Rolling bearing state evaluation method based on contribution weight and life prediction system
By using a rolling bearing condition assessment method based on contribution weights and employing a convex optimization model and double exponential function fitting, automated health condition assessment and life prediction of bearing equipment are achieved. This solves the problem of relying on human experience in existing technologies and improves the accuracy of fault identification and degradation trend prediction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NORTHEASTERN UNIV CHINA
- Filing Date
- 2026-04-22
- Publication Date
- 2026-05-29
AI Technical Summary
Existing technologies rely on human experience and prior knowledge in assessing the health status of bearing equipment, requiring complex feature processing and combined testing, making it difficult to achieve early fault identification and accurate prediction of performance degradation trends.
A rolling bearing condition assessment method based on contribution weights is adopted. Vibration signal data is collected, noise is reduced, and then square envelope spectrum transformation is performed to construct a convex optimization model. The fault characteristic frequency components are adaptively weighted, and the health index assessment and remaining life prediction are achieved by combining root mean square value and double exponential function fitting.
It achieves automated fault diagnosis without manual feature extraction and expert system support, accurately identifies initial faults and predicts performance degradation trends, and improves the safety and reliability of bearing equipment.
Smart Images
Figure CN122108602A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the technical field of fault diagnosis methods, specifically relating to a rolling bearing condition assessment method and life prediction system based on contribution weight. Background Technology
[0002] As a key component of rotor systems, bearings play a crucial role in maintaining the rotation of mechanical parts, reducing frictional resistance, and ensuring rotational accuracy. Bearings typically operate in harsh environments, and their health inevitably declines over time. When damage exceeds a certain threshold, they become unable to fulfill planned tasks and requirements, leading to economic losses or even serious accidents. Therefore, researching bearing health status assessment is of great significance for developing highly safe and reliable mechanical equipment. Equipment health status assessment generally involves constructing a full life-cycle health index to monitor early failures and predict performance degradation trends. However, implementing real-time monitoring of bearing operation and constructing a health index to accurately diagnose early failures and performance degradation trends presents considerable challenges.
[0003] A method and system for comprehensive bearing health detection based on multi-channel parameters is disclosed in invention document CN116558824A. The method includes the following steps: (1) collecting multi-channel full life cycle signals of the bearing, providing historical warning points, and extracting effective signal components using the information entropy graph method; (2) calculating the weighted comprehensive health index of negative entropy and spectral negative entropy for the effective signal components, and judging the normal and fault state indication sequences based on prior knowledge; (3) using the warning significance P-value of each channel test sequence as a vote to obtain the decision-fusion bearing monitoring results. Bearing condition monitoring relies on human experience and prior knowledge, requiring complex feature processing and combined testing processes.
[0004] The invention document with publication number CN115034137A discloses a two-stage hybrid prediction method for bearing remaining life based on RVM and degradation model. The method includes the following steps: (1) collecting bearing vibration signals and extracting the maximum amplitude value in the time domain to construct a health index sequence; (2) using the 3σ criterion to check the bearing failure start time, performing RVM regression with different kernel parameters on the sequence from the failure start time to the inspection time, and obtaining different RVs; (3) using a degradation model combining weighted single, double exponential and polynomial to fit the RVs, using Hausdorff distance to perform similarity analysis between the fitted curve and the true smooth curve, selecting the optimal degradation curve, extrapolating to the failure threshold, and then predicting the bearing remaining life. This method effectively overcomes the limitations of pure model-based and pure data-driven methods. However, similarly, the condition monitoring process relies on human participation and prior knowledge, and requires multiple tests to select the optimal parameters. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this application proposes a rolling bearing condition assessment method and life prediction system based on contribution weights. Based on convex optimization of conditional constraint equations, the method automatically assigns weights to the frequency components of fault features, adaptively detects the bearing's health status, and predicts the bearing's remaining life when the bearing's health status indicates a fault. This method further enhances the subtle features of the system or structure, combining them with evaluation indicators to form health indicators, enabling early fault identification and prediction of performance degradation trends. It does not require manual feature extraction and screening and does not rely on expert systems or prior knowledge.
[0006] In a first aspect, the present invention provides a method for evaluating the condition of rolling bearings based on contribution weights, comprising:
[0007] Vibration signal data during bearing operation are collected as the bearing's full life cycle vibration signal, and noise reduction processing is performed on the bearing's full life cycle vibration signal.
[0008] The square envelope spectrum transform is performed on the noise-reduced bearing full-life-cycle vibration signal to obtain the full-life-cycle square envelope spectrum amplitude.
