Heat dissipation monitoring method, device and system for gear reducer

Abstract

The invention relates to the technical field of temperature monitoring, in particular to a heat dissipation monitoring method, device and system of a gear reducer. And the data similar interval and the cut-off point of each data point are dynamically determined from the temperature time sequence data, so that the heat dissipation monitoring accuracy is improved. Thirdly, comprehensively analyzing the temperature difference and the interval overlapping condition in the data similar interval of each data point, and calculating an initial abnormal degree value; and then, correcting the initial abnormal degree value by using the difference between the local temperature change condition at the cut-off point and the local temperature change condition at the data point and the cut-off condition of the data similar interval to obtain an updated abnormal degree value. And screening out abnormal data points according to an updated abnormal degree value and a temperature local discrete condition, and carrying out model training by using the data to obtain an anomaly recognition model with high sensitivity and accuracy. And finally, monitoring the temperature condition of the gear reducer in real time by using the model, and realizing accurate monitoring of the heat dissipation condition. The method improves the accuracy of the monitoring result.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature monitoring, and in particular to a heat dissipation monitoring method, device and system for a gear reducer. Background Art

[0002] As an important component of mechanical equipment, the stability of the gear reducer's operating state has a vital impact on the performance of the entire mechanical system. During the operation of the gear reducer, its temperature will change due to the influence of various factors such as friction, load changes, and working environment. If the temperature rises abnormally, it may cause the gear reducer's performance to deteriorate, damage, or even cause a safety accident. Therefore, real-time monitoring of the gear reducer's temperature and timely detection of temperature abnormalities are of great significance to ensure the normal operation of the gear reducer.

[0003] In recent years, with the rapid development of sensor technology and data processing technology, intelligent monitoring methods based on the Internet of Things have gradually become a research hotspot. These methods can not only monitor the temperature changes of reducers in real time, but also achieve fault prediction and early warning through data analysis and cloud computing, providing a scientific basis for equipment maintenance and management.

[0004] When the prior art uses sensor technology to monitor the heat dissipation of the gear reducer, a fixed temperature threshold is usually set. When the temperature data collected by the temperature sensor exceeds the temperature threshold, it is determined that the heat dissipation of the gear reducer is abnormal. However, due to the complex operating conditions of the gear reducer, the normal temperature range under different conditions will also be different. Therefore, the change of the gear reducer temperature under different conditions has a certain pattern and trend. Therefore, using a fixed temperature threshold is difficult to adapt to the temperature changes under different conditions, which will affect the accuracy of the gear reducer heat dissipation monitoring results. Summary of the invention

[0005] In order to solve the complex operating conditions of the gear reducer, the normal temperature range under different working conditions will also be different. Therefore, the change of the temperature of the gear reducer under different working conditions has a certain pattern and trend; therefore, it is difficult to adapt to the temperature change under different working conditions using a fixed temperature threshold, which will affect the accuracy of the gear reducer heat dissipation monitoring result. The purpose of the present invention is to provide a gear reducer heat dissipation monitoring method, device and system, and the technical scheme adopted is as follows:

[0006] Get the temperature time series data of the gear reducer;

[0007] In the temperature time series data, the cutoff point corresponding to each data point is determined according to the fluctuation and change trend of the temperature value in the local range of each data point, so as to obtain the data similarity interval corresponding to each data point;

[0008] In the data similarity interval corresponding to each data point, the difference in temperature values ​​and the overlap of the data similarity intervals of all data points are analyzed to obtain the initial abnormality degree value of each data point; based on the difference in the change of temperature values ​​in the data similarity interval of each data point and the data similarity interval of the corresponding cutoff point, and combined with the cutoff of the data similarity interval of the data point, the initial abnormality degree value of each data point is corrected to obtain the updated abnormality degree value of each data point;

[0009] Based on the updated abnormality degree values ​​of all data points and the discreteness of the temperature values ​​in the data similarity intervals of all data points, abnormal data points are screened out from all data points; the model is trained according to the temperature changes of the abnormal data points, so as to monitor the temperature condition of the gear reducer at the current moment based on the trained abnormality recognition model.

[0010] Furthermore, the method for obtaining the cutoff point includes:

[0011] In the temperature time series data, any data point is selected as a point to be measured;

[0012] In terms of time series, the data is traversed forward in sequence starting from the point to be measured. In each traversal process, the fluctuation and change trend of the temperature values ​​of all data points between the data point to be traversed and the point to be measured are analyzed to obtain the merging index corresponding to the data point to be traversed. If the merging index corresponding to the data point to be traversed meets the interval judgment condition, the data point to be traversed is merged into the data similarity interval of the point to be measured, and the traversal is continued; if the interval judgment condition is not met, the traversal is stopped, and the data point to be traversed at this time is used as the cutoff point corresponding to the point to be measured;

[0013] The interval judgment condition is: the merging index is greater than a preset merging threshold.

[0014] Furthermore, the method for obtaining the combined index includes:

[0015] Get the timing curve of temperature timing data;

[0016] In each traversal process, the data point between the data point to be traversed and the previous data point of the data point to be measured in time sequence is taken as the target point, the absolute value of the difference in temperature between each target point and the point to be measured is taken as the temperature deviation factor, and the average of all temperature deviation factors is taken as the temperature deviation value corresponding to the data point to be traversed;

[0017] Obtaining the temperature slope value at the data point to be traversed on the timing curve;

[0018] The product of the temperature slope value at the data point to be traversed and the temperature deviation value corresponding to the data point to be traversed is negatively correlated and normalized to a value which is used as the merging index corresponding to the data point to be traversed.

