An incremental learning data flow analysis platform

Abstract

The present invention relates to the technical field of data processing, and specifically discloses an incremental learning data flow analysis platform, which is used to solve the problem that the prior art does not mention the relationship between the aging of rolling bearings caused by different load fluctuations and the sensitivity of aging to load fluctuations. The present invention obtains the working data flow of the rolling bearing; the working data flow includes the first working data flow obtained when the rolling bearing runs for the first time and the second working data flow of each run after the first run of the rolling bearing. Analyze the bearing standard coefficient according to the first working data flow, analyze the bearing aging coefficient based on the bearing standard coefficient and the second working data flow, and calculate the fluctuation sensitivity index according to the bearing aging coefficient and the second working data flow. The present invention not only extends the service life of the bearing, but also reduces the operation and maintenance costs.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and more specifically, to an incremental learning data stream analysis platform. Background Art

[0002] Rolling bearings are important components of mechanical equipment and are widely used in the industrial field and the transportation field. However, rolling bearings often operate under extreme conditions such as high temperature and heavy load, and are extremely prone to failure. In the field of rolling bearing fault diagnosis, traditional fault diagnosis models are usually trained based on large-scale offline data. By extracting features such as bearing vibration signals and acoustic emission signals, and combining machine learning or deep learning models for classification and prediction. These methods have good performance in the offline static environment, but face significant limitations in actual engineering applications, especially under long-term operating conditions of equipment. Due to the complex and variable environment during the operation of rolling bearings, factors such as aging and load fluctuations will cause the vibration signals and other features to evolve over time, resulting in data distribution drift or even pattern mutation. Existing offline training models usually have difficulty adapting to this change, manifested as a gradual decrease in model accuracy, and may even be unable to correctly diagnose new pattern faults.

[0003] Incremental learning technology can update the model without accessing a large amount of historical data, thus effectively reducing storage requirements and computational costs. At the same time, incremental learning supports adding new features or new fault patterns without forgetting old knowledge, improving the adaptability and generalization ability of the model. Existing literature ([1] Zhao Shangjun. Research on Rolling Bearing Fault Diagnosis Based on Deep Learning [D]. Shijiazhuang Tiedao University, 2023. DOI: 10.27334 / d.cnki.gstdy.2023.000619.) has studied problems such as low diagnostic accuracy of rolling bearings under variable working conditions, traditional batch learning models being unable to better identify newly generated data, and incremental learning models being unable to better diagnose data under different test benches and different working conditions, but did not mention the relationship between the aging of rolling bearings caused by different load fluctuations and the sensitivity of aging to load fluctuations, resulting in difficulty in adapting to the deterioration of bearing performance in high-load or frequently load-changing environments and a decrease in the reliability of diagnostic results.

[0004] To solve the above problems, a technical solution is provided now. Summary of the Invention

[0005] To overcome the above-mentioned defects of the prior art, the present invention provides an incremental learning data stream analysis platform, which is used to solve the problem that the prior art does not mention the connection between the aging of rolling bearings caused by different load fluctuations and the sensitivity of aging to load fluctuations, resulting in difficulty in adapting to the deterioration of bearing performance in an environment with high loads or frequent load changes and the decline in the reliability of diagnostic results, so as to solve the problems raised in the above background art.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An incremental learning data stream analysis platform includes a data stream import module, a first bearing analysis module, a second bearing analysis module, and a data sensitivity monitoring module. The data stream import module is used to obtain the working data stream of the rolling bearing; the working data stream includes the first working data stream obtained when the rolling bearing runs for the first time and the second working data stream for each run after the first run of the rolling bearing. The first bearing analysis module is used to analyze the bearing standard coefficient according to the first working data stream. The second bearing analysis module is used to analyze the bearing aging coefficient according to the bearing standard coefficient and the second working data stream. The second bearing analysis module includes a data update unit; the data update unit is used to update the data in real time according to the first working data stream and the second working data stream to obtain the bearing standard coefficient and the basic statistical data. The basic statistical data are the maximum value, minimum value, and average value corresponding to the first running vibration frequency, first running vibration amplitude, first running sound frequency, and first running sound amplitude respectively. When the second running data stream reaches the s-th moment, the j-th item among the second running vibration frequency, second running vibration amplitude, second running sound frequency, and second running sound amplitude at the s-th moment is updated. Among them, the average value update formula for each data item is:

