in the overall level of a sensed value (e.g., vibration) is detected, and the spectral components that are most correlated with the detected abnormal increase in the sensed value are isolated. Thus, an analyst can have a first indication of the type of fault affecting the equipment based on knowledge of the spectral components that contribute to the increase in the overall sensed value level (e.g., vibration of the rotating equipment).
[0014] In other words, a higher level of awareness of simple changes in sensed values (e.g., vibrations) is provided to the above-described diagnostic system, wherein the new technology enables abnormal sensed value levels to be detected and the associated spectrum (frequency components) determined as the primary cause of that abnormality to be identified.
[0015] In an example, a time period between adjacent batches of temporal sensor data is greater than a time period between adjacent sensor data within a batch.
[0016] In an example, the at least one statistical process algorithm includes Hotelling statistics.
[0017] In an example, for each batch of spectral sensor data, there is an index value determined by Hotelling statistics.
[0018] In an example, a batch of spectral sensor data of interest is determined when a value of an indicator determined by the Hotelling statistic for that batch is greater than a threshold associated with the Hotelling statistic.
[0019] In an example, at least one statistical process algorithm includes a squared prediction error or Q-statistic.
[0020] In an example, for each batch of spectral sensor data, there is an index value determined by the squared prediction error or Q-statistic.
[0021] In an example, a batch of spectral sensor data of interest is determined when a metric value determined for that batch by the squared prediction error or the Q-statistic is greater than a threshold associated with the squared prediction error or the Q-statistic.
[0022] In an example, a batch of spectral sensor data of interest is determined when an indicator value determined for that batch by the Hotelling statistic is greater than a threshold associated with the Hotelling statistic; or wherein a batch of spectral sensor data of interest is determined when an indicator value determined for that batch by the squared prediction error or Q statistic is greater than a threshold associated with the squared prediction error or Q statistic.
[0023] In other words, both indicators (determined using SPE or Q statistics or T2 Hotelling statistics) are monitored separately, and if one of them exceeds its limit, analysis of the corresponding batch is performed to determine the problematic spectral frequency.
[0024] In other words, two individual statistical control charts (calculated indices or indices values for batches) are used to detect abnormal observations by comparing them to threshold limits. Hotelling's T2 statistic is used in the principal component space, while the squared prediction error (SPE or Q) statistic is used in the remaining space.
[0025] In an example, the determining of the plurality of batches of spectral sensor data includes utilizing a Fourier transform algorithm on temporal sensor values for each of the plurality of batches of temporal sensor data.
[0026] In an example, the processing unit is configured to subdivide the spectral values of the batch for the spectral data of interest into a plurality of frequency ranges.The frequency range of interest is determined as a frequency range exhibiting values greater than values associated with other frequency ranges.
[0027] In a second aspect, there is provided a system for equipment monitoring, the system comprising: at least one sensor configured to acquire a plurality of batches of temporal sensor data; and the apparatus for equipment monitoring according to the first aspect.
[0028] In a third aspect, there is provided a method for equipment monitoring, the method comprising:
[0029] a) providing a plurality of batches of time sensor data for an item of equipment, wherein each batch of the time sensor data comprises a plurality of time sensor values as a function of time;
[0030] b) processing the plurality of batches of temporal sensor data to determine a plurality of batches of spectral sensor data, wherein each batch of spectral sensor data comprises a plurality of spectral sensor values as a function of frequency;
[0031] c) implementing at least one statistical process algorithm to process a plurality of spectral sensor values for a plurality of batches of spectral sensor data to determine a plurality of indicator values, wherein for each batch of spectral sensor data there is each indicator value determined by the statistical process algorithm;
[0032] d) determining, using the threshold value and the plurality of indicator values for each statistical process algorithm, batches of spectral sensor data of interest having indicator values greater than the threshold value for the associated statistical process algorithm; and
[0033] e) determining a frequency range of interest based on a plurality of spectral sensor values for the batch of spectral sensor data of interest.
