A method, device and equipment for monitoring the health of bearings under time-varying working conditions

Abstract

The present invention discloses a method, device and equipment for monitoring the operation health of bearings under time-varying working conditions. The root mean square value of the vibration signal at the current moment and the speed at the current moment are input into the bearing hybrid system response model. The root mean square value of the vibration signal at the next moment and the neural network parameters at the next moment are calculated according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm; the Euclidean distance between the neural network parameters at the current moment and the neural network parameters at the next moment is calculated, and the fault detection of the bearing to be monitored is performed in combination with the abnormal value detection criterion. The neural network parameters at each moment are subjected to principal component analysis according to the feature dimensionality reduction method, and the first-order principal component is taken as the real-time degradation process after the failure of the bearing to be monitored, so as to realize the operation health monitoring of the bearing. The present invention can realize the monitoring of the operation health status of the bearing under time-varying working conditions.

Description

Technical Field

[0001] The present invention belongs to the technical field of mechanical equipment status monitoring, and specifically relates to a bearing operation health monitoring method, device and equipment under time-varying working conditions. Background Art

[0002] Industrial equipment is developing towards intelligence. Real-time monitoring of the equipment's operating status and obtaining the equipment's current health status are important ways to ensure the equipment's continued healthy operation. As a key rotating component widely present in mechanical equipment, the health status of the bearing affects the operating status of the entire equipment. Therefore, real-time monitoring of the operation and health status of the bearings in the equipment and obtaining their potential health degradation information are important prerequisites for ensuring that the equipment does not experience unexpected downtime and implementing predictive health maintenance strategies.

[0003] At present, the monitoring of bearing health is mainly achieved by analyzing its vibration signal. Fault detection can be achieved by extracting fault characteristic frequency based on prior fault structure information, or by directly extracting statistical indicators in the time domain, frequency domain, and time-frequency domain to detect faults and track degradation trends. Moreover, most of the research focuses on bearing health monitoring under stable working conditions, which results in the research background not being able to effectively cover the operating conditions of equipment such as wind turbines and robotic arms. At the same time, due to the adjustment of production plans, the working conditions of some equipment may also change within a period of time. However, there is currently no good method to characterize the bearing health status under time-varying working conditions for real-time health status monitoring. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a method, device and equipment for monitoring the operating health of a bearing under time-varying working conditions, which can monitor the operating health status of a bearing under time-varying working conditions.

[0005] In order to solve the above technical problems, the present invention is implemented by the following technical solutions:

[0006] A method for monitoring the operation health of a bearing under time-varying working conditions, comprising:

[0007] Obtaining the root mean square value of the vibration signal of the bearing to be monitored at the current moment and the rotation speed of the bearing to be monitored at the current moment;

[0008] Initializing the neural network parameters of the current moment in the pre-built bearing hybrid system response model, and defining the process noise variance and the observation noise variance in the bearing hybrid system response model;

[0009] Inputting the root mean square value of the vibration signal at the current moment and the rotation speed at the current moment into the bearing hybrid system response model, and calculating the root mean square value of the vibration signal at the next moment and the neural network parameters at the next moment according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm;

[0010] The Euclidean distance between the neural network parameters at the current moment and the neural network parameters at the next moment is calculated, and fault detection is performed on the bearing to be monitored in combination with the outlier detection criterion. The neural network parameters at each moment are subjected to principal component analysis according to the feature dimensionality reduction method, and the first-order principal component is taken as the real-time degradation process characterizing the bearing to be monitored after a fault occurs, so as to realize bearing operation health monitoring.

[0011] Furthermore, the bearing hybrid system response model is:

[0012]

[0013] In the formula, and are the weights and biases of the neural network in the initial layer and hidden layer respectively; X t-1 and X t is the root mean square value of the vibration signal at time t-1 and time t; C t is the speed at time t; is the output of the initial layer of the neural network; f (0) and f (1) is the activation function of the initial layer and hidden layer of the neural network; ω t and ν t is the process noise variance and the observation noise variance; is the observation function in the observation equation; is the observed output of the bearing hybrid system response model.