[0008] ;

[0009] ;

[0010] In the formula: is the average value corresponding to the j-th item among the second running vibration frequency, second running vibration amplitude, second running sound frequency, and second running sound amplitude up to the s-th moment, is the average value corresponding to the j-th item among the second running vibration frequency, second running vibration amplitude, second running sound frequency, and second running sound amplitude up to the (s - 1)-th moment, is the j-th item among the second running vibration frequency, second running vibration amplitude, second running sound frequency, and second running sound amplitude at the s-th moment, is the number of data items of the j-th item up to the (s - 1)-th moment, The data quantity of the j-th item up to time s.

[0011] As a further solution of the present invention, the data stream merging module includes a first data stream summarizing unit and a second data stream summarizing unit;

[0012] The first data stream summarizing unit is used to obtain the first working data stream obtained during the initial operation of the rolling bearing; the first working data stream includes a first running vibration data stream and a first audio data stream; the first running vibration data stream includes the first running vibration frequency and the first running vibration amplitude of the rolling bearing; the first audio data stream includes the first running sound frequency and the first running sound amplitude of the rolling bearing;

[0013] The second data stream summarizing unit is used to obtain the second working data stream for each operation after the initial operation of the rolling bearing; the second working data stream includes a second running vibration data stream, a second audio data stream, and a first load data stream; the second running vibration data stream includes the second running vibration frequency and the second running vibration amplitude of the rolling bearing; the second audio data stream includes the second running sound frequency and the second running sound amplitude of the rolling bearing; the first load data stream includes the load value of the rolling bearing.

[0014] As a further solution of the present invention, the first bearing analysis module includes a primary data stream extraction unit and a bearing standard coefficient calculation unit; the primary data stream extraction unit is connected to the bearing standard coefficient calculation unit;

[0015] The primary data stream extraction unit is used to extract the first running vibration data stream and the first audio data stream;

[0016] The bearing standard coefficient calculation unit is used to substitute the first running vibration data stream and the first audio data stream into the bearing standard coefficient calculation formula to calculate the bearing standard coefficient. The bearing standard coefficient calculation formula is:

[0017] ;

[0018] In the formula: is the bearing standard coefficient, is the number of data items in the first running vibration frequency, the first running vibration amplitude, the first running sound frequency, and the first running sound amplitude, is the i-th item in the first running vibration frequency, the first running vibration amplitude, the first running sound frequency, and the first running sound amplitude at time t, is the initial running duration of the rolling bearing, is the maximum value corresponding to the i-th item in the first running vibration frequency, the first running vibration amplitude, the first running sound frequency, and the first running sound amplitude, is the minimum value corresponding to the i-th item among the first operating vibration frequency, the first operating vibration amplitude, the first operating sound frequency, and the first operating sound amplitude. is the average value corresponding to the i-th item among the first operating vibration frequency, the first operating vibration amplitude, the first operating sound frequency, and the first operating sound amplitude.

[0019] As a further solution of the present invention, the update formulas for the maximum and minimum values of each data item are:

[0020] ;

[0021] ;

[0022] In the formula: is the maximum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the s-th moment, is the maximum value screening function, is the minimum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the s-th moment, is the minimum value screening function, is the maximum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the (s - 1)-th moment, is the minimum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the (s - 1)-th moment, is the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude at the s-th moment, is the number of data items of the j-th item up to the (s - 1)-th moment.