[0019] Furthermore, the method for obtaining the data similarity interval includes:

[0020] In terms of time series, the data points between the first data point after the cutoff point time series corresponding to the test point and the test point form a data similarity interval corresponding to the test point.

[0021] Furthermore, the method for obtaining the initial abnormality degree value includes:

[0022] In the data similarity interval corresponding to the point to be tested, the number of data points that are repeated at any moment in the data similarity interval of each data point and the data similarity interval of the point to be tested is taken as the overlapping length corresponding to each data point, and the difference between the length of the data similarity interval of the point to be tested and the overlapping length corresponding to each data point is taken as the length difference factor, and the mean of all the length difference factors is taken as the initial abnormal factor of the point to be tested;

[0023] The difference between the maximum temperature value and the minimum temperature value in the data similarity interval corresponding to the test point is taken as the temperature range;

[0024] The value obtained by normalizing the product of the initial abnormal factor and the temperature extreme difference is used as the initial abnormal degree value corresponding to the point to be measured.

[0025] Furthermore, the method for obtaining the updated abnormality degree value includes: for any data point, taking the number of times the cutoff point corresponding to the data point is used as the cutoff point as the first adjustment factor;

[0026] Calculate the mean of all temperature values ​​in the data similarity interval of the data point as the first temperature characteristic value;

[0027] The mean of all temperature values ​​in the data similarity interval of the cutoff point corresponding to the data point is taken as the second temperature characteristic value;

[0028] A value obtained by negatively correlating and normalizing the absolute value of the difference between the first temperature characteristic value and the second temperature characteristic value is used as a second adjustment factor;

[0029] Normalizing the product of the first adjustment factor and the second adjustment factor as the adjustment coefficient corresponding to the initial abnormality value of the data point;

[0030] The product of the adjustment coefficient and the initial abnormality level value of the data point is normalized to a value which is used as the updated abnormality level value of the data point.

[0031] Furthermore, the method for obtaining abnormal data points includes:

[0032] In the data similarity interval corresponding to each data point, the variance of all temperature values ​​is taken as the discrete factor corresponding to each data point;

[0033] Constructing a coordinate system based on the discrete factors of the data points and the updated abnormality degree values, and mapping all data points to the coordinate system;

[0034] In the coordinate system, a preset radius and a preset minimum number are set, and cluster analysis is performed on all data points based on the DBSCAN algorithm to obtain all cluster clusters;

[0035] In each cluster, the Euclidean norm of the mean of the discrete factors of all data points and the mean of the updated abnormality values ​​of all data points is calculated; the product of the Euclidean norm and the distribution density of the data points is normalized to obtain the abnormality coefficient corresponding to each cluster;

[0036] Among all clusters, the data points in the cluster with the largest abnormal coefficient are regarded as abnormal data points.

[0037] Furthermore, the model is trained according to the temperature change of the abnormal data point, so as to monitor the temperature of the gear reducer at the current moment based on the trained abnormal recognition model, including:

[0038] In the temperature time series data, determining a preset neighborhood of each abnormal data point;

[0039] In a preset neighborhood of each abnormal data point, the variance of the temperature slope values ​​at all data points is calculated as the variation characteristic value of each abnormal data point;

[0040] Input the change characteristic values ​​of all abnormal data points into the SVM algorithm and train it to obtain a trained abnormality recognition model;

[0041] In the preset neighborhood corresponding to the data point at the current moment, the change feature value of the data point at the current moment is calculated and the change feature value is input into the trained anomaly recognition model to output the monitoring result.

[0042] A heat dissipation monitoring device for a gear reducer comprises a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and when the at least one instruction, at least one program, a code set or an instruction set is loaded and executed by the processor, any step of a heat dissipation monitoring method for a gear reducer is implemented.

[0043] A heat dissipation monitoring system for a gear reducer, the system comprising:

[0044] A data acquisition module, used to acquire the temperature time series data of the gear reducer;

[0045] An interval division module is used to determine the cutoff point corresponding to each data point in the temperature time series data according to the fluctuation and change trend of the temperature value in the local range of each data point, so as to obtain the data similarity interval corresponding to each data point;

[0046] The abnormality degree analysis module is used to analyze the difference of temperature values ​​and the overlap of data similarity intervals of all data points in the data similarity interval corresponding to each data point, and obtain the initial abnormality degree value of each data point; based on the difference of temperature value changes in the data similarity interval of each data point and the data similarity interval of the corresponding cutoff point, and combined with the cutoff of the data similarity interval of the data point, the initial abnormality degree value of each data point is corrected to obtain the updated abnormality degree value of each data point;

[0047] The heat dissipation monitoring module is used to screen abnormal data points from all data points based on the updated abnormal degree values ​​of all data points and the discreteness of the temperature values ​​in the data similarity intervals of all data points; the model is trained according to the temperature changes of the abnormal data points, so as to monitor the temperature condition of the gear reducer at the current moment based on the trained abnormal recognition model.