[0009] Based on the full lifecycle squared envelope spectrum amplitude, a convex optimization model based on the conditional constraint equation is constructed. The convex optimization model is solved to obtain the weight of the squared envelope spectrum amplitude at each running time.
[0010] The root mean square value is used to quantify the weight of the squared envelope spectrum amplitude at each running time to obtain the health index;
[0011] The health status of the bearing is assessed based on health indicators. If the bearing is in a fault state, a bi-exponential function with parameter constraints is used to fit and obtain the predicted remaining life of the bearing.
[0012] The step of performing a square envelope spectrum transformation on the noise-reduced bearing full-life-cycle vibration signal to obtain the full-life-cycle square envelope spectrum amplitude includes:
[0013] The noise-reduced bearing full-life-cycle vibration signal is subjected to Hilbert transform to obtain the analytic signal of the Hilbert transform;
[0014] The analytic signal of the Hilbert transform is subjected to envelope squaring to obtain the squared envelope signal;
[0015] Perform a Fourier transform on the square envelope signal to obtain the amplitude of the square envelope spectrum over the entire lifetime period.
[0016] Based on the amplitude of the squared envelope spectrum over the entire life cycle, a convex optimization model based on the conditional constraint equation is constructed, and the calculation formula is as follows:
[0017] ;
[0018] Constraints:
[0019] ;
[0020] in, Let i = 1, 2, 3, ..., m, where i is the weight corresponding to the amplitude component of the i-th squared envelope spectrum. Let be the weight corresponding to the m-th squared envelope amplitude component. It is the set of weights for the amplitude components of the squared envelope spectrum. Let n be the L1 normalized norm, n be the number of periodic vibration signals throughout the bearing's lifespan, and m be the number of sampling points for the periodic vibration signals throughout the bearing's lifespan. The squared envelope spectrum value of the m-th sampling point of the n-th vibration signal data. It is a diagonal matrix. It is a vector distance used to separate healthy and faulty states. This is a data matrix representing the squared envelope spectrum amplitude over the entire lifespan.
[0021] The method uses the root mean square value to quantify the weight of the squared envelope spectrum amplitude at each running time to obtain health indicators, including:
[0022] The weights of the squared envelope spectrum amplitude at each running time are quantified, and the root mean square value of the weights of the squared envelope spectrum amplitude is calculated. All the calculated root mean square values are used as health indicators.
[0023] The root mean square value of the weights used to calculate the amplitude of the squared envelope spectrum is used as a health indicator, and the calculation formula is as follows:
[0024] ;
[0025] ;
[0026] in, As a health indicator, For all root mean square values, , i=1,2,3,…,n, are the root mean square values of the weights of the i-th squared envelope spectrum amplitude, and m is the number of sampling points of the bearing's full-life-cycle vibration signal.
[0027] The method of obtaining the bearing remaining life prediction result by fitting a bi-exponential function with parameter constraints includes:
[0028] Determine the parameters of the double exponential function;
[0029] Based on health indicators, fit a parameterized biexponential function;
[0030] Based on a parameterized double exponential function, a remaining life prediction curve is plotted with the number of sample files on the horizontal axis and health indicators on the vertical axis. The remaining life prediction result of the bearing is obtained from the remaining life prediction curve.
[0031] The method involves fitting a parameter-constrained biexponential function based on health indicators, calculated as follows:
[0032] ;
[0033] Where a is the scaling factor of the control curve, b is the curve growth rate, h is the vertical displacement of the curve, y is the fitted predicted lifetime, and x is the number of sample files.
[0034] Secondly, this application proposes a lifetime prediction system based on contribution weights, including:
[0035] The noise reduction module is used to collect vibration signal data during the operation of the bearing as the vibration signal of the bearing throughout its entire life cycle, and to perform noise reduction processing on the vibration signal of the bearing throughout its entire life cycle.
[0036] The envelope spectrum transformation module is used to perform square envelope spectrum transformation on the noise-reduced bearing full-life-cycle vibration signal to obtain the full-life-cycle square envelope spectrum amplitude.
[0037] The weight acquisition module is used to construct a convex optimization model based on the conditional constraint equations according to the full life cycle squared envelope spectrum amplitude, solve the convex optimization model, and obtain the weight of the squared envelope spectrum amplitude at each running time.
[0038] The indicator determination module is used to quantify the weight of the squared envelope spectrum amplitude at each running time using the root mean square value to obtain the health indicator.
[0039] The predicted structure output module is used to evaluate the health status of the bearing based on health indicators. If the bearing health status is a fault state, the remaining life prediction result of the bearing is obtained by fitting a bi-exponential function with parameter constraints.
