Working condition threshold value automatic defining method suitable for dynamic equipment and application

Abstract

The invention relates to a working condition threshold value automatic defining method suitable for dynamic equipment and application. The method comprises the steps that a monitoring model is constructed, training data are collected and preprocessed, and a multi-dimensional measurement data set is obtained; training a monitoring model based on different working conditions by using the multi-dimensional measurement data set as the sample; obtaining a multi-dimensional measurement data set of the dynamic equipment, inputting the trained monitoring model, and obtaining a definition result of a working condition threshold value; the method is applied to consoles of all devices, and when the devices continuously operate in a preset time range and the working conditions change along with operation, the consoles collect multi-dimensional measurement data sets of the devices, obtain an automatic definition result of a working condition threshold value and obtain an SPC score. The method is high in adaptability, high in monitoring precision, capable of more accurately describing data distribution characteristics under different working conditions, high in generalization ability, capable of more comprehensively reflecting the operation state of the machine equipment and improving the monitoring robustness and generalization ability, and capable of being widely applied to state monitoring of the equipment.

Description

Technical Field

[0001] This invention relates to the technical field of electronic digital data processing, and in particular to a method and application for automatically defining operating thresholds for moving equipment. Background Technology

[0002] In industrial production, monitoring the operational status of machinery and equipment is crucial. Various sensors deployed on key equipment collect critical operating parameters in real time, such as vibration, temperature, pressure, and rotational speed. This not only allows staff to quickly understand the equipment's status, but the built-in intelligent analysis modules in existing technologies can also perform in-depth analysis of the collected data, identify abnormal equipment conditions, and automatically determine whether the equipment is in normal working condition by comparing historical data, setting thresholds, and using machine learning algorithms. This predicts potential faults and issues early warnings, providing accurate data support for equipment maintenance and production optimization.

[0003] However, traditional monitoring methods are usually based on fixed thresholds or models, which are difficult to adapt to scenarios where machines operate under changing conditions. When operating conditions change, characteristic signals such as vibration, sound, and magnetic flux usually change significantly, and the original thresholds or models are very likely no longer applicable to the current operating conditions. This often leads to misjudgments or omissions in monitoring the machine's operating status, resulting in losses of manpower, material resources, and financial resources.

[0004] Therefore, how to effectively conduct statistical monitoring under varying operating conditions has become an urgent problem to be solved. Summary of the Invention

[0005] This invention solves the problems existing in the prior art and provides a method and application for automatically defining the operating condition threshold of moving equipment.

[0006] The technical solution adopted in this invention is an automatic threshold definition method for operating equipment, wherein the method constructs a monitoring model;

[0007] Training data is collected and preprocessed to obtain multi-dimensional measurement data sets; the measurement data here is generated under real working conditions, which ensures the reliability of the monitoring model obtained after training.

[0008] The monitoring model is trained based on different working conditions using a multi-dimensional measurement data set as a sample.

[0009] Acquire multi-dimensional measurement data sets of the moving equipment, input them into the trained monitoring model, and obtain the definition results of the working condition threshold.

[0010] Preferably, the training data and operational data are characterization signal data during equipment operation and monitoring feature data at corresponding times.

[0011] Preferably, the characterization signal data includes vibration acceleration signal, sound signal, and magnetic flux signal, and the monitoring feature data includes loudness value and temperature value at the corresponding time; the measurement value of the characterization signal data is obtained, and a multi-dimensional measurement data group is established with the monitoring feature data.

[0012] Preferably, the monitoring model includes a working condition identification module, a working condition statistics module, and a threshold adjustment module arranged sequentially.

[0013] Preferably, the working condition identification module performs clustering based on one or more data points in the multi-dimensional measurement data group to confirm the working condition category.

[0014] Preferably, the working condition statistics module calculates the statistics of each measurement value under each working condition.

[0015] Preferably, based on the statistics, the threshold adjustment module adjusts the early warning monitoring threshold and the alarm monitoring threshold.

[0016] In this invention, the identification of working condition categories can be effectively defined using a small number of detection values. On the one hand, for edge devices with only tens or even a few megabytes of memory, clustering based on a single metric can greatly reduce memory usage compared to clustering based on multiple metrics. On the other hand, the effective value of vibration velocity is sufficient to distinguish the category of working condition, thus ensuring accuracy.