[0014] Further, the root mean square value of the vibration signal at the current moment and the speed at the current moment are input into the bearing hybrid system response model, and the root mean square value of the vibration signal at the next moment is calculated according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm, including:

[0015] Predict the RMS value of the vibration signal at the next moment and its corresponding covariance matrix. The prediction formula is as follows:

[0016]

[0017]

[0018]

[0019] In the formula, F is the mapping relationship of the neural network; is the Jacobian matrix of the neural network mapping function relative to the root mean square value X of the vibration signal; P X，t-1 is the covariance matrix of X at time t-1; is the covariance matrix of the predicted root mean square value of the vibration signal at time t; Q t is the covariance of the transmission process noise; θ is the set of neural network parameters W and B; T is the transposition operation;

[0020] The root mean square value of the vibration signal at the next moment and its corresponding covariance matrix are corrected and updated to obtain the root mean square value of the vibration signal at the next moment. The correction and update formula is as follows:

[0021]

[0022]

[0023]

[0024] In the formula, K X，t is the gain matrix of the root mean square value X of the vibration signal at time t, and the observation function is set is a linear function, that is, H is the observation matrix in the observation equation; R t is the observation noise covariance; is the root mean square value of the vibration signal updated at time t; is the root mean square value of the vibration signal updated at time t The covariance of .

[0025] Furthermore, the root mean square value of the vibration signal at the current moment and the speed at the current moment are input into the bearing hybrid system response model, and the neural network parameters at the next moment are calculated according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm, including:

[0026] Predict the neural network parameters and their corresponding covariance matrix at the next moment. The prediction formula is as follows:

[0027]

[0028]

[0029] In the formula, is the neural network parameter predicted at time t; is the covariance of the neural network parameter θ at the predicted time t-1; is the covariance of the neural network parameter θ at the next moment t predicted; Q t is the covariance of the transfer process noise at time t;

[0030] The neural network parameters at the next moment and their corresponding covariance matrix are corrected and updated to obtain the neural network parameters at the next moment. The corrected update formula is as follows:

[0031]

[0032]

[0033]

[0034]

[0035] In the formula, K θ，t is the Kalman gain matrix of the neural network parameter θ at time t; is the Jacobian matrix of the neural network's mapping function relative to the neural network parameter θ; R t is the cooperative defense difference of the observation noise of the observation equation; is the parameter of the neural network at time t after the update; Y t is the root mean square value of the vibration signal at the actual time t; is the covariance of the neural network parameters θ at the updated time t-1.

[0036] Further, the obtaining of the root mean square value of the vibration signal of the bearing to be monitored at the current moment and the rotation speed of the bearing to be monitored at the current moment includes:

[0037] Obtain the real-time vibration signal and real-time key phase signal of the bearing to be monitored;

[0038] The root mean square value of the vibration signal of the bearing to be monitored at the current moment is calculated according to the real-time vibration signal, and the rotational speed of the bearing to be monitored at the current moment is calculated according to the real-time key phase signal.

[0039] Furthermore, an acceleration sensor is used to obtain the real-time vibration signal, and a displacement sensor is used to obtain the real-time key phase signal.

[0040] Furthermore, the outlier detection criterion is the 3sigma criterion, and the feature dimension reduction method is the principal component analysis method.

[0041] A bearing operation health monitoring device under time-varying working conditions, comprising:

[0042] An acquisition module is used to acquire the root mean square value of the vibration signal of the bearing to be monitored at the current moment and the rotation speed of the bearing to be monitored at the current moment;

[0043] A parameter setting module, used to initialize the neural network parameters of the pre-built bearing hybrid system response model at the current moment, and define the process noise variance and observation noise variance in the bearing hybrid system response model;

[0044] A calculation module, used for inputting the root mean square value of the vibration signal at the current moment and the rotation speed at the current moment into the bearing hybrid system response model, and calculating the root mean square value of the vibration signal at the next moment and the neural network parameters at the next moment according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm;

[0045] The monitoring module is used to calculate the Euclidean distance between the neural network parameters at the current moment and the neural network parameters at the next moment, and perform fault detection on the bearing to be monitored in combination with the outlier detection criterion, perform principal component analysis on the neural network parameters at each moment according to the feature dimensionality reduction method, and take the first-order principal component as the real-time degradation process of the bearing to be monitored after a fault occurs, so as to realize bearing operation health monitoring.