[0048] The present invention has the following beneficial effects:

[0049] First, the temperature time series data of the gear reducer is obtained. Since the temperature change of the gear reducer under different working conditions may be different, and the temperature change of the gear reducer during operation is relatively linear, the present invention dynamically determines the data similarity interval and cutoff point of each data point through the local temperature fluctuation and change trend of each data point in the temperature time series data, thereby preliminarily improving the accuracy of subsequent heat dissipation monitoring; then, in the data similarity interval corresponding to each data point, the difference of the temperature value and the overlap of the data similarity interval are comprehensively analyzed to more comprehensively capture the trend and pattern of temperature change, thereby calculating the initial abnormality value of each data point. Further, since the local fluctuation of the temperature value at the cutoff point corresponding to each data point can reflect the abnormality of the cutoff point, thereby reflecting the abnormality of each data point, the present invention uses the difference between the local temperature change at the cutoff point corresponding to each data point and the local temperature change at each data point, as well as the cutoff of the data similarity interval of all data points, to correct the initial abnormality value of each data point, thereby obtaining the updated abnormality value of each data point. Next, based on the updated abnormality value of the data point and the local discreteness of the temperature value, the abnormal data points are screened out. At this time, the abnormal data points are screened out based on the local change pattern of the temperature value, so they will be more accurate. Therefore, the temperature change of the abnormal data points is used for model training, and the obtained abnormal recognition model will also have higher sensitivity and accuracy. Finally, the trained abnormal recognition model is used to monitor the temperature of the gear reducer at the current moment, and a more accurate monitoring result can be obtained, thereby realizing real-time monitoring of the heat dissipation of the gear reducer. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 A method flow chart of a heat dissipation monitoring method for a gear reducer provided by one embodiment of the present invention;

[0052] Figure 2 A schematic structural diagram of a heat dissipation monitoring device for a gear reducer provided by one embodiment of the present invention;

[0053] Figure 3 A system block diagram of a heat dissipation monitoring system for a gear reducer provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the heat dissipation monitoring method, device and system of a gear reducer proposed by the present invention, its specific implementation method, structure, characteristics and effects in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0055] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0056] The specific scheme of the heat dissipation monitoring method, device and system of a gear reducer provided by the present invention is described in detail below with reference to the accompanying drawings.

[0057] See also Figure 1 , which shows a method flow chart of a heat dissipation monitoring method for a gear reducer provided by an embodiment of the present invention, the method comprising the following steps:

[0058] Step S1: Obtaining the temperature time series data of the gear reducer.

[0059] In modern industrial production, gear reducers are widely used in mechanical transmission systems. The reliability of their performance is directly related to the operating stability and production efficiency of the entire equipment. With the increase of working load, gear reducers are prone to generate excessively high temperatures during long-term operation, leading to lubricant degradation, increased gear wear, and even equipment failure. Therefore, it is particularly important to monitor the heat dissipation of gear reducers in real time.

[0060] In view of the fact that the operating conditions of the gear reducer are relatively complex and the normal temperature range under different conditions may also be different, in the embodiment of the present invention, the heat dissipation of the gear reducer is not monitored relying on a fixed temperature threshold, but by analyzing the temperature change pattern and trend of the gear reducer to identify abnormal data points, and then constructing an abnormality recognition model to monitor the temperature condition of the gear reducer at the current moment.

[0061] First, it is necessary to obtain the temperature time series data of the gear reducer. The intelligent sensing system can be used to collect the temperature value of the gear reducer: install the temperature sensor at a suitable position that can accurately reflect the temperature change of the gear reducer, and then collect the data from the temperature sensor.

[0062] It should be noted that the frequency of data collection can be set to 50 Hz, the horizontal axis of the collected temperature time series data is time, and the vertical axis is the temperature value; the suitable position in the embodiment of the present invention can be set to the bearing shell, oil pan, bottom or top of the gear reducer, etc., and the specific position can be adjusted according to the implementation scenario and is not limited here; the collection length of the temperature time series data can be set to 10 minutes before the current time sequence, and the specific length can be adjusted according to the implementation scenario and is not limited here.

[0063] Step S2: In the temperature time series data, the cutoff point corresponding to each data point is determined according to the fluctuation and change trend of the temperature value in the local range of each data point, so as to obtain the data similarity interval corresponding to each data point.

[0064] Because the normal temperature range of the gear reducer will be different under different operating conditions, when monitoring the heat dissipation of the gear reducer, it is possible to determine whether the temperature is abnormal based on the long-term temperature change. And because the temperature change of the gear reducer during operation is relatively linear, when the temperature is abnormal, the most valuable reference data is usually data that is relatively close in time series. Therefore, in an embodiment of the present invention, the fluctuation and change trend of the temperature value in the local range of each data point in the temperature time series data are analyzed, and the characteristics of the temperature change can be identified, thereby determining a turning point of the temperature change in the local range as a cutoff point, thereby dividing the data similarity interval corresponding to each data point, that is, the interval where the temperature value fluctuates within a certain range and the change is relatively similar.

[0065] Preferably, in one embodiment of the present invention, the method for obtaining the cutoff point includes:

[0066] In order to facilitate subsequent explanation and illustration, a data point is selected as a test point in the temperature time series data, and the method for obtaining the cutoff point corresponding to the test point is described to explain the process of the cutoff point acquisition method.

[0067] In terms of time series, the data is traversed forward in sequence starting from the point to be measured. In each traversal process, the fluctuation and change trend of the temperature values ​​of all data points between the data point to be traversed and the point to be measured are analyzed to obtain the merging index corresponding to the data point to be traversed. If the merging index corresponding to the data point to be traversed meets the interval judgment condition, it means that when the data point to be traversed is incorporated into the data similarity interval of the point to be measured, the fluctuation and change trend of all temperature values ​​in the data similarity interval will not change significantly, so the data point to be traversed is incorporated into the data similarity interval of the point to be measured, and the traversal is continued to find the cutoff point; if the interval judgment condition is not met, it means that the temperature value of the data point to be traversed does not conform to the temperature value change pattern in the data similarity interval of the point to be measured, then the traversal is stopped, and the data point to be traversed at this time is used as the cutoff point corresponding to the point to be measured. Among them, the interval judgment condition is: the merging index is greater than the preset merging threshold.