[0017] Preferably, based on the monitoring model, a pre-defined polynomial model is used to fuse the scores of all measures to obtain the final SPC score, which is used as the SPC monitoring result. The SPC monitoring result is used to help users understand the current health status of the equipment. Different thresholds are needed to define whether the machine is operating normally under different operating conditions, so as to avoid the situation where the machine is operating normally but the monitoring result shows that the machine is operating abnormally. Therefore, the SPC score is used to represent the monitoring results of each measure under different thresholds, that is, to obtain the definition result of the operating condition threshold.

[0018] An application of the aforementioned method for automatically defining operating condition thresholds for moving equipment is applied to the control console of all the aforementioned equipment. When the equipment operates continuously within a preset time range and the operating conditions change with operation, the control console collects multi-dimensional measurement data sets of the equipment to obtain the automatic definition result of the operating condition thresholds and obtains the SPC score.

[0019] This invention relates to a method and application for automatically defining operating condition thresholds for moving equipment. The method constructs a monitoring model, collects and preprocesses training data to obtain a multi-dimensional measurement data set; uses the multi-dimensional measurement data set as samples to train the monitoring model based on different operating conditions; acquires the multi-dimensional measurement data set of the moving equipment, inputs it into the trained monitoring model, and obtains the definition result of the operating condition threshold; the method is applied to the control console of all the equipment. When the equipment runs continuously within a preset time range and the operating conditions change with operation, the control console collects the multi-dimensional measurement data set of the equipment, obtains the automatic definition result of the operating condition threshold, and obtains the SPC score.

[0020] The beneficial effects of this invention are as follows:

[0021] (1) Highly adaptable; Through historical data clustering, it can automatically identify and distinguish different working conditions without the need for manual setting of working condition modes, and is suitable for scenarios where machines run continuously and the working conditions do not change in a fixed pattern.

[0022] (2) High monitoring accuracy; monitoring models are established for each working condition category, which can more accurately describe the data distribution characteristics under different working conditions and improve monitoring accuracy;

[0023] (3) Strong generalization ability; based on multiple measures of multi-dimensional data (vibration, sound, magnetic flux), comprehensive monitoring can more comprehensively reflect the operating status of machinery and equipment, and improve the robustness and generalization ability of monitoring.

[0024] (4) It can be widely used in the condition monitoring of various rotating machinery, reciprocating machinery and other equipment. Attached Figure Description

[0025] Figure 1 This is a flowchart of the method of the present invention;

[0026] Figure 2 This is a schematic diagram of the monitoring model in this invention;

[0027] Figure 3 The graph shows the y-axis vibration acceleration, y-axis vibration displacement, and y-axis vibration velocity curves of the induced draft fan under two alternating operating conditions in an embodiment of the present invention.

[0028] Figure 4 The clustering results are obtained after performing cluster analysis using historical data of the y-axis vibration velocity of the induced draft fan.

[0029] Figure 5 To construct a monitoring model using clustering results, the monitoring results obtained from real-time monitoring of the equipment include curves of y-axis vibration acceleration, y-axis vibration displacement, and y-axis vibration velocity. Detailed Implementation

[0030] The present invention will be further described in detail below with reference to embodiments, but the scope of protection of the present invention is not limited thereto.

[0031] This invention relates to an automatic threshold definition method for operating conditions of moving equipment, enabling effective monitoring in scenarios where machines operate continuously and their operating conditions change. It should be clarified that multiple operating conditions generally refer to machines operating at different speeds or under different loads. Machine load refers to the various external or internal forces, torques, power, currents, etc., that the machine experiences during operation, and the impact of these factors on the machine's operating state. These factors all lead to changes in equipment vibration. Excessive sensitivity to vibration acceleration may result in too many operating condition classifications, while insensitivity to vibration displacement may lead to too few. Therefore, vibration velocity is a suitable measure for classifying operating conditions. With operating condition classification, other measures can be further classified and analyzed, and SPC scores can be calculated. This greatly helps to obtain the stability of health score calculations for moving equipment under different operating conditions and the robustness of fault diagnosis.

[0032] The method is used to construct a monitoring model;

[0033] Collect training data and preprocess it to obtain multi-dimensional measurement data sets;

[0034] The monitoring model is trained based on different working conditions using a multi-dimensional measurement data set as a sample.

