Rolling bearing residual life prediction method, device, equipment, medium and product
By combining EMD, CNN and LSTM methods, the vibration signals of rolling bearings are extracted and processed, and the problem of low prediction accuracy of remaining life of rolling bearings in the prior art is solved, and higher prediction accuracy and robustness are achieved.
Patent Information
- Application Number
- CN202411910068.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-24
- Publication Date
- 2025-04-18
AI Technical Summary
The existing method of rolling bearing residual life prediction is insufficient in accuracy, especially when dealing with long time series, it is difficult to capture long-term dependencies and is prone to gradient vanishing problems.
Combining empirical modal decomposition (EMD), convolutional neural network (CNN) and long and short-term memory network (LSTM), the vibration signal is first decomposed through EMD to extract the IMF signal, and then the abstract features in the IMF signal are extracted using CNN, and the time dependence and sequence mode are used to capture the time dependence and sequence mode, and the remaining life of the rolling bearing is finally predicted.
EMD decomposition removes redundant information, so that CNN can extract physical features more accurately. LSTM solves the shortcomings of CNN in long-term dependency processing, effectively alleviating the problem of gradient vanishing, thereby improving the accuracy of rolling bearing residual life prediction.
Smart Images
Figure CN119845583A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of bearing remaining life prediction, and in particular to a rolling bearing remaining life prediction method, device, equipment, medium and product. Background Art
[0002] Bearings are indispensable parts in mechanical equipment and are widely used in rotating machinery. Unexpected failure of bearings may lead to huge economic costs and catastrophic consequences. Research by Rai and Upadhyay shows that about 50% of motor failures are caused by rolling bearing failures. Accurate prediction of the remaining useful life (RUL) of bearings is of great reference value for the safe operation and maintenance decision-making of equipment systems. Therefore, predicting the RUL of bearings is of great significance. Taking effective remaining life prediction can timely detect bearing damage, avoid bearing failure, and minimize industrial production losses. In addition, RUL prediction is also the basis for establishing reasonable performance inspection and maintenance plans. Due to the huge potential of bearing RUL prediction, various prediction technologies have been developed.
[0003] Existing rolling bearing RUL prediction methods usually use convolutional neural networks (CNN) for RUL prediction. Although CNN networks have strong feature extraction capabilities, they are difficult to directly capture long-term dependencies when processing long time series due to the limitations of their convolution kernel size and receptive field. In addition, as the number of network layers increases, CNN may also face the problem of gradient disappearance, especially during the back-propagation process, the gradient may gradually decrease to zero, resulting in slow or even stagnant network parameter updates.
[0004] When the Long Short-Term Memory (LSTM) network is used for bearing RUL prediction, the statistical features extracted from the vibration signal cannot represent the non-stationary information well, because this transient information is amortized by processing the entire sequence. However, in the actual bearing degradation process, the non-stationary characteristics are of great significance for predicting the performance of the bearing. It can be seen that LSTM cannot provide valuable degradation information using the extracted statistical features with only time domain or frequency domain analysis.
[0005] In summary, the existing rolling bearing RUL prediction methods cannot accurately predict the remaining life of the bearing. Summary of the invention
[0006] The purpose of this application is to provide a method, device, equipment, medium and product for predicting the remaining life of a rolling bearing to solve the problem of low accuracy in predicting the remaining life of a bearing.
[0007] To achieve the above objectives, this application provides the following solutions:
[0008] In a first aspect, the present application provides a method for predicting the remaining life of a rolling bearing, comprising:
[0009] Perform empirical mode decomposition on the vibration signal of the rolling bearing to determine multiple IMF signals;
[0010] Using a CNN network to extract the abstract features hidden in the IMF signal;
[0011] Based on the loop feedback mechanism, gating mechanism and memory unit of the LSTM network, the LSTM network is used to capture the time dependency and sequence pattern in the abstract features;
[0012] forecasting time series data based on the time dependency and the sequence pattern;
[0013] The remaining life of the rolling bearing is predicted based on the predicted time series data.
[0014] In a second aspect, the present application provides a rolling bearing remaining life prediction device, comprising:
[0015] EMD decomposition module, used to perform empirical mode decomposition on the vibration signal of the rolling bearing and determine multiple IMF signals;
[0016] A CNN network is used to extract abstract features hidden in the IMF signal;
[0017] An LSTM network, for capturing temporal dependencies and sequence patterns in the abstract features based on a loop feedback mechanism, a gating mechanism, and a memory unit of the LSTM network;
[0018] A time series data prediction module, used for predicting time series data based on the time dependency and the sequence pattern;
[0019] The rolling bearing remaining life prediction module is used to predict the rolling bearing remaining life according to the prediction time series data.