[0034] According to a further aspect, a computer program element is provided for controlling an apparatus or a system as described above, which computer program element is suitable for carrying out the method steps as described above when it is executed by a processing unit.
[0035] According to another aspect, there is also provided a computer readable medium having stored thereon a computer element as previously described.
[0036] The above aspects and examples will be apparent from and explained with reference to the embodiments described hereinafter. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] Exemplary embodiments will be described below with reference to the following drawings:
[0038] Figure 1 An example showing the principal component space spanned by two principal components and the remaining space for a data set of three process variables;
[0039] Figure 2 a schematic representation showing examples of determined indicators for different batches calculated using the T2 Hotelling statistic or calculated using the PRE or Q statistics for a statistical limit for one of those statistics; and
[0040] Figure 3 Examples of spectral frequency energy spectra for batches exceeding the statistical limit are shown. DETAILED DESCRIPTION
[0041] In reference Figure 1-3In the case of, the apparatus, system and method for equipment monitoring currently provided are now described in detail. An example of an apparatus for equipment monitoring includes an input unit, a processing unit and an output unit. The input unit is configured to provide a plurality of batches of time sensor data for an item of equipment to the processing unit. Each batch of time sensor data includes a plurality of time sensor values as a function of time. The processing unit is configured to process the plurality of batches of time sensor data to determine a plurality of batches of spectrum sensor data. Each batch of spectrum sensor data includes a plurality of spectrum sensor values as a function of frequency. The processing unit is configured to implement at least one statistical process algorithm to process a plurality of spectrum sensor values for a plurality of batches of spectrum sensor data in order to determine a plurality of indicator values. For each batch of spectrum sensor data, there is an indicator value determined by each statistical process algorithm. Each statistical process algorithm has an associated threshold, and the processing unit is configured to determine a batch of spectrum sensor data of interest using at least one threshold and a plurality of indicator values, which has an indicator value greater than the threshold for the associated statistical process algorithm. The processing unit is configured to determine a frequency range of interest based on a plurality of spectrum sensor values for a batch of spectrum sensor data of interest.
[0042] Thus, the device can operate in real time and be at the site of a device and monitor that device. Alternatively, the device can analyze previously acquired data to determine abnormal behavior and what frequencies are problematic.
[0043] According to an example, a time period between adjacent batches of temporal sensor data is greater than a time period between adjacent sensor data within a batch.
[0044] In an example, the time period between adjacent batches of temporal sensor data is one of the following: 1 hour; 2 hours; 3 hours; 4 hours; 5 hours; 6 hours; 7 hours; 8 hours; 12 hours; 24 hours. The time period between batches can be different from the time periods described above.
[0045] In an example, a time period between adjacent sensor data within a batch is one of the following: 0.0001s; 0.0005s; 0.001s; 0.002s; 0.003s; 0.004s; 0.005s; 0.01s; 0.02s; 0.05s.
[0046] In an example, each batch of sensor data is acquired within one of the following time periods: 10s; 20s; 30s; 60s; 120s.
[0047] Thus, for example, a sensor may acquire data at a rate of 1 kHz for 60 seconds. Then wait 6 hours, and again acquire data at a rate of 1 kHz for 60 seconds. However, it is possible to utilize different rates of sensor data acquisition over different time periods, and have different durations between batches of sensor data.
[0048] According to an example, the at least one statistical process algorithm comprises Hotelling statistics.
[0049] By way of example, for each batch of spectral sensor data, there is an index value determined by Hotelling statistics.
[0050] According to an example, a batch of spectral sensor data of interest is determined when a value of an indicator determined by the Hotelling statistic for that batch is greater than a threshold value associated with the Hotelling statistic.
[0051] According to an example, at least one statistical process algorithm includes a squared prediction error or Q-statistic.
[0052] By way of example, for each batch of spectral sensor data, there is an index value determined by the squared prediction error or Q-statistic.
[0053] According to an example, a batch of spectral sensor data of interest is determined when a value of a metric determined for that batch by the squared prediction error or the Q-statistic is greater than a threshold associated with the squared prediction error or the Q-statistic.