[0040] Thirdly, this application proposes an electronic device comprising: one or more processors, and a memory for storing instructions that, when executed by the one or more processors, cause the one or more processors to perform the rolling bearing state assessment method based on contribution weights.
[0041] Fourthly, this application proposes a computer-readable storage medium storing executable instructions that, when executed, cause a processor to perform the aforementioned rolling bearing state evaluation method based on contribution weights.
[0042] Fifthly, this application proposes a computer program product, including a computer program or instructions that, when executed by a processor, implement the aforementioned rolling bearing condition assessment method based on contribution weights.
[0043] Beneficial effects:
[0044] This application proposes a rolling bearing condition assessment method and life prediction system based on contribution weights. It enables an adaptive fault diagnosis method that automatically assigns weights to the frequency components of fault features, eliminating the need for manual feature extraction and filtering, and avoiding reliance on expert systems and prior knowledge. The method utilizes a health index obtained through a convex optimization model based on conditional constraint equations, exhibiting clear initial fault initiation points and degradation trends. This allows for the performance degradation assessment and prediction of bearing equipment, unlike black-box models such as neural networks, and possesses a clear mathematical definition and theoretical foundation. Furthermore, fitting with a bi-exponential function with parameter constraints yields superior fitting results compared to natural functions. Attached Figure Description
[0045] Figure 1 Flowchart of the rolling bearing condition assessment method based on contribution weight according to an embodiment of the present invention;
[0046] Figure 2 A schematic diagram of the rolling bearing condition assessment method based on contribution weight according to an embodiment of the present invention;
[0047] Figure 3 The following is a schematic diagram of the fitting curve based on the natural exponential function in an embodiment of the present invention, wherein (a) simulation results of TSP+50 fault data; (b) simulation results of TSP+100 fault data; (c) simulation results of TSP+150 fault data; and (d) simulation results of TSP+200 fault data.
[0048] Figure 4 The fitting curves based on the double natural exponential function in this embodiment of the invention are as follows: (a) simulation results of TSP+50 fault data; (b) simulation results of TSP+100 fault data; (c) simulation results of TSP+150 fault data; (d) simulation results of TSP+200 fault data.
[0049] Figure 5 Sample diagram of bearing life vibration signal sequence according to an embodiment of the present invention;
[0050] Figure 6 Squared envelope spectrum of partial samples of bearing life-cycle vibration signal according to an embodiment of the present invention;
[0051] Figure 7 A weighted result diagram of the squared envelope spectrum of a portion of the bearing's full-life-cycle vibration signal according to an embodiment of the present invention;
[0052] Figure 8 The root mean square value of the squared envelope spectrum amplitude weights in this embodiment of the invention is used as a health indicator.
[0053] Figure 9 A block diagram illustrating the principle of the lifetime prediction system based on contribution weight according to an embodiment of the present invention. Detailed Implementation
[0054] The specific implementation methods of this application will be further described in detail below with reference to the accompanying drawings and embodiments.
[0055] Example 1:
[0056] This embodiment provides a rolling bearing condition assessment method based on contribution weights, such as... Figure 1 , Figure 2 As shown, it includes:
[0057] Step S1: Collect vibration signal data during bearing operation as the bearing's full life cycle vibration signal, and perform noise reduction processing on the bearing's full life cycle vibration signal;
[0058] In this embodiment, vibration signal data during bearing operation is collected by sensors, and the vibration signal data during bearing operation is used as the vibration signal of the bearing throughout its entire life cycle. The vibration signal of the bearing throughout its entire life cycle is then processed for noise reduction.
[0059] The time it takes for a bearing to reach a fully damaged state is very long. In the experiment of collecting bearing vibration signals throughout its entire life cycle, bearing vibration signal data is recorded for 1 second at fixed intervals. This effectively reduces the amount of data while preserving the basic state information of the bearing. In this embodiment, a total of n bearing vibration signal samples are collected, and m vibration signal data sampling points are collected in each sample. The bearing vibration signal throughout its entire life cycle is represented as follows:
[0060] ;
[0061] in, The signal represents the vibration signal throughout the bearing's entire life cycle, where n is the number of samples taken during the bearing's entire life cycle, and m is the number of sampling points during the sampling process. This represents the m-th sampling point in the n-th vibration signal sequence sample.