[0020] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-described methods for predicting the remaining life of a rolling bearing.
[0021] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for predicting the remaining life of a rolling bearing.
[0022] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for predicting the remaining life of a rolling bearing.
[0023] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0024] This application combines empirical mode decomposition (EMD), CNN network and LSTM network to predict the remaining life of rolling bearings. First, EMD decomposition is performed before the vibration signal is input into CNN to discard redundant information, so that CNN can more accurately extract the actual physical information in the signal and more comprehensively extract degradation features. Then, LSTM is used for sequence modeling to capture the time dependency and sequence pattern in the abstract features, solve the defects of CNN in the long-term dependency problem, and then predict the time series data based on the time dependency and sequence pattern to accurately predict the remaining life of the bearing. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0026] Figure 1 A flow chart of a method for predicting the remaining life of a rolling bearing provided in one embodiment of the present application;
[0027] Figure 2 A schematic diagram of a CNN network provided in an embodiment of the present application;
[0028] Figure 3 A schematic diagram of an LSTM network provided in an embodiment of the present application;
[0029] Figure 4 A schematic diagram of the overall network structure provided by an embodiment of the present application;
[0030] Figure 5 A schematic diagram of a computer device provided in accordance with an embodiment of the present application. DETAILED DESCRIPTION
[0031] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0032] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0033] The present application embodiment provides a method for predicting the remaining life of a rolling bearing. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or can be executed by a terminal and a server together. In the present application embodiment, Figure 1 As shown, the method includes the following steps.
[0034] S1: Perform empirical mode decomposition on the vibration signal of the rolling bearing to determine multiple IMF signals.
[0035] S2: Use the CNN network to extract the abstract features hidden in the IMF signal.
[0036] S3: Based on the loop feedback mechanism, gating mechanism and memory unit of the LSTM network, the LSTM network is used to capture the time dependency and sequence pattern in the abstract features.
[0037] S4: Predicting time series data based on the time dependency and the sequence pattern.
[0038] S5: Predict the remaining life of the rolling bearing according to the prediction time series data.
[0039] In an exemplary embodiment, S1 may be replaced by the following steps.
[0040] The vibration signal is subjected to empirical mode decomposition and decomposed into multiple intrinsic mode functions (IMFs) and a residual term. For example, for 2560 vibration data at each time point, the original data mode is (2560, 1). After EMD processing, the data mode is (2560, m), where m is the number of IMF components decomposed from a set of original vibration signals.
[0041] In an exemplary embodiment, S2 may be replaced by the following steps.
[0042] Based on the convolution layer and the pooling layer in the CNN network, the IMF signal is subjected to alternating convolution and pooling operations, the IMF signal is peeled off layer by layer, and the abstract features hidden in the IMF signal are mined.
[0043] CNN network has a deep structure, generally composed of input layer, convolution layer, pooling layer, fully connected layer and output layer, such as Figure 2 As shown, the core modules of the CNN network are the convolution layer and the pooling layer. This application adopts alternating convolution and pooling operations to complete the layer-by-layer peeling of data and mine the abstract features hidden in the data.
[0044] In an exemplary embodiment, the LSTM network specifically includes multiple memory units; the memory unit includes a forget gate, an input gate and an output gate; the forget gate is used to determine the information to be discarded in the abstract feature; the input gate is used to determine the information to be input in the abstract feature; based on the forget gate and the input gate, the output gate is used to determine the information to be output in the abstract feature; the information to be output includes time dependency and sequence pattern.
[0045] The topological structure of LSTM retains the loop feedback mechanism and controls the speed of information accumulation by adding the mechanism of gate units, including selectively remembering new information and forgetting previous information. Therefore, the problem of long-term dependency in the sequence is overcome. At the same time, the memory unit of LSTM can maintain information for a longer time span, and during the back-propagation process, the gradient can be propagated more stably through the gating mechanism and memory unit, thereby avoiding the gradient disappearance to a certain extent.
[0046] LSTM networks usually involve multiple memory cells, each of which contains three gates, namely the forget gate, input gate, and output gate. The forget gate tells us which information should be forgotten, the input gate determines which inputs need to be remembered, and the output gate determines which information needs to be output. The specific calculation process of these three gates is as follows: Figure 3 shown.
[0047] The LSTM network further processes the features extracted by the convolutional and pooling layers to capture the temporal dynamics and sequential patterns in the data. This helps the network better understand and predict time series data.