[0035] Acquire multi-dimensional measurement data sets of the moving equipment, input them into the trained monitoring model, and obtain the definition results of the working condition threshold.

[0036] First, clarify the acquisition of multi-dimensional measurement data sets;

[0037] The training data and operational data are the characterization signal data and corresponding monitoring feature data of the equipment during operation.

[0038] Furthermore, the characterization signal data includes vibration acceleration signals, sound signals, and magnetic flux signals, and the monitoring feature data includes loudness values ​​and temperature values ​​at corresponding times; the measurement values ​​of the characterization signal data are obtained, and a multi-dimensional measurement data set is established with the monitoring feature data.

[0039] Specifically, it involves collecting vibration acceleration signals, sound signals, and magnetic flux signals during the operation of machinery and equipment. Among these, the vibration acceleration signal (mm / s²) is... 2 The sampling frequency for the sound signal (dB) was 3200Hz, and the sampling duration was 10 seconds. The sampling frequency for the magnetic flux signal (Wb) was 96000Hz, and the sampling duration was 1 second. The sampling frequency for the magnetic flux signal (Wb) was 250Hz, and the sampling duration was 10 seconds. Additionally, the loudness value (dB) and temperature value (Wb) at the corresponding sampling times were also acquired. oC) Process the acquired signal data, performing operations including but not limited to noise reduction and filtering, to improve data quality;

[0040] The values ​​of each measure are calculated using the preprocessed data, including:

[0041] (1) Vibration signal correlation:

[0042] The values ​​of acceleration, velocity and displacement, root mean square, peak-to-average power ratio, skewness, kurtosis and envelope along the x, y and z axes, as well as the centroid of the y-axis spectrum, are obtained. The y-axis signal is the vibration signal perpendicular to the sensor's measurement plane.

[0043] (2) Sound signal related:

[0044] Effective sound values ​​in the 0-10kHz band, 10-20kHz band, 20-30kHz band, 30-40kHz band, and loudness value;

[0045] (3) Magnetic flux related:

[0046] The reciprocal of the root mean square triaxial maximum, the peak-to-mean ratio triaxial maximum, the skewness triaxial maximum, and the kurtosis triaxial maximum;

[0047] (4) Temperature value;

[0048] The measured values ​​obtained from each calculation are saved as a multi-dimensional measured data set.

[0049] Considering that different monitoring thresholds need to be matched for different working conditions, the monitoring model is designed to include a working condition identification module, a working condition statistics module and a threshold adjustment module set in sequence.

[0050] The working condition identification module clusters data from one or more data points in a multi-dimensional measurement data set to determine the working condition category.

[0051] The working condition statistics module calculates the statistics of each measurement value under each working condition;

[0052] Based on the statistics, the threshold adjustment module adjusts the early warning monitoring threshold and the alarm monitoring threshold.

[0053] Specifically, the processing of the monitoring model includes the following steps:

[0054] (2-1) Operating Condition Analysis (Classification)

[0055] Considering the following advantages of using y-axis vibration velocity as a classification criterion:

[0056] 1) Clustering algorithms save more space and time compared to multi-dimensional data that includes all measure values;

[0057] 2) Changes in operating conditions are most evident in vibration signals, and compared to when the sensor is parallel to the monitored plane, the sensor usually has the greatest sensitivity and the least interference when it is perpendicular to the monitored plane.

[0058] 3) Vibration acceleration is too sensitive to changes in working conditions, and the transition phase between working conditions will also be captured, resulting in unclear boundaries between working conditions and affecting the classification effect; while vibration displacement is not sensitive enough to changes in working conditions, resulting in indistinct differences between working conditions. Vibration velocity can not only reduce the problem of unclear boundaries between working conditions, but also has more obvious characteristics of changing working conditions compared to vibration displacement.

[0059] Therefore, the y-axis vibration velocity values ​​of each multi-dimensional measurement data group are extracted and K-means clustering analysis is performed. Each cluster contains at least 3 data points. The optimal number of clusters is determined using the Silhouette method. The data is divided into multiple clusters, which means that it is divided into multiple working conditions.

[0060] like Figure 3 The figures (a), (b), and (c) show the y-axis vibration acceleration, y-axis vibration displacement, and y-axis vibration velocity curves (sky blue curves in the figure) of the induced draft fan from May 15 to May 19, 2025, under the periodic alternation of two working conditions. The figure clearly shows that the variation characteristics of the vibration velocity under the working conditions are the most obvious, and the data is the most stable within one working condition.