[0054] According to an example, a batch of spectral sensor data of interest is determined when an indicator value determined for that batch by Hotelling statistics is greater than a threshold associated with Hotelling statistics. Alternatively, a batch of spectral sensor data of interest is determined when an indicator value determined for that batch by squared prediction error or Q statistics is greater than a threshold associated with squared prediction error or Q statistics.
[0055] According to an example, the determining of the plurality of batches of spectral sensor data includes utilizing a Fourier transform algorithm on temporal sensor values for each of the plurality of batches of temporal sensor data.
[0056] In an example, the Fourier transform algorithm is a Fast Fourier Transform algorithm.
[0057] According to an example, the processing unit is configured to subdivide the spectral values of the batch of spectral data of interest into a plurality of frequency ranges, and wherein the frequency range of interest is determined as a frequency range exhibiting larger values than values associated with other frequency ranges.
[0058] In an example, the frequency range of interest is determined as a frequency range that exhibits spectral power values that are larger than spectral power values associated with other frequency ranges.
[0059] In an example, the sensor data is rotation sensor data.
[0060] It will therefore be appreciated that the above-described apparatus, when coupled to at least one sensor that acquires batches of temporal sensor data, provides a system for equipment monitoring.
[0061] Furthermore, an example relates to a method for equipment monitoring, the method comprising:
[0062] a) providing a plurality of batches of time sensor data for an item of equipment, wherein each batch of the time sensor data comprises a plurality of time sensor values as a function of time;
[0063] b) processing the plurality of batches of temporal sensor data to determine a plurality of batches of spectral sensor data, wherein each batch of spectral sensor data comprises a plurality of spectral sensor values as a function of frequency;
[0064] c) implementing at least one statistical process algorithm to process a plurality of spectral sensor values for a plurality of batches of spectral sensor data to determine a plurality of indicator values, wherein for each batch of spectral sensor data there is each indicator value determined by the statistical process algorithm;
[0065] d) using the threshold value for each statistical process algorithm and the plurality of indicator values to determine batches of spectral sensor data of interest having an indicator value greater than the threshold value for the associated statistical process algorithm; and
[0066] e) determining a frequency range of interest based on a plurality of spectral sensor values for the batch of spectral sensor data of interest.
[0067] In an example, a time period between adjacent batches of temporal sensor data is greater than a time period between adjacent sensor data within a batch.
[0068] In an example, the time period between adjacent batches of temporal sensor data is one of: 1 hour; 2 hours; 3 hours; 4 hours; 5 hours; 6 hours; 7 hours; 8 hours; 12 hours; 24 hours.
[0069] In an example, a time period between adjacent sensor data within a batch is one of the following: 0.0001s; 0.0005s; 0.001s; 0.002s; 0.003s; 0.004s; 0.005s; 0.01s; 0.02s; 0.05s.
[0070] In an example, each batch of sensor data is acquired within one of the following time periods: 10s; 20s; 30s; 60s; 120s.
[0071] In an example, in step c), the at least one statistical process algorithm comprises Hotelling statistics.
[0072] In an example, for each batch of spectral sensor data, there is an index value determined by Hotelling statistics.
[0073] In an example, in step d), a batch of spectral sensor data of interest is determined when a value of an indicator determined for that batch by the Hotelling statistic is greater than a threshold associated with the Hotelling statistic.
[0074] In an example, in step c), the at least one statistical process algorithm comprises a squared prediction error or a Q-statistic.
[0075] In an example, for each batch of spectral sensor data, there is an index value determined by the squared prediction error or Q-statistic.
[0076] In an example, in step d), a batch of spectral sensor data of interest is determined when a value of an indicator determined for that batch by the squared prediction error or the Q-statistic is greater than a threshold associated with the squared prediction error or the Q-statistic.