[0068] The method for obtaining the combined index includes: firstly obtaining a time series curve of temperature time series data.

[0069] Then, in each traversal process, the data point between the data point to be traversed and the previous data point of the data point to be tested in time series is taken as the target point, and the absolute value of the difference in temperature values ​​between each target point and the point to be tested is taken as the temperature deviation factor. The larger the temperature deviation factor, the more dissimilar the temperature values ​​between the target point and the point to be tested. The mean of all temperature deviation factors is taken as the temperature deviation value corresponding to the data point to be traversed. The larger the temperature deviation value, the higher the temperature deviation value, indicating that if the data point to be traversed is added to the data similarity interval of the point to be tested at this time, then the similarity of the temperature values ​​in the data similarity interval corresponding to the point to be tested will be very low, so the possibility of the data point to be traversed being the cutoff point at this time will be higher.

[0070] Then, the temperature slope value at the data point to be traversed is obtained on the timing curve. The slope value at the data point to be traversed represents the instantaneous temperature change at the data point to be traversed. Similarly, the larger the value is, the higher the degree of instantaneous temperature change at the data point to be traversed is, and the possibility of the data point to be traversed as a cutoff point should be higher, that is, the possibility of being incorporated into the data similarity interval of the test point will be lower.

[0071] Finally, the product of the temperature slope value at the data point to be traversed and the temperature deviation value corresponding to the data point to be traversed is negatively correlated and normalized to achieve logical relationship correction, and the value after negative correlation mapping and normalization is used as the merging index corresponding to the data point to be traversed. At this time, the larger the merging index, the more the data point to be traversed should be incorporated into the data similarity interval of the test point. Conversely, the smaller the merging index, the more the data point to be traversed should be used as the cutoff point of the data similarity interval of the test point.

[0072] It should be noted that the negative correlation mapping and normalization method in the embodiment of the present invention can adopt the exp(-x) function, wherein exp() represents an exponential function with the natural constant e as the base, and x represents the independent variable; the preset merging threshold is set to 0.61, and the specific value can be adjusted according to the actual scenario and is not limited here.

[0073] Based on the above process, the cutoff point corresponding to each data point in the temperature time series data can be obtained, and the cutoff point represents the end position of the data similarity interval division of each data point, so the data similarity interval of each data point can be determined based on the cutoff point.

[0074] Preferably, the method for obtaining data similarity intervals includes:

[0075] Taking the test point as an example, in terms of time series, the data points between the first data point after the cutoff point time series corresponding to the test point and the test point (including the test point) form the data similarity interval corresponding to the test point. At this time, the temperature values ​​in the data similarity interval of each data point have a relatively consistent change pattern.

[0076] Here, an example is given to illustrate the process of obtaining the above cutoff point and data similarity interval: the length of the temperature time series data is 10, and the data point 9 is selected as the test point. Then, in the first traversal, data point 8 is used as the first data point to be traversed, and data point 8 is also used as the target point. The merging index corresponding to the data point 8 to be traversed is calculated. If the merging index is greater than the preset merging threshold, then data point 8 and the test point 9 together constitute the data similarity interval of the test point 9; then the second traversal process is performed, and the data point to be traversed at this time is data point 7, and the target point is data point 8 and the data point 7 to be traversed. The merging index corresponding to the data point 8 to be traversed is calculated. The merging index corresponding to the traversed data point 7 is calculated. If the merging index is still greater than the preset merging threshold at this time, data points 7, 8 and the point to be tested 9 together constitute the data similarity interval of the point to be tested 9; then the third traversal process is performed. At this time, the data point to be traversed is data point 6, and the target point is data points 7, 8 and the data point to be traversed 6. The merging index corresponding to the data point to be traversed 6 is calculated. If the merging index is less than or equal to the preset merging threshold at this time, the data point to be traversed 6 is used as the cutoff point of the point to be tested 9. At the same time, the data similarity interval of the point to be tested 9 is obtained, which is composed of data points 7, 8 and the point to be tested 9.

[0077] Step S3: within the data similarity interval corresponding to each data point, analyze the difference in temperature values ​​and the overlap of the data similarity intervals of all data points to obtain the initial abnormality degree value of each data point; based on the difference in the change of temperature values ​​in the data similarity interval of each data point and the data similarity interval of the corresponding cutoff point, and combined with the cutoff of the data similarity interval of the data point, correct the initial abnormality degree value of each data point to obtain the updated abnormality degree value of each data point.

[0078] In step S2, the data similarity interval of each data point can be obtained, and the temperature values ​​in the data similarity interval all have relatively similar change patterns. Therefore, due to the connection between the states of temperature values ​​that are close in time, the overlap of the similarity intervals of the data point can be analyzed in the data similarity interval of a certain data point as an indicator for quantifying the initial abnormality value of the data point; at the same time, the difference in temperature values ​​in the data similarity interval can also be used as an indicator for evaluating whether the data point is abnormal or not. Further, since the division of the data similarity interval depends on the fluctuation and change trend of the temperature value in the local range, the temperature change of the data similarity interval of the cutoff point and the cutoff of the data similarity interval of all data points can also be used as an indicator for evaluating the abnormality of the data point. Therefore, the initial abnormality value of the data point is corrected using the above-mentioned characteristics, thereby obtaining the updated abnormality value of each data point, and the updated abnormality value can more accurately describe the abnormality of each data point.