[0023] As a further solution of the present invention, the bearing aging coefficient calculation unit is used to obtain the bearing standard coefficient, substitute the bearing standard coefficient, the second operating vibration data stream, and the second audio data stream into the bearing aging coefficient calculation formula to calculate the bearing aging coefficient. The bearing aging coefficient calculation formula is:

[0024] ;

[0025] In the formula: is the bearing aging coefficient during the h-th operation of the rolling bearing, is the bearing standard coefficient, is the average value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the s-th moment. is the maximum value corresponding to the j-th item among the second operating vibration frequency, second operating vibration amplitude, second operating sound frequency, and second operating sound amplitude up to time s. is the minimum value corresponding to the j-th item among the second operating vibration frequency, second operating vibration amplitude, second operating sound frequency, and second operating sound amplitude up to time s. is the number of data items in the second operating vibration frequency, second operating vibration amplitude, second operating sound frequency, and second operating sound amplitude. is the j-th item among the second operating vibration frequency, second operating vibration amplitude, second operating sound frequency, and second operating sound amplitude at time s. is the operating duration of the rolling bearing for the h-th time.

[0026] As a further aspect of the present invention, the data sensitivity monitoring module includes a three-level data stream extraction unit, a load fluctuation coefficient calculation unit, and a fluctuation sensitivity index monitoring unit; the three-level data stream extraction unit is used to extract the first load data stream; the load fluctuation coefficient calculation unit is used to construct a load fluctuation model based on the first load data stream to obtain the load fluctuation coefficient of the bearing in real time. The formula of the load fluctuation model is:

[0027] ;

[0028] In the formula: is the load fluctuation coefficient during the h-th operation of the rolling bearing. is the load value of the rolling bearing at time s. is the load value of the rolling bearing at time s - 1. is the maximum load value of the rolling bearing up to time s. is the minimum load value of the rolling bearing up to time s. is the number of data items in the second operating vibration frequency, second operating vibration amplitude, second operating sound frequency, and second operating sound amplitude. is the operating duration of the rolling bearing for the h-th time.

[0029] As a further aspect of the present invention, the fluctuation sensitivity index monitoring unit is used to establish a fluctuation sensitivity analysis model based on the load fluctuation coefficient and the bearing aging coefficient to calculate the fluctuation sensitivity index. The formula of the fluctuation sensitivity analysis model is:

[0030] ;

[0031] In the formula: is the fluctuation sensitivity index. is the bearing aging coefficient during the h-th operation of the rolling bearing. is the load fluctuation coefficient during the h-th operation of the rolling bearing. is the bearing aging coefficient during the (h - 1)-th operation of the rolling bearing, is the load fluctuation coefficient during the (h - 1)-th operation of the rolling bearing, is the number of operations of the rolling bearing.

[0032] The technical effects and advantages of an incremental learning data stream analysis platform of the present invention: By obtaining the working data stream of the rolling bearing; the working data stream includes the first working data stream obtained during the initial operation of the rolling bearing and the second working data stream for each operation after the initial operation of the rolling bearing, analyzing the bearing standard coefficient according to the first working data stream, analyzing the bearing aging coefficient based on the bearing standard coefficient and the second working data stream, and based on the incremental learning model, it can analyze the working data stream of the rolling bearing in real time and dynamically update the bearing aging coefficient and the fluctuation sensitivity index; by comparing the initial operation data and the subsequent operation data, it can accurately evaluate the standard state and its aging trend of the bearing; by calculating the aging coefficient and the fluctuation sensitivity index, it can predict the possible fault trend of the bearing, avoid the interruption of the equipment due to bearing failure in critical tasks, and comprehensively improve the monitoring and management efficiency of the rolling bearing, not only extending the service life of the bearing, but also reducing the operation and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a schematic structural diagram of an incremental learning data stream analysis platform provided by the present invention;

[0034] Figure 2 is a schematic working flow diagram of the second bearing analysis module provided by the present invention;

[0035] Figure 3 is an analysis trend diagram of the second running vibration frequency provided by the present invention;

[0036] Figure 4 is an analysis trend diagram of the second running sound frequency provided by the present invention;

[0037] Figure 5 is an analysis trend diagram of the bearing aging coefficient provided by the present invention;

[0038] Figure 6 is a fluctuation analysis trend diagram of the load fluctuation coefficient provided by the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0039] Next, the technical solutions in the present invention will be clearly and completely described in conjunction with the accompanying drawings in the present invention. Obviously, the described technical solutions are only a part of the present invention, rather than all of them. Based on the technical solutions in the present invention, all other technical solutions obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present invention.