[0077] In an example, a batch of spectral sensor data of interest is determined when an index value determined for that batch by the Hotelling statistic is greater than a threshold associated with the Hotelling statistic. Alternatively, a batch of spectral sensor data of interest is determined when an index value determined for that batch by the squared prediction error or Q statistic is greater than a threshold associated with the squared prediction error or Q statistic.
[0078] In an example, step b) comprises utilizing a Fourier transform algorithm on the temporal sensor values for each batch of the plurality of batches of temporal sensor data.
[0079] In an example, the Fourier transform algorithm is a Fast Fourier Transform algorithm.
[0080] In an example, the sensor data is rotation sensor data.
[0081] The following detailed description relates to the monitoring of rotating equipment. In a new way, multivariate statistical process control is used in a new way. MSPC has been used to monitor industrial processes and to support process operators or process engineers in troubleshooting process anomalies - see Kresta, JV, Macgregor, JF and Marlin, TE (1991) "Multivariate statistical monitoring of process operating performance" (The Canadian Journal of Chemical Engineering, 69 (1), 35-47.)
[0082] However, variations of this approach have been used for monitoring equipment, such as rotating equipment. While the standard MSPC approach uses time series collected from process sensors, it has been found that rotating equipment can be better characterized by spectral features. It has been found that faults or malfunctions in rotating equipment have features that are localized in the frequency domain but not necessarily in the time domain due to periodicity caused by rotation.
[0083] In summary, in the new technique, the T2 Hotelling metric and the Q or squared prediction error (SPE) metric are calculated for each batch of collected sensor values and compared to appropriate statistical limits. In a specific example, the "vibration" measurement is performed in a "batch" manner, that is, the measurements are recorded periodically over a given duration. For example, acceleration / velocity sensor data measurements are recorded every 6 hours. Below, reference is made to a time index k, where k=1, 2, 3, ..., n represents a batch of data acquired at a relative time of 6 hours, 12 hours, 18 hours, ..., n×6 hours at a rate of 1 KHz over a duration of, for example, one minute. Therefore, in this specific example, each batch or record of data acquired every 6 hours has 1000×60=60,000 data samples or values.
[0084] The T2 Hotelling and Q contribution plots are determined from one or more identified batches of the plurality of batches acquired every 6 hours. The T2 and Q contribution plots are then used to evaluate the frequency bins that are most correlated with the T2 and Q index deviations from their corresponding statistical limits. This indication allows the user to monitor changes in the spectrum of the vibration signal and helps the user understand the type of faults that the rotating equipment is experiencing.
[0085] Therefore, two individual statistical control charts are used to detect anomalous observations by comparing them to threshold limits. Hotelling's T2 statistic is used in the principal component space and the squared prediction error (SPE or Q) statistic is used in the remaining space.
[0086] Monitoring in principal component space
[0087] When the scores follow a Gaussian multivariate distribution, the Hotelling statistic is used. The scores have zero mean, and the estimated sample covariance matrix is is a diagonal matrix. , the Hotelling statistic for the t-score is given by
[0088]
[0089] where P is a matrix whose columns are the loading vectors, NPC is the number of principal components retained in the model, is a diagonal matrix whose elements are the eigenvalues of the sample covariance matrix (singular values of the data matrix X) retained in the model in descending order of magnitude. T is the score in the principal component space.
[0090] Hotelling The statistic is the sum of the scaled scores. It combines information from all scores into a single metric. Only loadings corresponding to larger singular values are included in the statistic. Smaller singular values corresponding to noise are inverted in the calculation of this statistic. Not including these smaller singular values allows for a better representation of process behavior and robust anomaly detection within the model. The statistic is the distance between the projection of the observation into the principal component space and the origin of the principal component space. Threshold Use the following formula to calculate
[0091]
[0092] in is the confidence level and N is the number of observations in the data matrix. is a Fisher distribution with NPC and N-NPC degrees of freedom Key point. From the above formula, we can get the control limit of the score
[0093]
[0094] The above formula describes the elliptical area used for the confidence of the score (for the example with m=3, NPC=2 see Figure 1 ). For a given observed variable pair The individual contribution of
[0095]
[0096] Monitoring in the remaining space
[0097] Monitoring in the remaining space uses the Q statistic defined as,
[0098]
[0099] in is the row vector of the error matrix E. The Q statistic is the Euclidean norm of the deviations of the observations from their projections onto the principal component space. The control limit of Q can be approximated by
[0100]
[0101] in is with The value of the normal distribution corresponding to the percentile,
[0102] and .