[0079] Preferably, in one embodiment of the present invention, the method for obtaining the initial abnormality degree value includes:

[0080] Any data point is selected as the test point. In the data similarity interval corresponding to the test point, the number of data points that are repeated at any time in the data similarity interval of each data point and the data similarity interval of the test point is taken as the overlap length corresponding to each data point. The longer the overlap length, the higher the similarity between the test point and the data similarity interval of each data point in the interval, which may indicate that the test point is in a normal temperature fluctuation range. Here, an example is given to illustrate the method of obtaining the overlap length: for example, the test point is data point 9, and the data points in its data similarity interval are data points 5, 6, 7, 8 and 9 respectively. For data point 8, if the data points in its data similarity interval are data points 4, 5, 6, 7 and 8 respectively, then the data points that are repeated at any time in the data similarity interval of data point 8 and the data similarity interval of the test point 9 are data points 5, 6, 7 and 8, and the corresponding overlap length is 4.

[0081] Then, the difference between the length of the data similarity interval of the test point and the corresponding overlapping length of each data point is taken as the length difference factor, and the mean of all length difference factors is taken as the initial anomaly factor of the test point; the data points in the data similarity interval of the test point can be regarded as the similar data set of the test point. When the corresponding length difference factor between the test point and the data points in the similar data set is larger, it means that the degree of deviation between the test point and its similar data is larger, then the larger the initial anomaly factor is, the greater the difference between the test point and its similar data set is, and the higher the degree of anomaly of the test point is.

[0082] Then, the difference between the maximum temperature value and the minimum temperature value in the data similarity interval corresponding to the test point is taken as the temperature extreme difference. The larger the temperature extreme difference, the more drastic the temperature fluctuation in the data similarity interval corresponding to the test point is, which also suggests that the test point may be in an abnormal temperature change situation.

[0083] Finally, the product of the initial abnormal factor and the temperature extreme difference is normalized to obtain the value as the initial abnormality degree value corresponding to the test point. The larger the initial abnormality degree value, the higher the abnormality degree of the test point. Normalization is a technical means well known to those skilled in the art, and the normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0084] When calculating the initial abnormality degree value, it mainly depends on the difference in temperature values ​​in the data similarity interval and the overlap of the data similarity intervals of the data points; further, the change in temperature value between each data point and the corresponding cutoff point can also reflect the abnormality of the data point. Therefore, in an embodiment of the present invention, based on the difference in the change of temperature values ​​in the data similarity interval of each data point and the data similarity interval of the corresponding cutoff point, and combined with the cutoff conditions of the data similarity intervals of all data points, the initial abnormality degree value of each data point is corrected to obtain an updated abnormality degree value for each data point.

[0085] Preferably, in one embodiment of the present invention, the method for obtaining the updated abnormality degree value includes:

[0086] For any data point, the number of times the cutoff point corresponding to the data point is used as the cutoff point is counted, and the number is used as the first adjustment factor; when a data point is used as the cutoff point multiple times, it is considered that the degree of abnormality of this cutoff point is higher. Therefore, when the first adjustment factor corresponding to the cutoff point of a data point is larger, it is considered that the degree of abnormality of the cutoff point corresponding to the data point is larger.

[0087] Then, the mean of all temperature values ​​in the data similarity interval of the data point is calculated as the first temperature characteristic value. By calculating the mean, some random fluctuations can be smoothed out, so that the first temperature characteristic value more accurately reflects the average temperature state of the gear reducer within the local time range of the data point.

[0088] Similarly, the mean of all temperature values ​​in the data similarity interval of the cutoff point corresponding to the data point is taken as the second temperature characteristic value, and the second temperature characteristic value reflects the average temperature state of the gear reducer within the local time range of the cutoff point corresponding to the data point.

[0089] When the deviation between the first temperature characteristic value and the second temperature characteristic value is large, it means that the similarity between the average temperature state at the data point and the corresponding cutoff point in the local range is lower; conversely, when the deviation between the first temperature characteristic value and the second temperature characteristic value is small, it is considered that the similarity between the average temperature state at the data point and the corresponding cutoff point in the local range is high; the absolute value of the difference between the first temperature characteristic value and the second temperature characteristic value is negatively correlated and normalized, and the value is used as the second adjustment factor. At this time, the larger the second adjustment factor is, the more similar the average temperature change state between the data point and the corresponding cutoff point is. In view of the cutoff point as a point that is different from the data point, the larger the second adjustment factor is, the greater the abnormality of the data point is. Among them, the negative correlation mapping and normalization method can use the exp(-x) function, where exp() represents an exponential function with the natural constant e as the base, and x represents the independent variable.

[0090] Then, the first adjustment factor and the second adjustment factor are comprehensively analyzed. Since the larger the first adjustment factor, the more abnormal the cutoff point is; and the larger the second adjustment factor, the more consistent the change in the local temperature state between a data point and the corresponding cutoff point is, the product of the first adjustment factor and the second adjustment factor is normalized to the value obtained as the adjustment coefficient corresponding to the initial abnormality value of the data point. At this time, the larger the adjustment coefficient corresponding to the data point, the higher the abnormality of the data point.

[0091] Finally, the product of the adjustment coefficient and the initial abnormality value of the data point is normalized to obtain the value, which is used as the updated abnormality value of the data point. The larger the updated abnormality value, the higher the temperature abnormality at the data point. Normalization is a technical means well known to those skilled in the art, and the normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0092] Step S4: Based on the updated abnormality degree values ​​of all data points and the discreteness of the temperature values ​​in the data similarity intervals of all data points, abnormal data points are screened out from all data points; the model is trained according to the temperature change of the abnormal data points, so as to monitor the temperature condition of the gear reducer at the current moment based on the trained abnormality recognition model.