[0040] Figure 1 This is a schematic structural diagram of an incremental learning data flow analysis platform provided by the present invention. As Figure 1 shown, an incremental learning data flow analysis platform includes a data flow input module, a first bearing analysis module, a second bearing analysis module, and a data sensitivity monitoring module;

[0041] The data flow input module is used to obtain the working data flow of the rolling bearing; the working data flow includes the first working data flow obtained when the rolling bearing runs for the first time and the second working data flow for each run after the first run of the rolling bearing;

[0042] The first bearing analysis module is used to analyze the bearing standard coefficient according to the first working data flow;

[0043] The second bearing analysis module is used to analyze the bearing aging coefficient according to the bearing standard coefficient and the second working data flow;

[0044] The data sensitivity monitoring module is used to calculate the fluctuation sensitivity index according to the bearing aging coefficient and the second working data flow.

[0045] Specifically, the data flow input module includes a first data flow summary unit and a second data flow summary unit;

[0046] The first data flow summary unit is used to obtain the first working data flow obtained when the rolling bearing runs for the first time; the first working data flow includes the first running vibration data flow and the first audio data flow; the first running vibration data flow includes the first running vibration frequency and the first running vibration amplitude of the rolling bearing; the first audio data flow includes the first running sound frequency and the first running sound amplitude of the rolling bearing;

[0047] The second data flow summary unit is used to obtain the second working data flow for each run after the first run of the rolling bearing; the second working data flow includes the second running vibration data flow, the second audio data flow, and the first load data flow; the second running vibration data flow includes the second running vibration frequency and the second running vibration amplitude of the rolling bearing; the second audio data flow includes the second running sound frequency and the second running sound amplitude of the rolling bearing; the first load data flow includes the load value of the rolling bearing;

[0048] As Figure 3 shown in the analysis trend chart of the provided second running vibration frequency, the value of the second running vibration frequency is approximately between 140 and 160, the overall curve fluctuates slightly but is relatively stable, the values from January to December generally fluctuate around 150, and the difference between the highest point and the lowest point is not large, indicating that the vibration noise is relatively stable throughout the year; As Figure 4As shown in the analysis trend chart of the second operating sound frequency provided, the vertical axis value is approximately between 70 and 90. The curve also has slight fluctuations throughout the year. It is about 80 in January, and then fluctuates up and down between 80 and 85. It slightly drops or stabilizes at the end of the year. Generally speaking, the abnormal sound level, like the vibration noise, does not show obvious sharp increases or decreases and remains within a relatively controllable range.

[0049] By dividing the first data stream aggregation unit and the second data stream aggregation unit, the data of the rolling bearing from initial operation to long-term use is covered. The first working data stream provides the "benchmark data" of the rolling bearing in the initial stage, reflecting the standard state of the equipment and helping to compare with subsequent data. The second working data stream contains vibration, audio, and load data, reflecting the dynamic behavior of the bearing during operation and covering various environments and working conditions; The vibration frequency, vibration amplitude, and audio data during the initial operation provide an accurate benchmark for subsequent aging assessment and help to identify manufacturing defects or installation problems in the initial state; By real-time monitoring the second operating vibration data stream and the second audio data stream, the state changes of the bearing during operation can be accurately captured. The addition of load data enables the performance analysis of the bearing to combine with the actual working conditions and evaluate the impact of load fluctuations on the bearing state. The changes in vibration frequency and vibration amplitude can reflect the aging of internal components of the bearing, insufficient lubrication, or external environmental impacts.