[0048] In an exemplary embodiment, using the forget gate to determine the information to be discarded in the abstract feature specifically includes:
[0049] At the forget gate, using the formula Determine the information to be discarded in the abstract feature; where f t is the information to be discarded; σ is the sigmoid activation function, which compresses the value to between 0 and 1; xt is the data at time t; h t-1 is the output at time t-1; W f and R f For and f t The associated weight matrix; b f For and f t The associated deviation vector; “·” is the dot product operation.
[0050] In an exemplary embodiment, using the input gate to determine the information to be input in the abstract feature specifically includes:
[0051] At the input gate, using the formula Identify the information to be entered t , and use the formula
[0052] Determine candidate values for memory cell state values in, is the tanh activation function, which compresses the value between -1 and 1; W i and R i For i t The associated weight matrix; b i For i t The associated deviation vector; W c and R c For The associated weight matrix; b c For The associated bias vector.
[0053] In an exemplary embodiment, based on the forget gate and the input gate, the output gate is used to determine the information to be output in the abstract feature, specifically including:
[0054] At the output gate, using the formula Determine the state value C of the memory unit at time t t ; Where “×” represents the element-by-element multiplication of vectors;
[0055] Based on the state value C t , the input x of the memory unit t and the output h of the previous memory unit t-1 , using the formula Determine the information to be output in the abstract feature; wherein, t is the intermediate state, that is, h t-1 and x t The value after σ processing, * is the symbol for vector multiplication; W o and R o For Ot The associated weight matrix; b o For O t The associated bias vector.
[0056] The overall network structure is as follows Figure 4 As shown, after the convolution layer, pooling layer, and LSTM network processing and analysis, the application learns the relationship between a vibration signal and the remaining life, and outputs the prediction result, i.e., the remaining life of the rolling bearing, through the fully connected layer.
[0057] This application uses the PHM2012 bearing dataset for experimental verification and compares it with the deep separable convolution neural network (DSCN), multiscale dilated convolutional neural network (MDCN), convolutional long short-term memory (CLSTM) and deep convolution neural network (DCNN). The results show that this method has obvious advantages in prediction accuracy and generalization ability.
[0058] This application uses EMD to decompose nonlinear and non-stationary vibration signals, extracts IMF signals that are more representative of the actual physical information during the operation of the bearing, uses the CNN network to automatically extract feature information layer by layer, and combines the LSTM network for modeling and prediction. In this process, the vibration signal processed by EMD eliminates redundant information, provides CNN with more accurate bearing physical information, and can make CNN more accurate in extracting features and improve the prediction accuracy of the network. At the same time, the use of LSTM solves the difficulties that CNN has in dealing with long-term dependencies, and effectively alleviates the problem of gradient disappearance that CNN is prone to when dealing with long-term series problems. Feature extraction through CNN can also solve the problem of low computational efficiency of LSTM when processing large-scale data, and the statistical features extracted from the vibration signal cannot represent non-stationary information well. The organic combination of the three methods can improve the effectiveness and efficiency of degradation trend feature extraction and improve the accuracy and robustness of bearing RUL prediction.
[0059] Based on the same inventive concept, the embodiment of the present application also provides a rolling bearing remaining life prediction device for implementing the above-mentioned rolling bearing remaining life prediction method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in one or more embodiments of the rolling bearing remaining life prediction device provided below can refer to the limitations of the rolling bearing remaining life prediction method above, and will not be repeated here.
[0060] In an exemplary embodiment, a rolling bearing remaining life prediction device is provided, comprising:
[0061] The EMD decomposition module is used to perform empirical mode decomposition on the vibration signal of the rolling bearing and determine multiple IMF signals.
[0062] The CNN network is used to extract the abstract features hidden in the IMF signal.
[0063] The LSTM network is used to capture the temporal dependency and sequence pattern in the abstract features based on the LSTM network's loop feedback mechanism, gating mechanism, and memory unit.
[0064] The time series data prediction module is used to predict the time series data based on the time dependency and the sequence pattern.
[0065] The rolling bearing remaining life prediction module is used to predict the rolling bearing remaining life according to the prediction time series data.
[0066] In an exemplary embodiment, a computer device is provided, such as Figure 5 As shown, the computer device can be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store rolling bearing remaining life prediction data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for predicting the remaining life of a rolling bearing is implemented.
[0067] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the above method is implemented when the processor executes the computer program.
[0068] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the above method when executed by a processor.