[0061] like Figure 4 The image shows the clustering results obtained after performing cluster analysis using historical data of the y-axis vibration velocity of the induced draft fan.

[0062] (2-2) Calculate the statistics of each measure under each working condition, including the mean and standard deviation;

[0063] (2-3) Determine the monitoring threshold based on the mean and standard deviation. The early warning monitoring threshold is calculated as "mean + 3 * standard deviation". The alarm monitoring threshold is set to "mean value + 6 * standard deviation". For subsequent data monitoring; among which, This represents the mean. , for Measurements under each working condition within each working condition. Standard deviation ;

[0064] Through the above clustering and statistical calculations, a basic reference standard for variable operating condition monitoring can be obtained. Subsequent data can then be monitored using SPC (Statistical Process Control) following the steps outlined below.

[0065] Based on the monitoring model, a pre-defined polynomial model is used to fuse the scores of all measures to obtain the final SPC score, which serves as the SPC monitoring result. Specifically, the following steps are included:

[0066] (3-1) Calculate the y-axis vibration velocity value of the current data, and take the cluster center closest to this value as the category, i.e. the working condition;

[0067] (3-2) Calculate the control line based on the average and standard deviation of each measure in the category. The control line includes the warning line and the alarm line. The value of the warning line is: average value + 3 * standard deviation, and the value of the alarm line is: average value + 6 * standard deviation. If the measure value is smaller than its warning value, the measure is normal. If it is greater than the warning value but less than the alarm value, the measure is in a warning state. If it is greater than the alarm value, the measure is in an alarm state.

[0068] (3-3) The corresponding measure score (range 0-100) is calculated according to a unified standard, which is:

[0069] A score of 100 is given when the measure value is less than the minimum threshold (mean + 2 * standard deviation).

[0070] A score of 0 is given when the score is greater than or equal to the maximum threshold (mean + 6 * standard deviation).

[0071] When the score is greater than or equal to the minimum threshold but less than the maximum threshold, a linear interpolation method is used to determine the score.

[0072] (3-4) According to the preset polynomial model, the scores of all measures are merged to obtain the final SPC score (range 0-100), which is used as the SPC monitoring result.

[0073] This invention can adaptively change the warning and alarm values ​​to ensure that the SPC score remains stable under varying operating conditions, thus achieving stability in the monitoring of varying operating conditions.

[0074] In practical applications, the number of historical data required to build the monitoring model can be set according to the actual situation. However, the number of historical data must be at least greater than or equal to the minimum set number to ensure the effectiveness of the monitoring model. Since the real-time monitoring process is all calculated and processed at the edge, considering the running efficiency and memory usage, a maximum number of historical data to build the monitoring model is also set to ensure the computing performance at the edge without affecting the monitoring effect. The maximum set number should be able to cover the working conditions that need to be monitored next. For example, if data is collected every two hours, and the minimum number of historical data points is set to 10 and the maximum to 100, then the number of data collections should be set to a number greater than or equal to 10. When the number of collections reaches the set number, a clustering algorithm is performed on the y-axis vibration velocity, and warning and alarm values ​​are calculated to monitor the machine equipment online. If the set number of collections is greater than 100, then only the latest 100 data points are used to build the monitoring model using the clustering algorithm. For example, if the number of collections is set to 200, then when more than 100 data points are collected, the oldest data point is deleted each time, and the latest data point is saved until the 200th data point. Then, the latest 100 data points are used to build the monitoring model and execute the real-time monitoring process.

[0075] In rare cases, if the machine changes to a new operating condition and the current monitoring model did not collect data for the new operating condition during its construction, the number of times the new operating condition is set can be increased to add the data for the new operating condition to the latest 100 data points, and the monitoring model can be rebuilt to monitor the new operating condition in real time.

[0076] This patented method is applicable not only to equipment with periodically changing operating conditions but also to equipment with continuously changing operating conditions. The difference between periodically and continuously changing operating conditions lies in the fact that the operating conditions in periodically changing conditions change instantaneously, such as the rotational speed changing directly from 700 r / min to 900 r / min, while the operating conditions in continuously changing conditions change more smoothly, with the rotational speed gradually increasing from 700 r / min to 900 r / min. For equipment with continuously changing operating conditions, increasing the number of clusters can divide the gradual change in operating conditions into multiple sub-conditions, achieving a better monitoring effect.