[0050] Specifically, the first bearing analysis module includes a first-level data stream extraction unit and a bearing standard coefficient calculation unit; The first-level data stream extraction unit is connected to the bearing standard coefficient calculation unit;

[0051] The first-level data stream extraction unit is used to extract the first operating vibration data stream and the first audio data stream;

[0052] The bearing standard coefficient calculation unit is used to substitute the first operating vibration data stream and the first audio data stream into the bearing standard coefficient calculation formula to calculate the bearing standard coefficient. The bearing standard coefficient calculation formula is:

[0053] ;

[0054] In the formula: is the bearing standard coefficient, is the number of data items in the first operating vibration frequency, the first operating vibration amplitude, the first operating sound frequency, and the first operating sound amplitude, is the i-th item in the first operating vibration frequency, the first operating vibration amplitude, the first operating sound frequency, and the first operating sound amplitude at time t, is the initial operating duration of the rolling bearing, is the maximum value corresponding to the i-th item among the first operating vibration frequency, the first operating vibration amplitude, the first operating sound frequency, and the first operating sound amplitude. is the minimum value corresponding to the i-th item among the first operating vibration frequency, the first operating vibration amplitude, the first operating sound frequency, and the first operating sound amplitude. is the average value corresponding to the i-th item among the first operating vibration frequency, the first operating vibration amplitude, the first operating sound frequency, and the first operating sound amplitude.

[0055] The first-level data stream extraction unit extracts multi-dimensional data including vibration frequency, vibration amplitude, sound frequency, and sound amplitude, which can comprehensively capture the dynamic characteristics of the bearing during initial operation, provide complete information on the bearing's working state, and ensure the accuracy of calculations; the bearing standard coefficient, as the reference value of the bearing, reflects the health state of the bearing in the initial operation stage, can be used as a comparison benchmark during subsequent operation, helps analyze the aging or fault trend of the bearing. By comparing with subsequent operation data, it can accurately identify whether there are abnormalities in the bearing and perform targeted repairs; the calculated coefficient is a dynamic standard value used to monitor the working state of the bearing in real-time. When there are abnormal fluctuations in the bearing standard coefficient, potential faults can be predicted in advance to avoid sudden failure of the bearing during operation; based on the analysis of the coefficient, the maintenance cycle can be scientifically planned to avoid premature or late maintenance decisions. Only when there are obvious changes in the bearing standard coefficient will a maintenance warning be triggered, thereby optimizing the equipment's usage efficiency and maintenance costs.

[0056] Specifically, the second bearing analysis module includes a second-level data stream extraction unit, a data update unit, and a bearing aging coefficient calculation unit; the second-level data stream extraction unit is connected to the data update unit, and the data update unit is connected to the bearing aging coefficient calculation unit;

[0057] The second-level data stream extraction unit is used to extract the second operating vibration data stream and the second audio data stream;

[0058] The data update unit is used to update the data in real-time according to the first working data stream and the second working data stream to obtain the bearing standard coefficient and the basic statistical data. The basic statistical data are the maximum value, the minimum value, and the average value corresponding to the first operating vibration frequency, the first operating vibration amplitude, the first operating sound frequency, and the first operating sound amplitude respectively. When the second operating data stream reaches the s-th moment, the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude at the s-th moment is updated. Among them, the average value update formula for each data item is:

[0059] ;

[0060] ;

[0061] where: is the average value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the s-th moment, is the average value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the (s - 1)-th moment, is the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude at the s-th moment, is the number of data items of the j-th item up to the (s - 1)-th moment, is the number of data items of the j-th item up to the s-th moment;

[0062] The update formulas for the maximum and minimum values of each data item are:

[0063] ;

[0064] ;

[0065] where: is the maximum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the s-th moment, is the maximum value screening function, is the minimum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the s-th moment, is the minimum value screening function, is the maximum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the (s - 1)-th moment, is the minimum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the (s - 1)-th moment, is the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude at the s-th moment, is the number of data items of the j-th item up to the (s - 1)-th moment;

[0066] The bearing aging coefficient calculation unit is used to obtain the bearing standard coefficient, substitute the bearing standard coefficient, the second operating vibration data stream, and the second audio data stream into the bearing aging coefficient calculation formula to calculate the bearing aging coefficient. The bearing aging coefficient calculation formula is:

[0067] ;

[0068] where: is the bearing aging coefficient during the h-th operation of the rolling bearing, is the bearing standard coefficient, is the average value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the s-th moment, is the maximum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the s-th moment, is the minimum value corresponding to the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude up to the s-th moment, is the number of data items among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude, is the j-th item among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude at the s-th moment, is the operating duration of the rolling bearing for the h-th time.