[0046] A device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of a method for monitoring the operating health of a bearing under time-varying working conditions when executing the computer program.

[0047] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a bearing operation health monitoring method under time-varying working conditions are implemented.

[0048] Compared with the prior art, the present invention has at least the following beneficial effects:

[0049] Most of the current bearing health status methods focus on the situation under stable working conditions. The general bearing health monitoring methods under stable working conditions cannot effectively realize the bearing health status monitoring under time-varying working conditions. Even though some dimensionless indicators such as kurtosis can realize fault detection under time-varying working conditions, they cannot realize real-time tracking of fault evolution after the fault occurs, and cannot effectively characterize the health degradation process. The present invention provides a bearing operation health monitoring method under time-varying working conditions, which effectively solves the problem of monitoring the bearing health status under time-varying working conditions. By combining neural networks and state space models, a new bearing hybrid system response model is proposed to describe the operating state of the bearing. The health state of the bearing is described by the neural network of the bearing hybrid system response model. By introducing the dual extended Kalman filter algorithm, the neural network parameter estimation problem in the bearing hybrid system response model is effectively solved. Experiments have proved the effectiveness of the monitoring method of the present invention, as well as its superiority over general dimensionless statistical parameters in eliminating working condition interference and degradation monotonicity evaluation indicators.

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the specific implementation modes of the present invention, the drawings required for use in the description of the specific implementation modes will be briefly introduced below. Obviously, the drawings described below are some implementation modes of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 It is a flowchart of a method for monitoring the health of a bearing under time-varying working conditions according to the present invention;

[0053] Figure 2 This is a structural diagram of the bearing hybrid system response model proposed by the present invention;

[0054] Figure 3 The bearing condition monitoring strategy under time-varying working conditions proposed by the present invention;

[0055] Figure 4 It is the trend diagram of the system status change of the bearing during the whole life stage;

[0056] Figure 5 It is the trend diagram of the bearing's working condition changes during its entire life cycle;

[0057] Figure 6 The early fault detection results of the bearing monitored by the bearing operation health monitoring method of the present invention are as follows:

[0058] Figure 7 It is the characterization result of the bearing degradation stage monitored by the bearing operation health monitoring method of the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] As a specific embodiment of the present invention, Figure 1 As shown, a bearing operation health monitoring method under time-varying working conditions specifically includes the following steps:

[0061] Step 1: Obtain the root mean square value of the vibration signal of the bearing to be monitored at the current moment and the rotation speed of the bearing to be monitored at the current moment.

[0062] In this implementation, the root mean square value of the vibration signal of the bearing to be monitored at the current moment and the rotation speed of the bearing to be monitored at the current moment are obtained, as follows:

[0063] Step 1.1, obtain the real-time vibration signal and real-time key phase signal of the bearing to be monitored.

[0064] Preferably, an acceleration sensor is used to obtain a real-time vibration signal, and a displacement sensor is used to obtain a real-time key phase signal.

[0065] Step 1.2: Calculate the root mean square value of the vibration signal of the bearing to be monitored at the current moment according to the real-time vibration signal, and calculate the rotation speed of the bearing to be monitored at the current moment according to the real-time key phase signal.

[0066] Step 2: Initialize the neural network parameters of the pre-built bearing hybrid system response model at the current moment, and define the process noise variance and observation noise variance in the bearing hybrid system response model.

[0067] The bearing hybrid system response model pre-constructed in the present invention is:

[0068]

[0069] In the formula, and are the weights and biases of the neural network in the initial layer and hidden layer respectively; X t-1 and X t is the root mean square value of the vibration signal at time t-1 and time t; C t is the speed at time t; is the output of the initial layer of the neural network; f (0) and f (1) is the activation function of the initial layer and hidden layer of the neural network; ω t and ν t is the process noise variance and the observation noise variance; is the observation function in the observation equation; is the observed output of the bearing hybrid system response model.