[0062] The bearing life-cycle vibration signals collected in experiments are often accompanied by complex environmental noise, which may cause errors in subsequent frequency domain feature analysis. Removing interference components from the signal can highlight fault characteristic frequencies or patterns, improving analysis accuracy. Wavelet transform denoising has excellent multi-resolution analysis capabilities and is very effective in processing non-stationary signals and multi-scale analysis. Using wavelet denoising to denoise the collected vibration signal sample sequence can improve analysis accuracy. (Image: Bearing life-cycle vibration signal after denoising) Represented as:
[0063] ;
[0064] in, This is the vibration signal of the bearing throughout its entire life cycle after noise reduction processing. This represents the wavelet decomposition, thresholding, and signal reconstruction operations of the i-th vibration signal sample sequence. , is represented as:
[0065] ;
[0066] in, Represents the approximate component (low-frequency signal) of scale j; Represents the detail components (high-frequency signals) at scale j; J represents the maximum number of decomposition levels. Indicates details of the components Thresholding is performed to remove noise; The threshold value for the j-th layer can be calculated using an empirical formula or statistical method. The calculation method is as follows:
[0067] ;
[0068] in, The standard deviation of noise can be calculated using median estimation. For ease of representation, we will use the median of the denoised vibration signal sequence sample. Still recorded as .
[0069] Step S2: Perform square envelope spectrum transformation on the noise-reduced bearing full-life-cycle vibration signal to obtain the full-life-cycle square envelope spectrum amplitude;
[0070] In this embodiment, the collected bearing full-life-cycle vibration signal is subjected to square envelope spectrum transformation to obtain the full-life-cycle square envelope spectrum amplitude. , is represented as:
[0071] ;
[0072] in, The amplitude of the squared envelope spectrum over the entire life cycle. Represents the i-th vibration signal sequence sample The result of the squared envelope spectrum transformation, ; This represents the amplitude of the j-th sampling point in the i-th vibration signal sequence sample. .
[0073] The step of performing a square envelope spectrum transformation on the noise-reduced bearing full-life-cycle vibration signal to obtain the full-life-cycle square envelope spectrum amplitude includes:
[0074] Step S2.1: Perform Hilbert transform on the noise-reduced bearing life-cycle vibration signal to obtain the analytic signal of the Hilbert transform;
[0075] In this embodiment, firstly, the n vibration signals of the bearing throughout its entire life cycle after noise reduction are subjected to Hilbert transform to obtain the analytic signal of the Hilbert transform. :
[0076] ;
[0077] in, For the analytic signal of the Hilbert transform, Indicates the modulo operation of a complex number; This represents the operating data throughout the entire life cycle of the equipment; This represents the i-th vibration signal sequence sample. ; This represents the analytic signal corresponding to the j-th sampling point of the i-th vibration signal sequence sample. ; This is represented by the Hilbert transform of the vibration signal sample sequence acquired in the i-th acquisition:
[0078] ;
[0079] in, Indicates to Perform the Hilbert transform.
[0080] Step S2.2: Perform envelope squaring on the analytic signal of the Hilbert transform to obtain the squared envelope signal, calculated as follows:
[0081] ;
[0082] in, The signal has a squared envelope. To square the signal, This represents the squared envelope amplitude corresponding to the j-th sampling point of the i-th vibration signal sequence sample. . Let Hilbert be the analytic signal corresponding to the m-th sampling point of the nth vibration signal sequence sample. This represents the square envelope amplitude corresponding to the m-th sampling point of the nth vibration signal sequence sample.
[0083] Step S2.3: Perform a Fourier transform on the square envelope signal to obtain the amplitude of the square envelope spectrum over the entire lifetime period.
[0084] ;
[0085] in, The amplitude of the squared envelope spectrum over the entire life cycle. This represents the squared envelope spectrum of the i-th vibration signal sequence. ; This represents the square envelope spectrum amplitude corresponding to the j-th sampling point of the i-th vibration signal sequence sample. , For the Fourier transform, since the spectrum of the Fourier transform has conjugate symmetry, we only need to take the squared amplitude of the first half of the envelope spectrum, which is expressed as:
[0086] ;
[0087] Step S3: Based on the full life cycle squared envelope spectrum amplitude, construct a convex optimization model based on the conditional constraint equation, solve the convex optimization model, and obtain the weight of the squared envelope spectrum amplitude at each running time.
[0088] In this embodiment, a convex optimization model based on the full-lifetime squared envelope spectrum amplitude (SES) is constructed using conditional constraint equations to adaptively determine the frequency weight of each frequency of the squared envelope spectrum amplitude. .