[0069] As Figure 5 is the analysis trend chart of the bearing aging coefficient shown. The aging coefficient shows a slow upward trend, indicating that with the passage of time, the aging degree of the equipment or material gradually deepens. The aging coefficient in January is about 0.09, and then it rises slowly, reaching 0.12 around October, and there are slight fluctuations in November and December. Overall, the aging trend meets the expectations, that is, with the passage of time, the material or equipment gradually ages, but there is no sudden aging acceleration phenomenon.

[0070] The data stream extraction module integrates the second running vibration data stream and the second audio data stream, covering multi-dimensional data such as vibration frequency, vibration amplitude, sound frequency, and sound amplitude. This data integration can comprehensively reflect the bearing operation status, capture physical vibration characteristics, and identify acoustic anomalies. Using the bearing standard coefficients in the first working data stream and the second working data stream, a comparison system from the initial standard to dynamic operation is established, providing a strong benchmark for aging assessment. Through the comparison of aging coefficients in multiple runs, the long-term aging trend of the bearing can be captured to help predict its service life. The real-time updated statistical data and aging coefficients can quickly capture anomalies during operation (such as abnormal increase in vibration amplitude or severe fluctuation in audio signal), provide early warnings, and prevent the expansion of faults. According to the changes in aging coefficients, the maintenance cycle can be scientifically planned to prevent premature or late maintenance, optimize maintenance resources, and quantitatively evaluate the specific impacts of different environmental factors (such as load, lubrication conditions, operating temperature) on bearing aging by combining statistical analysis of different data dimensions (vibration and audio).

[0071] As Figure 2 shown in the schematic diagram of the working process of the second bearing analysis module, including data stream input, data update, and coefficient calculation. As Figure 2 shown, the data stream 1 is the first working data stream and the second working data stream. The data stream 1 is respectively extracted, updated, and calculated. The data stream 2, that is, the required data stream, is obtained by extracting the data stream 1. The data stream 2 and the data stream 1 are merged for an update operation to obtain the data stream 3. The data stream 3 and the data stream 1 are merged again to calculate the bearing aging coefficient during the operation of the rolling bearing.

[0072] Specifically, the data sensitivity monitoring module includes a three-level data stream extraction unit, a load fluctuation coefficient calculation unit, and a fluctuation sensitivity index monitoring unit;

[0073] The three-level data stream extraction unit is used to extract the first load data stream;

[0074] The load fluctuation coefficient calculation unit is used to construct a load fluctuation model based on the first load data stream to obtain the load fluctuation coefficient of the bearing in real time. The formula of the load fluctuation model is:

[0075] ;

[0076] In the formula: is the load fluctuation coefficient during the hth operation of the rolling bearing, is the load value of the rolling bearing at time s, is the load value of the rolling bearing at time s - 1, is the maximum load value of the rolling bearing up to time s, is the minimum load value of the rolling bearing up to time s, is the number of data items among the second operating vibration frequency, the second operating vibration amplitude, the second operating sound frequency, and the second operating sound amplitude. is the operating duration of the rolling bearing at the h-th time;

[0077] such as Figure 6 As shown in the fluctuation analysis trend chart of the load fluctuation coefficient, the load shows a certain fluctuation between January and December, and the overall value varies between 2.0 KN and 2.7 KN. The load has obvious peaks in March, June, and September, indicating that there may be higher load pressures in these months, while it slightly decreases in April and August, which may be the result of load reduction or equipment adjustment.

[0078] The fluctuation sensitivity index monitoring unit is used to establish a fluctuation sensitivity analysis model based on the load fluctuation coefficient and the bearing aging coefficient to calculate the fluctuation sensitivity index. The formula of the fluctuation sensitivity analysis model is:

[0079] ;

[0080] In the formula: is the fluctuation sensitivity index, is the bearing aging coefficient during the h-th operation of the rolling bearing, is the load fluctuation coefficient during the h-th operation of the rolling bearing, is the bearing aging coefficient during the (h - 1)-th operation of the rolling bearing, is the load fluctuation coefficient during the (h - 1)-th operation of the rolling bearing, is the number of operations of the rolling bearing.