[0093] Based on the above steps, the updated abnormality degree value of each data point can be obtained as an indicator for evaluating the abnormal situation of a single data point. In an embodiment of the present invention, by further analyzing the updated abnormality degree values ​​of all data points and combining the discrete situation of the temperature values ​​in the data similarity intervals of all data points, outlier abnormal data points can be identified more accurately; then, the abnormality recognition model is trained based on the temperature change of the identified outlier abnormal data points, and finally, the temperature condition of the gear reducer at the current moment is monitored based on the trained abnormality recognition model.

[0094] Preferably, in one embodiment of the present invention, the method for obtaining abnormal data points includes:

[0095] In the data similarity interval corresponding to each data point, the variance of all temperature values ​​is used as the discrete factor corresponding to each data point. The discrete factor is used to measure the degree of dispersion of the temperature values ​​in the data similarity interval of the data point, and the larger the value, the greater the degree of dispersion.

[0096] A coordinate system is constructed based on the discrete factor of the data point and the updated abnormality value. The coordinate system can be constructed with the discrete factor as the horizontal coordinate and the updated abnormality value as the vertical coordinate, and all data points are mapped to the coordinate system. At this time, each data point can be represented by the coordinate of (discrete factor, updated abnormality value).

[0097] In this coordinate system, a preset radius and a preset minimum number are set, and cluster analysis is performed on all data points based on the DBSCAN algorithm to obtain all cluster clusters. Through cluster analysis, data points with the same state or mode can be classified into one category. It should be noted that the preset radius is 3, the preset minimum number is 3, and the specific values ​​can be adjusted according to the implementation scenario, which is not limited here; the DBSCAN algorithm is a well-known technology, and the specific process is not repeated here.

[0098] Then, in each cluster, the Euclidean norm between the mean of the discrete factors of all data points and the mean of the updated abnormality values ​​of all data points is calculated. The larger the Euclidean norm, the higher the temperature discreteness of the data points in the cluster, and the higher the abnormality, so the possibility that the data points in the cluster are the final abnormal data points is higher; since the distribution density of the data points in each cluster can reflect the distribution characteristics between the data points, the distribution density of all data points in each cluster is calculated. In the embodiment of the present invention, the calculation method of the distribution density is: in the coordinate system, the average value of the distance between all data points in each cluster is used as the distribution density, wherein the distance of the data point quality inspection is calculated based on the coordinates of the two data point quality inspections. The larger the distribution density, the more discrete the distribution of the data points in the cluster, which means that it is more likely to be an outlier data point and the higher the abnormality.

[0099] Then, the value after normalizing the product of the Euclidean norm and the distribution density of the data points corresponding to each cluster is used as the abnormality coefficient corresponding to each cluster. The larger the abnormality coefficient, the higher the abnormality of the data points in the cluster. Normalization is a technical means well known to those skilled in the art, and the normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0100] Finally, among all clusters, the data points in the cluster with the largest abnormal coefficient are taken as abnormal data points.

[0101] Based on the above process, all abnormal data points in the temperature time series data can be obtained, and since the temperature time series data is a section of data before the current time series, the abnormality recognition model can be trained based on the temperature change of the abnormal data points, so as to use the abnormality recognition model to monitor the temperature of the gear reducer at the current moment, thereby improving the accuracy of the heat dissipation monitoring results.

[0102] Preferably, in one embodiment of the present invention, the model is trained according to the temperature change of the abnormal data point, so as to monitor the temperature condition of the gear reducer at the current moment based on the trained abnormal recognition model, including:

[0103] First, the preset neighborhood of each abnormal data point is determined.

[0104] Then, in the preset neighborhood of each abnormal data point, the variance of the temperature slope values ​​at all data points is calculated as the variation characteristic value of each abnormal data point. The variation characteristic value quantifies the temperature change fluctuation of the data points around each abnormal data point, which is the key feature for identifying abnormal patterns.

[0105] Then, the changing characteristic values ​​of all abnormal data points are input into the SVM algorithm and trained so that the SVM algorithm can learn the characteristics of the abnormal data points, thereby obtaining a trained anomaly recognition model.

[0106] Finally, in the preset neighborhood corresponding to the data point at the current moment, the change characteristic value of the data point at the current moment is calculated, and the change characteristic value is input into the trained anomaly recognition model to output the monitoring result.

[0107] When the monitoring result is abnormal, an early warning is required to notify the staff to inspect and repair the gear reducer; when the monitoring result is normal, no early warning is required.

[0108] It should be noted that the preset neighborhood is to take each abnormal data point as the starting point and take the previous 5 data points in time series. The specific number can be adjusted according to the implementation scenario and is not limited here. The training process of the SVM algorithm is a well-known technology and will not be described in detail here.