[0070] Step 3: Input the root mean square value of the vibration signal at the current moment and the speed at the current moment into the bearing hybrid system response model, and calculate the root mean square value of the vibration signal at the next moment and the neural network parameters at the next moment according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm.

[0071] In this implementation, the root mean square value of the vibration signal at the current moment and the speed at the current moment are input into the bearing hybrid system response model, and the root mean square value of the vibration signal at the next moment is calculated according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm, as follows:

[0072] First, predict the RMS value of the vibration signal at the next moment and its corresponding covariance matrix. The prediction formula is as follows:

[0073]

[0074]

[0075]

[0076] In the formula, F is the mapping relationship of the neural network; is the Jacobian matrix of the neural network mapping function relative to the root mean square value X of the vibration signal; P X，t-1 is the covariance matrix of X at time t-1; is the covariance matrix of the predicted root mean square value of the vibration signal at time t; Q t is the covariance of the transmission process noise; θ is the set of neural network parameters W and B; T is the transposition operation;

[0077] Finally, the root mean square value of the vibration signal at the next moment and its corresponding covariance matrix are corrected and updated to obtain the root mean square value of the vibration signal at the next moment. The correction and update formula is as follows:

[0078]

[0079]

[0080]

[0081] In the formula, K X，t is the gain matrix of the root mean square value X of the vibration signal at time t, and the observation function is set is a linear function, that is, H is the observation matrix in the observation equation; R t is the observation noise covariance; is the root mean square value of the vibration signal updated at time t; is the root mean square value of the vibration signal updated at time t The covariance of .

[0082] In this implementation, the root mean square value of the vibration signal at the current moment and the speed at the current moment are input into the bearing hybrid system response model, and the neural network parameters at the next moment are calculated according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm, as follows:

[0083] First, predict the neural network parameters and their corresponding covariance matrix at the next moment. The prediction formula is as follows:

[0084]

[0085]

[0086] In the formula, is the neural network parameter predicted at time t; is the covariance of the neural network parameter θ at the predicted time t-1; is the covariance of the neural network parameter θ at the next moment t predicted; Q t is the covariance of the transfer process noise at time t;

[0087] Finally, the neural network parameters and their corresponding covariance matrix at the next moment are corrected and updated to obtain the neural network parameters at the next moment. The corrected update formula is as follows:

[0088]

[0089]

[0090]

[0091]

[0092] In the formula, K θ，t is the Kalman gain matrix of the neural network parameter θ at time t; is the Jacobian matrix of the neural network's mapping function relative to the neural network parameter θ; R t is the cooperative defense difference of the observation noise of the observation equation; is the parameter of the neural network at time t after the update; Y tis the root mean square value of the actual vibration signal at time t collected and calculated by the sensor; is the covariance of the neural network parameters θ at the updated time t-1.

[0093] Step 4: Calculate the Euclidean distance between the neural network parameters at the current moment and the neural network parameters at the next moment to form a time difference sequence of the neural network parameters, and perform fault detection on the monitored bearing in combination with the outlier detection criterion. Perform principal component analysis on the neural network parameters at each moment according to the feature dimensionality reduction method, and take the first-order principal component as the representation of the real-time degradation process of the monitored bearing after a fault occurs, so as to realize the bearing operation health monitoring.

[0094] Preferably, the outlier detection criterion is the 3sigma criterion, and the feature dimension reduction method is the principal component analysis method.

[0095] The following is a detailed explanation of the method for constructing the bearing hybrid system response model provided by the present invention.

[0096] Build a hybrid system response model to characterize the operation of the bearing under time-varying conditions:

[0097] Specifically, the hybrid system response model is used to model the input real-time speed C t and the measured real-time observed RMS Y t :

[0098]

[0099] in, is the system response function of the system at time t. After considering the measurement system error, introducing state transfer and system observation noise, the hybrid system response model is rewritten as:

[0100]

[0101]

[0102] in, and are the state transfer function and observation function respectively; ω t With ν t is the variance of process noise and observation noise; X t is the real state of the system (represented by the RMS value of the vibration signal).