[0081] Normalize the real-time load change, dynamically quantify the load fluctuation situation during each bearing operation. By considering the load value change at each time point and the current maximum and minimum values, it can sensitively capture the degree and trend of load fluctuation; comprehensively consider the change amounts of the load fluctuation coefficient and the bearing aging coefficient, quantitatively describe the sensitivity of the bearing to load fluctuation. By comparing the aging coefficients and load fluctuation coefficients in different operation cycles, accurately predict the change trend of bearing performance; if the change amplitude of the load fluctuation coefficient is too large, it may indicate the instability of the load condition, thus identifying potential mechanical problems in advance. According to the result of the fluctuation sensitivity index, it is possible to identify the load fluctuation modes that have a greater impact on the bearing and adjust the load parameters to reduce the impact on the bearing.

[0082] In an embodiment of the present invention, the working data stream of a rolling bearing is obtained; the working data stream includes a first working data stream obtained when the rolling bearing runs for the first time and a second working data stream for each run after the first run of the rolling bearing. The bearing standard coefficient is analyzed based on the first working data stream, and the bearing aging coefficient is analyzed based on the bearing standard coefficient and the second working data stream. Based on the incremental learning model, the working data stream of the rolling bearing can be analyzed in real time, and the bearing aging coefficient and the fluctuation sensitivity index can be dynamically updated; by comparing the initial operation data and the subsequent operation data, the standard state and the aging trend of the bearing can be accurately evaluated; by calculating the aging coefficient and the fluctuation sensitivity index, the possible fault trend of the bearing can be predicted, avoiding the interruption of the equipment due to bearing failure during critical tasks, comprehensively improving the monitoring and management efficiency of the rolling bearing, not only extending the service life of the bearing, but also reducing the operation and maintenance costs.

[0083] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

[0084] Finally: The above is only the preferred solution of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An incremental learning data flow analysis platform, comprising a data flow import module, a first bearing analysis module, a second bearing analysis module and a data sensitivity monitoring module, characterized in that: The data stream import module is used to obtain the working data stream of the rolling bearing; the working data stream includes a first working data stream obtained when the rolling bearing is first run and a second working data stream obtained when the rolling bearing is run for the first time and each time the rolling bearing runs after the first run; The first bearing analysis module is used to analyze the bearing standard coefficient according to the first working data stream; the second bearing analysis module is used to analyze the bearing aging coefficient according to the bearing standard coefficient and the second working data stream; the second bearing analysis module includes a data update unit; the data update unit is used to update data in real time according to the first working data stream and the second working data stream to obtain the bearing standard coefficient and basic statistical data, the basic statistical data are the maximum value, minimum value and average value corresponding to the first operation vibration frequency, the first operation vibration amplitude, the first operation sound frequency and the first operation sound amplitude respectively. When the second operation data stream reaches time s, the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude at time s is Make updates; The data sensitivity monitoring module includes a three-level data flow extraction unit, a load fluctuation coefficient calculation unit and a fluctuation sensitivity index monitoring unit; the three-level data flow extraction unit is used to extract the first load data flow; The load fluctuation coefficient calculation unit is used to construct a load fluctuation model according to the first load data stream to obtain the load fluctuation coefficient of the bearing in real time. The formula of the load fluctuation model is: ； Where: is the load fluctuation coefficient of the hth rolling bearing operation, is the load value of the rolling bearing at time s, is the load value of the rolling bearing at time s-1, is the maximum load value of the rolling bearing up to time s, is the minimum load value of the rolling bearing up to time s, is the number of data items in the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency, and the second operation sound amplitude, is the h-th running time of the rolling bearing; The fluctuation sensitivity index monitoring unit is used to establish a fluctuation sensitivity analysis model according to the load fluctuation coefficient and the bearing aging coefficient to calculate the fluctuation sensitivity index. The formula of the fluctuation sensitivity analysis model is: ； Where: is the volatility sensitivity index, is the bearing aging coefficient during the hth rolling bearing operation, is the load fluctuation coefficient of the hth rolling bearing operation, is the bearing aging coefficient during the h-1th rolling bearing operation, is the load fluctuation coefficient of the rolling bearing during the h-1th operation, is the number of rolling bearing operations.