[0077] Specifically, if the current moving equipment is a continuously variable operating condition equipment, then the theoretical slope set of the characteristic measurement data of the current moving equipment under the variable operating condition is picked up; the sampled data is preprocessed, the characteristic slope of the fitted data within a preset range is fitted, the slope interpolation is performed based on the difference between the characteristic slope and the theoretical slope set, the operating conditions are divided based on the interpolation points as sub-operating conditions, the data in each sub-operating condition is clustered, the current operating condition category is confirmed, and monitoring is realized.

[0078] The present invention also relates to the application of the aforementioned method for automatically defining operating condition thresholds for moving equipment, which is applied to the control console of all the aforementioned equipment. When the equipment runs continuously within a preset time range and the operating condition changes with the operation, the control console collects multi-dimensional measurement data sets of the equipment to obtain the automatic definition result of the operating condition threshold and obtain the SPC score.

[0079] like Figure 5 As shown, the monitoring results are obtained by using the clustering results to build a monitoring model and monitor the equipment in real time. (a), (b), and (c) are the y-axis vibration acceleration, y-axis vibration displacement, and y-axis vibration velocity, respectively. The blue solid line in the figure is the actual measurement value, and the yellow dashed line is the warning line. Finally, the monitoring results are used to calculate the SPC score, and the score is stable at 100 points, which means that the current equipment is working normally and all measurements have not exceeded the warning line.

[0080] In this invention, a computer-readable storage medium is used to store an automatic threshold determination program for operating equipment. When the program is executed by a processor, it implements the above-mentioned automatic threshold determination method for operating equipment.

[0081] The present invention also employs a computer device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the program, it implements the above-mentioned method for automatically defining operating condition thresholds applicable to moving equipment.

[0082] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0083] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0084] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0085] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0086] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0087] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for automatically defining operating condition thresholds for moving equipment, characterized in that: The method is used to construct a monitoring model; Collect training data and preprocess it to obtain multi-dimensional measurement data sets; The monitoring model is trained based on different working conditions using a multi-dimensional measurement data set as a sample. Acquire multi-dimensional measurement data sets of the moving equipment, input them into the trained monitoring model, and obtain the definition results of the working condition threshold.

2. The method for automatically defining operating condition thresholds for moving equipment according to claim 1, characterized in that: The training data and operational data are the characterization signal data and corresponding monitoring feature data of the equipment during operation.

3. The method for automatically defining operating condition thresholds for moving equipment according to claim 2, characterized in that: The characterization signal data includes vibration acceleration signal, sound signal, and magnetic flux signal; the monitoring feature data includes loudness value and temperature value at the corresponding time. The measurement value of the characterization signal data is obtained, and a multi-dimensional measurement data group is established with the monitoring feature data.

4. The method for automatically defining operating condition thresholds for moving equipment according to claim 1, characterized in that: The monitoring model includes a working condition identification module, a working condition statistics module, and a threshold adjustment module set in sequence.

5. The method for automatically defining operating condition thresholds for moving equipment according to claim 4, characterized in that: The operating condition identification module clusters data from one or more data points in a multi-dimensional measurement data set to determine the operating condition category.

6. The method for automatically defining operating condition thresholds for moving equipment according to claim 5, characterized in that: The working condition statistics module calculates the statistics of each measurement value under each working condition.

7. The method for automatically defining operating condition thresholds for moving equipment according to claim 6, characterized in that: Based on the statistics, the threshold adjustment module adjusts the early warning monitoring threshold and the alarm monitoring threshold.

8. The method for automatically defining operating condition thresholds for moving equipment according to claim 4, characterized in that: Based on the monitoring model, a pre-defined polynomial model is used to fuse the scores of all measures to obtain the final SPC score, which serves as the SPC monitoring result.

9. The application of the automatic threshold determination method for operating conditions of moving equipment as described in any one of claims 1 to 8, characterized in that: The console, applied to all the aforementioned devices, collects multi-dimensional measurement data sets of the devices when the devices operate continuously within a preset time range and the operating conditions change with operation, obtains the automatic definition result of the operating condition threshold, and obtains the SPC score.

Images