[0103] Because of the function The bearing system state transfer process represented is a complex nonlinear process. In order to better simulate the process, the present invention introduces a neural network to replace the explicit transfer function expression to achieve the modeling of the nonlinear transfer process. Specifically, after the introduction of the neural network, the system transfer equation represented by formula (2) can be rewritten as shown in formulas (4) and (5), where W and B are the weights and biases of the neurons in the neural network, respectively; f (0) and f (1) is the activation function of the initial layer and hidden layer of the neural network.

[0104] X t =F(C t , X t-1 , W t-1 , B t-1 )+ω t (4)

[0105]

[0106] In this way, the nonlinear transfer function of the unknown bearing system is effectively replaced by embedding the neural network in the system transfer equation. Combined with the observation equation, the final bearing hybrid system response model is shown in formula (6):

[0107]

[0108] The following is a detailed description of the invention using a set of accelerated bearing fatigue life tests as a validation of the invention. Figure 2 and Figure 3 As shown, the specific implementation steps are as follows:

[0109] Step 1: Obtain the vibration signal of the experimental bearing seat through the acceleration sensor, and the vibration signal sampling frequency is 20480Hz. The coupling key phase signal is collected through the displacement sensor to calculate the key phase signal length of 3s. The rotation speed of the bearing changes from 600r / min to 3000r / min and then to 600r / min, changing by 600r / min every 240 minutes, and each speed remains unchanged for 30 minutes.

[0110] Step 2: Calculate the root mean square value Y of the vibration signal of the bearing to be monitored at the current moment based on the real-time vibration signal t ,like Figure 4 As shown, the speed C of the bearing to be monitored at the current moment is calculated based on the key phase signal. t .

[0111] Step 3: Normalize the speed signal: like Figure 5 shown.

[0112] Step 4: Set the process noise variance Q and observation noise variance R in the bearing hybrid system response model to 0.2 and 0.0025 respectively.

[0113] Step 5: Initialize the neural network parameters at the current moment in the pre-built bearing hybrid system response model and the parameters updated by the model estimation:

[0114] Step 6: Set the neural network structure parameters to 2-10-1, which means two neurons in the input layer, 10 neurons in the hidden layer, and one neuron in the output layer. The activation function is as shown below:

[0115]

[0116] Step 7: Estimate the actual system state value and neural network parameters in real time.

[0117] Step 7.1: Predict the parameters of the neural network at the next time t:

[0118]

[0119]

[0120] Step 7.2: Calculate the Jacobian matrix

[0121]

[0122] Step 7.3: Predict the RMS value of the vibration signal of the system at the next moment

[0123]

[0124]

[0125] Step 7.4: Update the RMS value of the vibration signal at time t

[0126]

[0127]

[0128]

[0129] Step 7.5: Calculate the Jacobian matrix

[0130]

[0131] Step 7.6: Update the values ​​of the neural network parameters at time t

[0132]

[0133]

[0134]

[0135] Step 8: Application The differential sequence combined with 3sigma can accurately detect the early faults of bearings, such as Figure 6 shown.

[0136] Step 8.1: Use adjacent moments The Euclidean distance calculation difference sequence d t :

[0137] d t =||(θ t -θ t-1 )|| 2

[0138] Step 8.2: Use the 3 sigma criterion to detect early bearing failures.

[0139] Specifically, the first 200 data points are selected as health data to calculate their mean and standard deviation:

[0140]

[0141]

[0142] Determination: As follows:

[0143] If:|d t -μ|>3σ, then the bearing has an initial fault at the current time t.

[0144] It can be found that Figure 6 The moment shown is when initial failure of the bearing occurred.