[0089] The specific process is as follows:
[0090] A convex optimization model based on conditional constraint equations is constructed to adaptively determine the frequency weight of each frequency in the squared envelope spectrum amplitude. The optimization model is expressed as:
[0091] ;
[0092] Constraints: ;
[0093] in, This is represented by the weight corresponding to the amplitude component of the i-th squared envelope spectrum; Let be the weight corresponding to the m-th squared envelope amplitude component. It is the set of weights for the amplitude components of the squared envelope spectrum. It is represented as the L1 normalization norm, which is used to enhance the sparsity of fault feature frequency weights.
[0094] In the constraints, It is represented as a diagonal matrix to distinguish between healthy and abnormal data; It is represented as a vector distance, used to separate healthy and abnormal data; Represented as a squared envelope spectrum data matrix, the matrix is defined as follows: It consists of health data and hypothetical "abnormal" data. Let m be the squared envelope spectrum value of the m-th sampling point of the n-th vibration signal data. The initial 10 runs of the bearing are considered healthy data, and the subsequent 10 runs are considered hypothetical "abnormal" data. The hypothetical "abnormal" data in the matrix is continuously updated as the bearing operates (at this time...). ),matrix Represented as:
[0095] ;
[0096] Constraints This indicates that the squared envelope spectrum amplitude is weighted to separate healthy and abnormal data. The specific expression is as follows:
[0097] ;
[0098] Solving the constraint-condition extremum equation yields the squared envelope spectrum weights at the i-th running time. The weights of the squared envelope spectrum amplitude at each running time step are solved based on the constructed convex optimization model. .
[0099] The specific process is as follows:
[0100] Solving the constraint-condition extremum equation yields the squared envelope spectrum weights at the i-th runtime. While keeping the healthy data unchanged, a new matrix is continuously formed by updating the "abnormal" data. For example: Using data 1-10 as healthy data and 12-21 as hypothetical "abnormal" data, solving the constraint extremum equation yields the squared envelope spectrum weight result for the 11th running time. Then, using 13-22 as hypothetical "abnormal" data, solving the constraint extremum equation yields the squared envelope spectrum weight result for the 12th running time, and so on. Solving for the squared envelope spectrum amplitude weight result for all running sequence samples yields the same result. .
[0101] Step S4: Use the root mean square value to quantify the weight of the squared envelope spectrum amplitude at each running time to obtain the health index;
[0102] In this embodiment, the root mean square (RMS) value is used to quantify the weighting results. Obtain health indicators This enables early fault monitoring and health status assessment of bearings.
[0103] The specific process is as follows:
[0104] The weighted results of all operating vibration signal sequence samples cannot directly reflect the occurrence and degradation trend of faults. The root mean square (RMS) value of the weighted results can be used to obtain the bearing's health index, enabling the determination of the initial fault initiation point and performance degradation trend. Health Index Represented as:
[0105] ;
[0106] in, This represents the solution for the root mean square value of the weighting result of the i-th iteration:
[0107] ;
[0108] in, As a health indicator, For all root mean square values, Let i = 1, 2, 3, ..., n, be the root mean square value of the weight of the i-th squared envelope spectrum amplitude, and m be the number of sampling points of the bearing's full-life-cycle vibration signal.
[0109] Health indicator results Plotting in a two-dimensional plane, with the x-axis representing time and the y-axis representing the health index (HI), allows observation of the timing of failures and assessment of performance degradation trends.
[0110] By acquiring bearing vibration signal data in real time and establishing bearing operating status evaluation indicators through a convex optimization method based on conditional constraint equations, early fault warning and real-time status assessment of bearing equipment can be achieved.
[0111] Step S5: Evaluate the health status of the bearing based on health indicators. If the bearing health status is a fault state, use a bi-exponential function with parameter constraints to fit and obtain the bearing remaining life prediction result.
[0112] The method of obtaining the bearing remaining life prediction result by fitting a bi-exponential function with parameter constraints includes:
[0113] Step S5.1: Determine the parameters of the double exponential function;
[0114] Step S5.2: Fit a parametric biexponential function based on health indicators;
[0115] In this embodiment, based on health indicators, the degradation curves all resemble exponential function curves. Therefore, a natural exponential function fitting curve can be designed to fit the performance degradation curve after the initial failure. A commonly used form of the natural exponential function is:
[0116] ;
[0117] Where 'a' is the scaling factor controlling the curve; 'b' determines the curve's growth rate, increasing when b > 0 and decreasing when b < 0; and 'h' is the vertical displacement of the curve. However, a single natural exponential function can only describe a single growth rate or decay rate. For complex trend curves, a combination of two natural exponential functions can be introduced to improve the adaptability and accuracy of the fitted function. The double natural exponential function takes the form:
[0118] ;
[0119] Where a is the scaling factor of the control curve, b is the curve growth rate, y is the fitted predicted lifetime, and x is the number of sample files.