[0109] In summary, firstly, the temperature time series data of the gear reducer is obtained. Since the temperature change of the gear reducer under different working conditions may be different, and the temperature change of the gear reducer during operation is relatively linear, the present invention dynamically determines the data similarity interval and cutoff point of each data point through the local temperature fluctuation and change trend of each data point in the temperature time series data, thereby preliminarily improving the accuracy of subsequent heat dissipation monitoring; then, in the data similarity interval corresponding to each data point, the difference of the temperature value and the overlap of the data similarity interval are comprehensively analyzed to more comprehensively capture the trend and pattern of temperature change, thereby calculating the initial abnormality degree value of each data point. Further, since the local fluctuation of the temperature value at the cutoff point corresponding to each data point can reflect the abnormality of the cutoff point, thereby reflecting the abnormality degree of each data point, the present invention uses the difference between the local temperature change at the cutoff point corresponding to each data point and the local temperature change at each data point, as well as the cutoff of the data similarity interval of all data points, to correct the initial abnormality degree value of each data point, thereby obtaining the updated abnormality degree value of each data point. Next, based on the updated abnormality value of the data point and the local discreteness of the temperature value, the abnormal data points are screened out. At this time, the abnormal data points are screened out based on the local change pattern of the temperature value, so they will be more accurate. Therefore, the temperature change of the abnormal data points is used for model training, and the obtained abnormal recognition model will also have higher sensitivity and accuracy. Finally, the trained abnormal recognition model is used to monitor the temperature of the gear reducer at the current moment, and a more accurate monitoring result can be obtained, thereby realizing real-time monitoring of the heat dissipation of the gear reducer.

[0110] An embodiment of the present invention also provides a heat dissipation monitoring device for a gear reducer, including a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and when the at least one instruction, at least one program, a code set or an instruction set is loaded and executed by the processor, the steps of a heat dissipation monitoring method for a gear reducer are implemented.

[0111] See also Figure 2, which shows a schematic structural diagram of a heat dissipation monitoring device for a gear reducer provided by an embodiment of the present invention, including a processor 200, a memory 201, a bus 202 and a communication interface 203, wherein the processor 200, the communication interface 203 and the memory 201 are connected via the bus 202; wherein the memory 201 may include a high-speed random access memory, the bus 202 may be an ISA bus, a PCI bus or an EISA bus, etc., and the processor 200 may be an integrated circuit chip with signal processing capabilities; the memory 201 stores at least one instruction, at least one program, a code set or an instruction set, and when the at least one instruction, at least one program, a code set or an instruction set is loaded and executed by the processor, the steps in a heat dissipation monitoring method for a gear reducer are implemented.

[0112] The embodiment of the present invention also provides a heat dissipation monitoring system for a gear reducer, see Figure 3 , which shows a system block diagram of a heat dissipation monitoring system for a gear reducer, the system comprising: a data acquisition module 301, used to acquire the temperature time series data of the gear reducer; an interval division module 302, used to determine the cutoff point corresponding to each data point in the temperature time series data according to the fluctuation and change trend of the temperature value in the local range of each data point, so as to obtain the data similarity interval corresponding to each data point; an abnormality degree analysis module 303, used to analyze the difference of the temperature value and the overlap of the data similarity intervals of all data points in the data similarity interval corresponding to each data point, so as to obtain the initial abnormality degree value of each data point ; Based on the difference in the change of temperature values ​​in the data similarity interval of each data point and the data similarity interval of the corresponding cutoff point, and combined with the cutoff of the data similarity interval of the data point, the initial abnormality degree value of each data point is corrected to obtain the updated abnormality degree value of each data point; the heat dissipation monitoring module 304 is used to screen abnormal data points from all data points based on the updated abnormality degree values ​​of all data points and the discreteness of the temperature values ​​in the data similarity intervals of all data points; the model is trained according to the temperature change of the abnormal data points, so as to monitor the temperature condition of the gear reducer at the current moment based on the trained abnormality recognition model.

[0113] It should be noted that the system provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the heat dissipation monitoring system for a gear reducer and the heat dissipation monitoring method for a gear reducer provided in the above embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0114] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0115] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A heat dissipation monitoring method for a gear reducer, characterized in that: The method comprises: Get the temperature time series data of the gear reducer; In the temperature time series data, the cutoff point corresponding to each data point is determined according to the fluctuation and change trend of the temperature value in the local range of each data point, so as to obtain the data similarity interval corresponding to each data point; In the data similarity interval corresponding to each data point, the difference in temperature values ​​and the overlap of the data similarity intervals of all data points are analyzed to obtain the initial abnormality degree value of each data point; based on the difference in the change of temperature values ​​in the data similarity interval of each data point and the data similarity interval of the corresponding cutoff point, and combined with the cutoff of the data similarity interval of the data point, the initial abnormality degree value of each data point is corrected to obtain the updated abnormality degree value of each data point; Based on the updated abnormality degree values ​​of all data points and the discreteness of the temperature values ​​in the data similarity intervals of all data points, abnormal data points are screened out from all data points; the model is trained according to the temperature changes of the abnormal data points, so as to monitor the temperature condition of the gear reducer at the current moment based on the trained abnormality recognition model.

2. The heat dissipation monitoring method of a gear reducer according to claim 1, characterized in that: The method for obtaining the cutoff point comprises: In the temperature time series data, any data point is selected as a point to be measured; In terms of time series, the data is traversed forward in sequence starting from the point to be measured. In each traversal process, the fluctuation and change trend of the temperature values ​​of all data points between the data point to be traversed and the point to be measured are analyzed to obtain the merging index corresponding to the data point to be traversed. If the merging index corresponding to the data point to be traversed meets the interval judgment condition, the data point to be traversed is merged into the data similarity interval of the point to be measured, and the traversal is continued; if the interval judgment condition is not met, the traversal is stopped, and the data point to be traversed at this time is used as the cutoff point corresponding to the point to be measured; The interval judgment condition is: the merging index is greater than a preset merging threshold.