[0145] Step 9: Apply principal component analysis to neural network parameters The first-order principal component is used to construct the bearing health index to characterize its health degradation trend. Figure 7 As shown, the degradation trend at the current moment is represented in real time. From the figure, it can be clearly seen that the health status of the bearing degrades over time after the initial failure occurs.

[0146]

[0147] This embodiment provides a bearing operation health monitoring device under time-varying working conditions, including:

[0148] The acquisition module is used to obtain the root mean square value of the vibration signal of the bearing to be monitored at the current moment and the rotation speed of the bearing to be monitored at the current moment.

[0149] Preferably, the acquisition module includes an acquisition unit and a calculation unit, wherein:

[0150] The acquisition unit is used to acquire the real-time vibration signal and real-time key phase signal of the bearing to be monitored.

[0151] The calculation unit is used to calculate the root mean square value of the vibration signal of the bearing to be monitored at the current moment according to the real-time vibration signal, and calculate the speed of the bearing to be monitored at the current moment according to the real-time key phase signal.

[0152] The parameter setting module is used to initialize the neural network parameters at the current moment in the pre-built bearing hybrid system response model, and define the process noise variance and observation noise variance in the bearing hybrid system response model.

[0153] The calculation module is used to input the root mean square value of the vibration signal and the speed at the current moment into the bearing hybrid system response model, and calculate the root mean square value of the vibration signal at the next moment and the neural network parameters at the next moment according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm.

[0154] The monitoring module is used to calculate the Euclidean distance between the neural network parameters at the current moment and the neural network parameters at the next moment, and to perform fault detection on the monitored bearing in combination with the outlier detection criterion. The principal component analysis of the neural network parameters at each moment is performed according to the feature dimensionality reduction method, and the first-order principal component is taken as the real-time degradation process of the monitored bearing after a fault occurs, so as to realize the bearing operation health monitoring.

[0155] In one embodiment of the present invention, a computer device is provided, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used to implement the operation of a bearing operation health monitoring method under time-varying working conditions.

[0156] In one embodiment of the present invention, a method for monitoring the health of a bearing under time-varying working conditions can be stored in a computer-readable storage medium if it is implemented in the form of a software functional unit and sold or used as an independent product. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. Computer-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data.

[0157] The computer storage medium can be any available medium or data storage device that can be accessed by a computer, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NANDFLASH), solid-state drive (SSD)), etc.

[0158] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0159] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products of the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0160] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0161] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0162] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for monitoring the operating health of a bearing under time-varying working conditions, characterized in that: include: Obtaining the root mean square value of the vibration signal of the bearing to be monitored at the current moment and the rotation speed of the bearing to be monitored at the current moment; Initializing the neural network parameters of the current moment in the pre-built bearing hybrid system response model, and defining the process noise variance and the observation noise variance in the bearing hybrid system response model; The bearing hybrid system response model is: In the formula, and are the weights and biases of the neural network in the initial layer and hidden layer respectively; X t-1 and X t is the root mean square value of the vibration signal at time t-1 and time t; C t is the speed at time t; is the output of the initial layer of the neural network; f (0) and f (1) is the activation function of the initial layer and hidden layer of the neural network; ω t and v t is the process noise variance and the observation noise variance; is the observation function in the observation equation; is the observed output of the bearing hybrid system response model; The root mean square value of the vibration signal at the current moment and the speed at the current moment are input into the bearing hybrid system response model, and the root mean square value of the vibration signal at the next moment and the neural network parameters at the next moment are calculated according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm, specifically: Predict the RMS value of the vibration signal at the next moment and its corresponding covariance matrix. The prediction formula is as follows: In the formula, F is the mapping relationship of the neural network; is the Jacobian matrix of the neural network mapping function relative to the root mean square value X of the vibration signal; P X，t-1 is the covariance matrix of X at time t-1; is the covariance matrix of the predicted root mean square value of the vibration signal at time t; Q t is the covariance of the transfer process noise; θ is the set of neural network parameters W and B; T is the transpose operation; The root mean square value of the vibration signal at the next moment and its corresponding covariance matrix are corrected and updated to obtain the root mean square value of the vibration signal at the next moment. The correction and update formula is as follows: In the formula, K X，t is the gain matrix of the root mean square value X of the vibration signal at time t, and the observation function is set is a linear function, that is, H is the observation matrix in the observation equation; R t is the observation noise covariance; is the root mean square value of the vibration signal updated at time t; is the root mean square value of the vibration signal updated at time t The covariance of Predict the neural network parameters and their corresponding covariance matrix at the next moment. The prediction formula is as follows: In the formula, is the neural network parameter predicted at time t; is the covariance of the neural network parameter θ at the predicted time t-1; is the covariance of the neural network parameter θ at the next moment t predicted; Q t is the covariance of the transfer process noise at time t; The neural network parameters at the next moment and their corresponding covariance matrix are corrected and updated to obtain the neural network parameters at the next moment. The corrected update formula is as follows: In the formula, K θ，t is the Kalman gain matrix of the neural network parameter θ at time t; is the Jacobian matrix of the neural network's mapping function relative to the neural network parameter θ; R t is the cooperative defense difference of the observation noise of the observation equation; is the parameter of the neural network at time t after the update; Y t is the root mean square value of the vibration signal at the actual time t; is the covariance of the neural network parameter θ at the updated time t-1; The Euclidean distance between the neural network parameters at the current moment and the neural network parameters at the next moment is calculated, and fault detection is performed on the bearing to be monitored in combination with the outlier detection criterion. The neural network parameters at each moment are subjected to principal component analysis according to the feature dimensionality reduction method, and the first-order principal component is taken as the real-time degradation process of the bearing to be monitored after a fault occurs.