[0120] By designing five parameters for the double natural exponential function, a stronger fit to the sample data can be achieved. However, more parameters also bring greater fitting difficulty. If there are few data points, the fitted function may exhibit an excessively high rate of decline, leading to incorrect predictions of future performance degradation trends. Therefore, when designing the double natural exponential function fitting, it is necessary to impose certain constraints on the parameter value range to avoid overfitting. The time point of the detected initial fault occurrence is taken as the starting prediction data point, denoted as TSP.
[0121] The blue curve represents the health index curve based on RMS, while the red curve represents the health index fitting curve obtained using the three fitting methods described above. Figure 3 In (a), the bearing's TSP time is sample data number 527. Its health indicators are continuously monitored. When 50 fault sample data points have been run, a natural exponential function is fitted using the health indicator data from TSP to TSP+50 fault data points, resulting in the red performance degradation curve. The bearing's expected final lifespan is determined to be at operating sample time number 765. Similarly, from... Figure 3 In (b), when 100 fault sample data points were collected, a natural exponential function was fitted, resulting in a red performance degradation curve. The predicted final bearing life is at the time of the 850th operating sample. Figure 3 In (c), when 150 fault sample data were collected, a natural exponential function fitting was performed, resulting in the red performance degradation curve. The predicted final bearing life is at the time of operating sample number 873. Figure 3 In (d), when 200 fault sample data were collected, a natural exponential function was fitted, and the resulting red performance degradation curve predicted the final bearing life at the time of the 870th operating sample.
[0122] Depend on Figure 4 In (a), when 50 fault sample data were collected, a natural exponential function was fitted, resulting in the red performance degradation curve. The predicted final bearing life is at the time of the 890th operating sample. Figure 4In (b), when 100 fault sample data points were collected, a natural exponential function was fitted, resulting in a red performance degradation curve. The predicted final bearing life was at the time of operating sample number 973. Figure 4 In (c), when 150 fault sample data were collected, a natural exponential function fitting was performed, resulting in the red performance degradation curve. The predicted final bearing life is at the time of operating sample number 936. Figure 4 In (d), when 200 fault sample data were collected, the natural exponential function was fitted to obtain the red performance degradation curve, and the final bearing life was predicted to be at the time of the 879th operating sample.
[0123] Depend on Figure 3 and Figure 4 It can be observed that the prediction results based on the natural exponential function and the double natural exponential function are basically consistent: the single natural exponential function predicts a gradually increasing final lifespan as the amount of data increases, eventually converging at sample time 870; the double natural exponential function predicts a gradually decreasing final lifespan, converging at sample time 879. The predicted final lifespan is close to 90% of the actual operating life of the bearing. At this point, equipment maintenance is economically efficient, effectively avoiding wasted bearing performance and unexpected downtime due to damage.
[0124] Step S5.3: Based on the parameter-constrained double exponential function, plot the remaining life prediction curve with the number of sample files on the horizontal axis and health indicators on the vertical axis. From the remaining life prediction curve, obtain the bearing remaining life prediction result.
[0125] To further illustrate the beneficial effects of this embodiment, this embodiment uses theoretical analysis of a set of bearing life vibration signals to illustrate the invention's ability to identify early faults and judge performance degradation trends in bearing equipment.
[0126] Figure 5 The image shows a sample of vibration signal sequences throughout the entire life cycle of the bearing used in this embodiment. Figure 6 The square envelope spectrum of a portion of the bearing's full-life-cycle vibration signal is obtained using the formula... Calculate the theoretical outer ring fault characteristic frequency At the same time Figure 6 The corresponding ones were found in all of them. Second harmonic and triple frequency . Figure 7 The diagram shows the weighting results of the square envelope spectrum of a portion of the vibration signal throughout the bearing's life cycle. It is observed that when a fault occurs, the amplitude weights of the square envelope spectrum become significantly concentrated. The weights of the fault frequency characteristic amplitudes are significantly different from those of other amplitudes, indicating that this characteristic frequency component plays a major role in the bearing fault. Figure 8This diagram illustrates the variation of the root mean square value of the bearing's full life cycle weighting results over time. The convex optimization method based on the conditional constraint equation assigns the root mean square value of the square envelope spectrum amplitude weight to have better sensitivity at the early failure time. Furthermore, the performance degradation trend over time approximates an exponential function, and under certain conditions, it can perform prediction of the finite remaining service life.