[0103] Control Limits represents a threshold for random variation in the process. The individual contribution of a variable to Q at a given observation is given by the element of E corresponding to the observation.
[0104] Reference Figure 1 , which shows a data set consisting of three variables. It can be seen that the three-dimensional data points (open circles) can be reduced to two dimensions spanned by the principal components, with the first principal component (along the major axis of the ellipse) and the second principal component (the minor axis of the ellipse) having higher explained variance. Consider a new observation (filled circle labeled PC1), which is located in the two-dimensional space spanned by the principal components, but is located at the score The new data points marked by the other solid circles are found to be acceptable when projected into the principal component space, even though the data points are different from the other points in the model. This change is captured by the Q statistic.
[0105] Therefore, multivariate statistical process control (MSPC) has been used to monitor industrial equipment in a new way. Principal component analysis (PCA) can be used. PCA allows the data set to be represented on a lower dimensional space. In addition, it separates the observation space into two subspaces. One subspace captures the process trend, while the other subspace captures the effects of random noise or new abnormal changes that are not part of the model.
[0106] Assume data matrix There are n observations collected from m sensors. PCA is a data reduction technique that extracts the maximum variance in the data in orthogonal directions called principal components. These principal components are linear combinations of variables that contain useful information (variability of the process data). PCA decomposes the data matrix into orthogonal vectors called loading vectors (p) and score vectors (t). The decomposition is performed using singular value decomposition (SVD) of the data matrix X:
[0107]
[0108] in Contains the real non-negative singular values in decreasing order of magnitude ( ). Right eigenvector is the load vector. In order to avoid modeling the noise present in the data, only the r larger eigenvalues and their corresponding load vectors are retained. The amount of variance captured in the direction of principal component i is The data matrix X, when projected onto the space formed by the individual load vectors (p), is
[0109]
[0110] in is called the score matrix, and the columns of t are orthogonal. The variance captured by the first principal component of X is greater than that captured by the second column The variance captured should be large. The score is projected to the original m-dimensional space.
[0111]
[0112] in It is modeling information. and The difference between is called the residual matrix E, and the space spanned by the residual matrix is called the residual space. This residual space corresponds to the variance in the smaller eigenvalues that are not included in the model.
[0113]
[0114] To test the new observation vector , the vector is projected onto the model. The projection of the test vector onto the space formed by the load vectors gives the score of the test data.
[0115]
[0116] Therefore, returning to the acquisition and processing of sensor data, let A vector representing collected sensor values, such as accelerometer measurements collected over time interval T starting at time k.
[0117]
[0118] set up is calculated in the time interval (k, k+T) using, for example, the Fast Fourier Transform (FFT) algorithm For each batch of collected sensor values, k=1, 2, 3, 4, ... n, where n is the number of batches at m frequency channels. As described above, each batch can be acquired every 6 hours, and each batch can have, for example, 6000 data samples acquired every 0.001 seconds. However, each batch can be acquired with a different time period between them, so that, for example, there are 2 hours between two batches, and then the next batch is acquired, for example, after 6 hours. However, the sensor values are read every Ts seconds (for example, every 0.001s) within a batch.
[0119] A batch is thus a series of sensor values collected during a time interval T. For each batch, the Fourier transform is calculated in order to estimate the corresponding power spectrum. The actual implementation of the power spectrum relies on a numerical algorithm that calculates the Fourier transform: the Fast Fourier Transform algorithm (FFT). The FFT calculates a discrete spectrum, i.e., the values of the spectrum at a finite number of frequencies. The frequencies or frequency bands at which the power spectrum is evaluated are called "frequency channels".