3. The heat dissipation monitoring method of a gear reducer according to claim 2, characterized in that: The method for obtaining the combined index includes: Get the timing curve of temperature timing data; In each traversal process, the data point between the data point to be traversed and the previous data point of the data point to be measured in time sequence is taken as the target point, the absolute value of the difference in temperature between each target point and the point to be measured is taken as the temperature deviation factor, and the average of all temperature deviation factors is taken as the temperature deviation value corresponding to the data point to be traversed; Obtaining the temperature slope value at the data point to be traversed on the timing curve; The product of the temperature slope value at the data point to be traversed and the temperature deviation value corresponding to the data point to be traversed is negatively correlated and normalized to a value which is used as the merging index corresponding to the data point to be traversed.

4. The heat dissipation monitoring method of a gear reducer according to claim 2, characterized in that: The method for obtaining the data similarity interval includes: In terms of time series, the data points between the first data point after the cutoff point time series corresponding to the test point and the test point form a data similarity interval corresponding to the test point.

5. The heat dissipation monitoring method of a gear reducer according to claim 2, characterized in that: The method for obtaining the initial abnormality degree value includes: In the data similarity interval corresponding to the point to be tested, the number of data points that are repeated at any moment in the data similarity interval of each data point and the data similarity interval of the point to be tested is taken as the overlapping length corresponding to each data point, and the difference between the length of the data similarity interval of the point to be tested and the overlapping length corresponding to each data point is taken as the length difference factor, and the mean of all the length difference factors is taken as the initial abnormal factor of the point to be tested; The difference between the maximum temperature value and the minimum temperature value in the data similarity interval corresponding to the test point is taken as the temperature range; The value obtained by normalizing the product of the initial abnormal factor and the temperature extreme difference is used as the initial abnormal degree value corresponding to the point to be measured.

6. The heat dissipation monitoring method of a gear reducer according to claim 1, characterized in that: The method for acquiring the update abnormality degree value includes: For any data point, the number of times the cutoff point corresponding to the data point is used as the cutoff point is used as the first adjustment factor; Calculate the mean of all temperature values ​​in the data similarity interval of the data point as the first temperature characteristic value; The mean of all temperature values ​​in the data similarity interval of the cutoff point corresponding to the data point is taken as the second temperature characteristic value; A value obtained by negatively correlating and normalizing the absolute value of the difference between the first temperature characteristic value and the second temperature characteristic value is used as a second adjustment factor; Normalizing the product of the first adjustment factor and the second adjustment factor as the adjustment coefficient corresponding to the initial abnormality value of the data point; The product of the adjustment coefficient and the initial abnormality level value of the data point is normalized to a value which is used as the updated abnormality level value of the data point.

7. The heat dissipation monitoring method of a gear reducer according to claim 1, characterized in that: The method for obtaining abnormal data points includes: In the data similarity interval corresponding to each data point, the variance of all temperature values ​​is taken as the discrete factor corresponding to each data point; Constructing a coordinate system based on the discrete factors of the data points and the updated abnormality degree values, and mapping all data points to the coordinate system; In the coordinate system, a preset radius and a preset minimum number are set, and cluster analysis is performed on all data points based on the DBSCAN algorithm to obtain all cluster clusters; In each cluster, the Euclidean norm of the mean of the discrete factors of all data points and the mean of the updated abnormality values ​​of all data points is calculated; the product of the Euclidean norm and the distribution density of the data points is normalized to obtain the abnormality coefficient corresponding to each cluster; Among all clusters, the data points in the cluster with the largest abnormal coefficient are regarded as abnormal data points.

8. The heat dissipation monitoring method of a gear reducer according to claim 3, characterized in that: The model is trained according to the temperature change of the abnormal data point, so as to monitor the temperature condition of the gear reducer at the current moment based on the trained abnormal recognition model, including: In the temperature time series data, determining a preset neighborhood of each abnormal data point; In a preset neighborhood of each abnormal data point, the variance of the temperature slope values ​​at all data points is calculated as the variation characteristic value of each abnormal data point; Input the change characteristic values ​​of all abnormal data points into the SVM algorithm and train it to obtain a trained abnormality recognition model; In the preset neighborhood corresponding to the data point at the current moment, the change feature value of the data point at the current moment is calculated and the change feature value is input into the trained anomaly recognition model to output the monitoring result.

9. A heat dissipation monitoring device for a gear reducer, characterized in that: It includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and when the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor, the steps of a heat dissipation monitoring method for a gear reducer as described in any one of claims 1 to 8 are implemented.

10. A heat dissipation monitoring system for a gear reducer, characterized in that: The system comprises: A data acquisition module, used to acquire the temperature time series data of the gear reducer; An interval division module is used to determine the cutoff point corresponding to each data point in the temperature time series data according to the fluctuation and change trend of the temperature value in the local range of each data point, so as to obtain the data similarity interval corresponding to each data point; The abnormality degree analysis module is used to analyze the difference of temperature values ​​and the overlap of data similarity intervals of all data points in the data similarity interval corresponding to each data point, and obtain the initial abnormality degree value of each data point; based on the difference of temperature value changes in the data similarity interval of each data point and the data similarity interval of the corresponding cutoff point, and combined with the cutoff of the data similarity interval of the data point, the initial abnormality degree value of each data point is corrected to obtain the updated abnormality degree value of each data point; The heat dissipation monitoring module is used to screen abnormal data points from all data points based on the updated abnormal degree values ​​of all data points and the discreteness of the temperature values ​​in the data similarity intervals of all data points; the model is trained according to the temperature changes of the abnormal data points, so as to monitor the temperature condition of the gear reducer at the current moment based on the trained abnormal recognition model.

Images