2. The incremental learning data flow analysis platform according to claim 1, characterized in that: For the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude at time s The formula for updating is: ； ； Where: is the average value corresponding to the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude up to time s, is the average value corresponding to the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude up to time s-1, is the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude at time s, is the number of data of the jth item up to time s-1, is the number of data of the jth item up to time s.

3. The incremental learning data flow analysis platform according to claim 1, characterized in that: The data stream import module includes a first data stream aggregation unit and a second data stream aggregation unit; The first data stream aggregation unit is used to obtain a first working data stream obtained when the rolling bearing is first operated; the first working data stream includes a first operating vibration data stream and a first audio data stream; the first operating vibration data stream includes a first operating vibration frequency and a first operating vibration amplitude of the rolling bearing; The first audio data stream includes a first operating sound frequency and a first operating sound amplitude of the rolling bearing; The second data stream aggregation unit is used to obtain the second working data stream of each operation of the rolling bearing after the initial operation; The second working data stream includes a second operating vibration data stream, a second audio data stream and a first load data stream; The second operation vibration data stream includes the second operation vibration frequency and the second operation vibration amplitude of the rolling bearing; the second audio data stream includes the second operation sound frequency and the second operation sound amplitude of the rolling bearing; and the first load data stream includes the load value of the rolling bearing.

4. The incremental learning data flow analysis platform according to claim 1, characterized in that: The update formula for the maximum and minimum values ​​of each data item is: ； ； Where: is the maximum value corresponding to the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude up to time s, is the maximum value screening function, is the minimum value corresponding to the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude up to time s, is the minimum value screening function, is the maximum value corresponding to the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude up to time s-1, is the minimum value corresponding to the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude up to time s-1, is the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude at time s, is the number of data of the jth item up to time s-1.

5. The incremental learning data flow analysis platform according to claim 1, characterized in that: The first bearing analysis module includes a primary data stream extraction unit and a bearing standard coefficient calculation unit; the primary data stream extraction unit is connected to the bearing standard coefficient calculation unit; The primary data stream extraction unit is used to extract the first operating vibration data stream and the first audio data stream; The bearing standard coefficient calculation unit is used to substitute the first running vibration data stream and the first audio data stream into the bearing standard coefficient calculation formula to calculate the bearing standard coefficient. The bearing standard coefficient calculation formula is: ； Where: is the bearing standard coefficient, is the number of data items in the first operation vibration frequency, the first operation vibration amplitude, the first operation sound frequency, and the first operation sound amplitude, is the i-th item among the first operation vibration frequency, the first operation vibration amplitude, the first operation sound frequency and the first operation sound amplitude at time t, The initial running time of the rolling bearing. is the maximum value corresponding to the i-th item among the first operation vibration frequency, the first operation vibration amplitude, the first operation sound frequency and the first operation sound amplitude, is the minimum value corresponding to the i-th item among the first operation vibration frequency, the first operation vibration amplitude, the first operation sound frequency and the first operation sound amplitude, It is the average value corresponding to the i-th item among the first operation vibration frequency, the first operation vibration amplitude, the first operation sound frequency and the first operation sound amplitude.

6. The incremental learning data flow analysis platform according to claim 5, characterized in that: The bearing aging coefficient calculation unit is used to obtain the bearing standard coefficient, substitute the bearing standard coefficient, the second operating vibration data stream and the second audio data stream into the bearing aging coefficient calculation formula, and calculate the bearing aging coefficient. The bearing aging coefficient calculation formula is: ； Where: is the bearing aging coefficient during the hth rolling bearing operation, is the bearing standard coefficient, is the average value corresponding to the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude up to time s, is the maximum value corresponding to the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude up to time s, is the minimum value corresponding to the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude up to time s, is the number of data items in the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency, and the second operation sound amplitude, is the jth item among the second operation vibration frequency, the second operation vibration amplitude, the second operation sound frequency and the second operation sound amplitude at time s, is the h-th running time of the rolling bearing.

Images