[0120] Superimposing the calculated spectra for all batches yields the following matrix
[0121]
[0122] The principal component (PCA) decomposition of X yields the following approximation:
[0123]
[0124] in
[0125]
[0126] and E is the residual (approximation) error, t is the vector of scores, and p is the vector of loadings.
[0127] The T2 Hotelling index and the Q(SPE) index are then calculated for each batch index and compared with appropriate statistical limits. Figure 2 In Figure 2 In , the T2 Hotelling index and the PRE or Q index are calculated for each batch, where k is the batch (dataset index) as described above. Figure 2 As shown in , batch 12 exceeds the limit, where only one set of index values calculated for the T2 Hotelling statistic and only one threshold are shown for simplicity, but there are actually two plots, one for the T2 Hotelling statistic and one for the PRE or Q statistic analysis. Therefore, as shown, batch k=12 is above the statistical limit T2 lim or Q lim This batch is above the statistical limit T2 lim or Q lim Other batches of are associated with abnormal vibration levels. Therefore, these data sets (batches) are then investigated by analyzing the contribution plots in order to determine which frequency(s) best explain the high values of T2 or Q. Such contribution plots are Figure 3 12 and is determined from the FFT determination of the spectral content for that batch with abnormal vibration levels. In this way, the T2 and Q contribution plots are used to evaluate the frequency bins that are most correlated with the T2 and Q index deviations corresponding to the statistical limits. Figure 3, the frequency bin [100 Hz-150 Hz] for batch 12 is significantly higher than the other spectral components, indicating that the detected change in the Q or T2 index is explained by the presence of components in the [100 Hz-150 Hz] frequency range. This indication allows the user to monitor changes in the spectrum of the vibration signal and helps him understand the type of fault that the rotating equipment is experiencing.
[0128] In a further exemplary embodiment, a computer program or a computer program element is provided, which is characterized in that it is configured to execute the method steps of the method according to one of the preceding exemplary embodiments on a suitable system.
[0129] Therefore, the computer program element can be stored on a computer unit, which can also be a component of an embodiment. This computing unit can be configured to perform or cause the steps of the above method to be performed. In addition, it can be configured to operate the components of the above device and / or system. The computing unit can be configured to automatically operate and / or execute the user's command. The computer program can be loaded into the working memory of the data processor. Therefore, the data processor can be equipped to perform the method described in one of the aforementioned embodiments.
[0130] According to a further exemplary embodiment of the present invention, a computer-readable medium (eg a CD-ROM) is proposed, wherein the computer-readable medium has a computer program element stored thereon, the computer program element being described by way of the preceding section.
[0131] Although the present invention has been illustrated and described in detail in the drawings and the foregoing description, such illustration and description are to be considered illustrative or exemplary rather than restrictive. The present invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and implemented by those skilled in the art in practicing the claimed invention upon study of the drawings, the present disclosure, and the appended claims.
Claims
1. A device for equipment monitoring, the device comprising: - input unit; - processing unit; as well as - output unit; wherein the input unit is configured to provide the processing unit with a plurality of batches of temporal sensor data for an item of equipment, wherein each batch of temporal sensor data comprises a plurality of temporal sensor values as a function of time; wherein the processing unit is configured to process the plurality of batches of temporal sensor data to determine a plurality of batches of spectral sensor data, wherein each batch of spectral sensor data comprises a plurality of spectral sensor values as a function of frequency; wherein the processing unit is configured to implement at least one statistical process algorithm to process the plurality of spectral sensor values for the plurality of batches of spectral sensor data to determine a plurality of indicator values, wherein for each batch of spectral sensor data there is an indicator value determined by each of the statistical process algorithms, the indicator value being determined from an FFT-determined energy spectral content of that batch; wherein each statistical process algorithm has an associated threshold, and wherein the processing unit is configured to utilize at least one threshold and the plurality of indicator values to determine batches of spectral sensor data of interest having indicator values greater than the threshold for the associated statistical process algorithm; wherein the processing unit is configured to subdivide the batch of spectral values of the spectral data of interest into a plurality of frequency ranges; wherein the processing unit is configured to determine a frequency range of interest based on the plurality of spectral sensor values for the batch of spectral sensor data of interest; wherein the frequency range of interest is determined as a frequency range that exhibits values greater than values associated with other frequency ranges; and Wherein the output unit is configured to output an indication of the frequency range of interest to a user.