[0127] Example 2:
[0128] This embodiment proposes a lifetime prediction system based on contribution weights, such as... Figure 9 As shown, it includes: a noise reduction processing module, an envelope spectrum transformation module, a weight acquisition module, an index determination module, and a prediction structure output module. The noise reduction processing module is connected to the envelope spectrum transformation module, the envelope spectrum transformation module is connected to the weight acquisition module, the weight acquisition module is connected to the index determination module, and the index determination module is connected to the prediction structure output module.
[0129] The noise reduction module is used to collect vibration signal data during the operation of the bearing as the vibration signal of the bearing throughout its entire life cycle, and to perform noise reduction processing on the vibration signal of the bearing throughout its entire life cycle.
[0130] The envelope spectrum transformation module is used to perform square envelope spectrum transformation on the noise-reduced bearing full-life-cycle vibration signal to obtain the full-life-cycle square envelope spectrum amplitude.
[0131] The weight acquisition module is used to construct a convex optimization model based on the conditional constraint equations according to the full life cycle squared envelope spectrum amplitude, solve the convex optimization model, and obtain the weight of the squared envelope spectrum amplitude at each running time.
[0132] The indicator determination module is used to quantify the weight of the squared envelope spectrum amplitude at each running time using the root mean square value to obtain the health indicator.
[0133] The predicted structure output module is used to evaluate the health status of the bearing based on health indicators. If the bearing health status is a fault state, the remaining life prediction result of the bearing is obtained by fitting a bi-exponential function with parameter constraints.
[0134] Example 3:
[0135] This embodiment proposes an electronic device, including: one or more processors, and a memory for storing instructions, which, when executed by the one or more processors, cause the one or more processors to execute the rolling bearing state evaluation method based on contribution weight.
[0136] The electronic device may be a mobile phone, computer, or tablet computer, etc., and includes a memory and a processor. The memory stores a computer program, which, when executed by the processor, implements the rolling bearing state evaluation method based on contribution weights as described in the embodiments. It is understood that the electronic device may also include input / output (I / O) interfaces and communication components.
[0137] The processor is used to execute all or part of the steps in the rolling bearing condition assessment method based on contribution weights as described in the above embodiments. The memory is used to store various types of data, which may include, for example, instructions for any application or method in the electronic device, as well as application-related data.
[0138] The processor may be implemented as an Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), controller, microcontroller, microprocessor, or other electronic components, and is used to execute the rolling bearing state evaluation method based on contribution weight described in the above embodiments.
[0139] Example 4:
[0140] This embodiment proposes a computer-readable storage medium that stores executable instructions. When these instructions are executed, if they are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.
[0141] The computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the rolling bearing condition assessment method based on contribution weights described in the various embodiments of this application.
[0142] The aforementioned storage media include: flash memory, hard disks, multimedia cards, card-type memory (e.g., SD (Secure Digital Memory Card) or DX (Memory Data Register, MDR) memory), random access memory (RAM), static random-access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic storage, disks, optical discs, servers, APP (Application) app stores, and other media capable of storing program verification codes. These media store computer programs, which, when executed by a processor, can implement the various steps of the aforementioned rolling bearing state evaluation method based on contribution weights.
[0143] Example 5:
[0144] This embodiment proposes a computer program product, including a computer program or instructions, which, when executed by a processor, implements the aforementioned rolling bearing state evaluation method based on contribution weights.
[0145] Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a computer program product.
[0146] The various embodiments in this application are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
[0147] The scope of protection of this application is not limited to the embodiments described above. Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from the scope and spirit of this disclosure. If such modifications and variations fall within the scope of equivalent technology of this disclosure, then the intent of this disclosure also includes such modifications and variations.
Claims
1. A method for assessing the condition of rolling bearings based on contribution weights, characterized in that, include: Vibration signal data during bearing operation are collected as the bearing's full life cycle vibration signal, and noise reduction processing is performed on the bearing's full life cycle vibration signal. The square envelope spectrum transform is performed on the noise-reduced bearing full-life-cycle vibration signal to obtain the full-life-cycle square envelope spectrum amplitude. Based on the full lifecycle squared envelope spectrum amplitude, a convex optimization model based on the conditional constraint equation is constructed. The convex optimization model is solved to obtain the weight of the squared envelope spectrum amplitude at each running time. The root mean square value is used to quantify the weight of the squared envelope spectrum amplitude at each running time to obtain the health index; The health status of the bearing is assessed based on health indicators. If the bearing is in a fault state, a bi-exponential function with parameter constraints is used to fit and obtain the predicted remaining life of the bearing.