2. The method for monitoring the bearing operation health under time-varying working conditions according to claim 1 is characterized in that: The step of obtaining the root mean square value of the vibration signal of the bearing to be monitored at the current moment and the rotation speed of the bearing to be monitored at the current moment includes: Obtain the real-time vibration signal and real-time key phase signal of the bearing to be monitored; The root mean square value of the vibration signal of the bearing to be monitored at the current moment is calculated according to the real-time vibration signal, and the rotational speed of the bearing to be monitored at the current moment is calculated according to the real-time key phase signal.

3. The method for monitoring the bearing operation health under time-varying working conditions according to claim 2 is characterized in that: An acceleration sensor is used to obtain the real-time vibration signal, and a displacement sensor is used to obtain the real-time key phase signal.

4. The method for monitoring the bearing operation health under time-varying working conditions according to claim 1, characterized in that: The outlier detection criterion is the 3sigma criterion, and the feature dimension reduction method is the principal component analysis method.

5. A bearing operation health monitoring device under time-varying working conditions, characterized in that: Used to implement a bearing operation health monitoring method under time-varying working conditions as described in any one of claims 1 to 4, the bearing operation health monitoring device under time-varying working conditions comprises: An acquisition module is used to acquire the root mean square value of the vibration signal of the bearing to be monitored at the current moment and the rotation speed of the bearing to be monitored at the current moment; A parameter setting module, used to initialize the neural network parameters of the pre-built bearing hybrid system response model at the current moment, and define the process noise variance and observation noise variance in the bearing hybrid system response model; A calculation module, used for inputting the root mean square value of the vibration signal at the current moment and the rotation speed at the current moment into the bearing hybrid system response model, and calculating the root mean square value of the vibration signal at the next moment and the neural network parameters at the next moment according to the process noise variance, the observation noise variance and the dual extended Kalman filter algorithm; The monitoring module is used to calculate the Euclidean distance between the neural network parameters at the current moment and the neural network parameters at the next moment, and perform fault detection on the bearing to be monitored in combination with the outlier detection criterion, perform principal component analysis on the neural network parameters at each moment according to the feature dimensionality reduction method, and take the first-order principal component as the real-time degradation process of the bearing to be monitored after a fault occurs, so as to realize bearing operation health monitoring.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for monitoring the operating health of a bearing under time-varying working conditions as described in any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a method for monitoring the operating health of a bearing under time-varying working conditions as described in any one of claims 1 to 4 are implemented.

Images