2. The device according to claim 1, wherein: The time period between adjacent batches of temporal sensor data is greater than the time period between adjacent sensor data within a batch.
3. The device according to claim 1, wherein: The at least one statistical process algorithm comprises Hotelling statistics.
4. The device according to claim 3, wherein: For each batch of spectral sensor data, there is an index value determined by the Hotelling statistic.
5. The device according to claim 3, wherein: The batch of spectral sensor data of interest is determined when the indicator value determined by the Hotelling statistic for that batch is greater than the threshold associated with the Hotelling statistic.
6. The device according to claim 5, wherein: The at least one statistical process algorithm comprises a squared prediction error or Q-statistic.
7. The device according to claim 6, wherein: For each batch of spectral sensor data, there is an index value determined by the squared prediction error or Q-statistic.
8. The device according to any one of claims 6 to 7, wherein: The batch of spectral sensor data of interest is determined when the index value determined for that batch by the squared prediction error or Q-statistic is greater than the threshold associated with the squared prediction error or Q-statistic.
9. The device of claim 8, wherein: The batch of spectral sensor data of interest is determined when the indicator value determined for that batch by the Hotelling statistic is greater than the threshold associated with the Hotelling statistic; or the batch of spectral sensor data of interest is determined when the indicator value determined for that batch by the squared prediction error or Q statistic is greater than the threshold associated with the squared prediction error or Q statistic.
10. The device according to any one of claims 1 to 7, wherein: The determining of the plurality of batches of spectral sensor data includes utilizing a Fourier transform algorithm on the temporal sensor values for each of the plurality of batches of temporal sensor data.
11. A system for equipment monitoring, the system comprising: - at least one sensor; as well as -A device for equipment monitoring according to any one of claims 1 to 10; Wherein the at least one sensor is configured to acquire the plurality of batches of temporal sensor data.
12. A method for equipment monitoring, the method comprising: a) providing a plurality of batches of temporal sensor data for an item of equipment, wherein each batch of temporal sensor data comprises a plurality of temporal sensor values as a function of time; b) processing the plurality of batches of temporal sensor data to determine a plurality of batches of spectral sensor data, wherein each batch of spectral sensor data comprises a plurality of spectral sensor values as a function of frequency; c) implementing at least one statistical process algorithm to process a plurality of spectral sensor values for a plurality of batches of spectral sensor data in order to determine a plurality of indicator values, wherein for each batch of spectral sensor data there is an indicator value determined by each of the statistical process algorithms, wherein the processing unit is configured to subdivide the spectral values of said batch of spectral data of interest into a plurality of frequency ranges; d) using the threshold value of each statistical process algorithm and the plurality of indicator values to determine a batch of spectral sensor data of interest having an indicator value greater than the threshold value of the associated statistical process algorithm; e) determining a frequency range of interest based on the plurality of spectral sensor values for the batch of spectral sensor data of interest; wherein the frequency range of interest is determined as a frequency range that exhibits values greater than values associated with other frequency ranges; and f) outputting an indication of said frequency range of interest to a user.
13. A computer program product, comprising a computer program, which, when executed by a processor, is configured to control the apparatus according to any one of claims 1 to 10 and / or the system according to claim 11 to perform the method according to claim 12.
14. A computer-readable medium storing a computer program, wherein when the computer program is executed by a processor, the computer program is configured to control the apparatus according to any one of claims 1 to 10 and / or the system according to claim 11 to perform the method according to claim 12.
Citation Information
Patent Citations
Diagnostic system for rotary machine
JP2015021901A