2. The rolling bearing condition assessment method based on contribution weights according to claim 1, characterized in that, The step of performing a square envelope spectrum transformation on the noise-reduced bearing full-life-cycle vibration signal to obtain the full-life-cycle square envelope spectrum amplitude includes: The noise-reduced bearing full-life-cycle vibration signal is subjected to Hilbert transform to obtain the analytic signal of the Hilbert transform; The analytic signal of the Hilbert transform is subjected to envelope squaring to obtain the squared envelope signal; Perform a Fourier transform on the square envelope signal to obtain the amplitude of the square envelope spectrum over the entire lifetime period.
3. The rolling bearing condition assessment method based on contribution weights according to claim 1, characterized in that, Based on the amplitude of the squared envelope spectrum over the entire life cycle, a convex optimization model based on the conditional constraint equation is constructed, and the calculation formula is as follows: ； Constraints: ； in, Let i = 1, 2, 3, ..., m, where i is the weight corresponding to the amplitude component of the i-th squared envelope spectrum. Let be the weight corresponding to the m-th squared envelope amplitude component. It is the set of weights for the amplitude components of the squared envelope spectrum. Let n be the L1 normalized norm, n be the number of periodic vibration signals throughout the bearing's lifespan, and m be the number of sampling points for the periodic vibration signals throughout the bearing's lifespan. The squared envelope spectrum value of the m-th sampling point of the n-th vibration signal data. It is a diagonal matrix. It is a vector distance used to separate healthy and faulty states. This is a data matrix representing the squared envelope spectrum amplitude over the entire lifespan.
4. The rolling bearing condition assessment method based on contribution weights according to claim 1, characterized in that, The method uses the root mean square value to quantify the weight of the squared envelope spectrum amplitude at each running time to obtain health indicators, including: The weights of the squared envelope spectrum amplitude at each running time are quantified, and the root mean square value of the weights of the squared envelope spectrum amplitude is calculated. All the calculated root mean square values are used as health indicators.
5. The rolling bearing condition assessment method based on contribution weight according to claim 4, characterized in that, The root mean square value of the weights used to calculate the amplitude of the squared envelope spectrum is used as a health indicator, and the calculation formula is as follows: ； ； in, As a health indicator, For all root mean square values, , i=1,2,3,…,n, are the root mean square values of the weights of the i-th squared envelope spectrum amplitude, and m is the number of sampling points of the bearing's full-life-cycle vibration signal.
6. The rolling bearing condition assessment method based on contribution weights according to claim 1, characterized in that, The method of obtaining the bearing remaining life prediction result by fitting a bi-exponential function with parameter constraints includes: Determine the parameters of the double exponential function; Based on health indicators, fit a parameterized biexponential function; Based on a parameterized double exponential function, a remaining life prediction curve is plotted with the number of sample files on the horizontal axis and health indicators on the vertical axis. The remaining life prediction result of the bearing is obtained from the remaining life prediction curve.
7. A life prediction system based on contribution weights, implemented using the rolling bearing condition assessment method based on contribution weights as described in any one of claims 1 to 6, characterized in that, include: The noise reduction module is used to collect vibration signal data during the operation of the bearing as the vibration signal of the bearing throughout its entire life cycle, and to perform noise reduction processing on the vibration signal of the bearing throughout its entire life cycle. The envelope spectrum transformation module is used to perform square envelope spectrum transformation on the noise-reduced bearing full-life-cycle vibration signal to obtain the full-life-cycle square envelope spectrum amplitude. The weight acquisition module is used to construct a convex optimization model based on the conditional constraint equations according to the full life cycle squared envelope spectrum amplitude, solve the convex optimization model, and obtain the weight of the squared envelope spectrum amplitude at each running time. The indicator determination module is used to quantify the weight of the squared envelope spectrum amplitude at each running time using the root mean square value to obtain the health indicator. The predicted structure output module is used to evaluate the health status of the bearing based on health indicators. If the bearing health status is a fault state, the remaining life prediction result of the bearing is obtained by fitting a bi-exponential function with parameter constraints.
8. An electronic device, characterized in that, include: One or more processors, and a memory for storing instructions that, when executed by the one or more processors, cause the one or more processors to perform the rolling bearing condition assessment method based on contribution weights as described in any one of claims 1 to 6.
9. A computer-readable storage medium, characterized in that, It stores executable instructions that, when executed, cause the processor to perform the rolling bearing condition assessment method based on contribution weights as described in any one of claims 1 to 6.
10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by the processor, they implement the rolling bearing condition assessment method based on contribution weight as described in any one of claims 1 to 6.