[0069] In an exemplary embodiment, a computer program product is provided, including a computer program, which implements the above method when executed by a processor.
[0070] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ReadOnlyMemory, ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (Magnetoresistive RandomAccess Memory, MRAM), ferroelectric random access memory (Ferroelectric RandomAccess Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory can include random access memory (RandomAccess Memory, RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0071] In this application, all actions to obtain signals, information or data are carried out in compliance with the relevant data protection laws and policies of the country where they are located and with the authorization given by the owner of the corresponding device.
[0072] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.
[0073] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0074] The present application uses specific examples to illustrate the principles and implementation methods of the present application. The above examples are only used to help understand the method and core ideas of the present application. At the same time, for those skilled in the art, according to the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present application.
Claims
1. A method for predicting the remaining life of a rolling bearing, characterized in that: The method for predicting the remaining life of a rolling bearing comprises: Perform empirical mode decomposition on the vibration signal of the rolling bearing to determine multiple IMF signals; Using a CNN network to extract the abstract features hidden in the IMF signal; Based on the loop feedback mechanism, gating mechanism and memory unit of the LSTM network, the LSTM network is used to capture the time dependency and sequence pattern in the abstract features; forecasting time series data based on the time dependency and the sequence pattern; The remaining life of the rolling bearing is predicted based on the predicted time series data.
2. The method for predicting the remaining life of a rolling bearing according to claim 1, characterized in that: The CNN network is used to extract the abstract features hidden in the IMF signal, including: Based on the convolution layer and the pooling layer in the CNN network, the IMF signal is subjected to alternating convolution and pooling operations, the IMF signal is peeled off layer by layer, and the abstract features hidden in the IMF signal are mined.
3. The method for predicting the remaining life of a rolling bearing according to claim 1, characterized in that: The LSTM network specifically includes a plurality of memory units; the memory unit includes a forget gate, an input gate and an output gate; Using the forget gate to determine the information to be discarded in the abstract feature; Determining the information to be input in the abstract feature by using the input gate; Based on the forget gate and the input gate, the output gate is used to determine the information to be output in the abstract feature; the information to be output includes time dependency and sequence pattern.
4. The method for predicting the remaining life of a rolling bearing according to claim 3, characterized in that: The forget gate is used to determine the information to be discarded in the abstract feature, specifically including: At the forget gate, using the formula Determine the information to be discarded in the abstract feature; where f t is the information to be discarded; σ is the Sigmoid activation function; x t is the abstract feature at time t; h t-1 is the output at time t-1; W f and R f For and f t The associated weight matrix; b f For and f t The associated deviation vector; "·" is the dot product operation.
5. The method for predicting the remaining life of a rolling bearing according to claim 4, characterized in that: The input gate is used to determine the information to be input in the abstract feature, specifically including: At the input gate, using the formula Identify the information to be entered t , and using the formula Determine candidate values for memory cell state values in, is the Tanh function; W i and R i For i t The associated weight matrix; b i For i t The associated deviation vector; W c and R c For The associated weight matrix; b c For The associated bias vector.
6. The method for predicting the remaining life of a rolling bearing according to claim 5, characterized in that: Based on the forget gate and the input gate, the output gate is used to determine the information to be output in the abstract feature, specifically including: At the output gate, using the formula Determine the state value C of the memory unit at time t t ; Where "×" represents the element-by-element multiplication of vectors; Based on the state value C t , the input x of the memory unit t and the output h of the previous memory unit t-1 , using the formula Determine the information to be output in the abstract feature; wherein, t is the intermediate state, * is the symbol for vector multiplication; W o and R o For O t The associated weight matrix; b o For O t The associated bias vector.
7. A rolling bearing remaining life prediction device, characterized in that: The rolling bearing remaining life prediction device comprises: EMD decomposition module, used to perform empirical mode decomposition on the vibration signal of the rolling bearing and determine multiple IMF signals; A CNN network is used to extract abstract features hidden in the IMF signal; An LSTM network, for capturing temporal dependencies and sequence patterns in the abstract features based on a loop feedback mechanism, a gating mechanism, and a memory unit of the LSTM network; A time series data prediction module, used for predicting time series data based on the time dependency and the sequence pattern; The rolling bearing remaining life prediction module is used to predict the rolling bearing remaining life according to the prediction time series data.
8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for predicting the remaining life of a rolling bearing according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting the remaining life of a rolling bearing according to any one of claims 1 to 6 is implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for predicting the remaining life of a rolling bearing according to any one of claims 1 to 6 is implemented.