A device failure trend warning method and system based on MVMD and gradient judgment

Abstract

This invention provides a method and system for warning equipment failure trends based on MVMD and gradient judgment. This method employs a gradient judgment algorithm to segment operating conditions, performs median filtering for each operating condition to eliminate sudden jumps, and finally employs a sliding average algorithm to track trends and issue warnings, achieving accurate warnings of equipment failures. To address false alarms and missed alarms in equipment warnings, this invention employs an MVMD decomposition algorithm to continuously extract all IMFs. This method eliminates the need to know the number of modes, reduces computational complexity, effectively removes noise interference from the signal, enables timely changes in the warning line, and reduces false alarms and missed alarms.

Description

Technical Field

[0001] The present invention belongs to the technical field of equipment trend warning, and in particular relates to an equipment failure trend warning method and system based on MVMD and gradient judgment. Background Art

[0002] Failure to detect faults in a timely manner during equipment operation often leads to equipment damage, resulting in significant economic losses. Existing vibration monitoring technology relies on vibration characteristic values ​​for fault warning. However, when operating conditions change, the warning line does not change in a timely manner, resulting in false alarms and missed alarms.

[0003] Existing vibration monitoring technologies, such as variational modal decomposition (MVMD), suffer from performance degradation due to an inaccurate number of decomposed modes. Multivariate variational modal decomposition (MVMD) continuously extracts all intrinsic mode function (IMF) components. Compared to VMD, this method does not require knowledge of the number of modes and has lower computational complexity, effectively removing noise from the signal. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for early warning of equipment failure trends based on MVMD and gradient judgment, which are used to accurately warn of equipment failures.

[0005] The technical solution adopted by the present invention to solve the above technical problems is: a device failure trend early warning method based on MVMD and gradient judgment, comprising the following steps:

[0006] S1: Obtain the vibration signal of the equipment and perform MVMD decomposition to reconstruct the signal according to the correlation;

[0007] S2: compose the acceleration vector and impact vector respectively according to the reconstructed signal;

[0008] S3: Call the gradient judgment algorithm according to the cumulative length of the signal, and divide the working conditions according to the severity of the changes in the acceleration vector and the impact vector;

[0009] S4: Under each working condition, the early warning line and alarm line following the data trend are calculated through sliding average, and trend early warning and alarm are performed.

[0010] According to the above scheme, in step S1, the specific steps are:

[0011] S11: Obtain vibration signals for a period of time and perform normalization;

[0012] S12: Perform MVMD decomposition on the vibration signal to obtain the basic modal components and then normalize them;

[0013] S13: Calculate the correlation coefficient between each normalized vibration signal and the corresponding normalized fundamental modal component, add the fundamental modal components whose correlation coefficients are greater than a preset value, and obtain a reconstructed signal.

[0014] According to the above scheme, in step S2, the specific steps are:

[0015] S21: Calculate the effective value of acceleration and impact characteristic value of each reconstructed signal;

[0016] S22: forming an acceleration vector and an impact vector according to the effective value of acceleration and the impact characteristic value respectively.

[0017] According to the above scheme, in step S3, the specific steps are:

[0018] S31: Setting the gradient judgment window, the number of successful gradient judgments, and the gradient judgment threshold;

[0019] S32: Set loop: When the cumulative length of the signal is greater than or equal to the gradient judgment window, execute the next step; otherwise, output the data directly;

[0020] S33: Set the first outer loop, with the range of [gradient judgment window - 1, total data length - gradient judgment window]; if the ratio of adjacent data is greater than the gradient judgment threshold or less than the inverse of the gradient judgment threshold, execute the next step;

[0021] S34: Set the first inner loop, with the range of [0, number of successful gradient judgments], and if the ratio of the first and last data is greater than the gradient judgment threshold or less than the inverse of the gradient judgment threshold, record the last data;

[0022] S35: If the length of the tail data is greater than or equal to the number of successful gradient judgments, intercept the data according to the starting point of the first outer loop, and store the data before the first outer loop as a working condition;

[0023] S36: If the length of the intercepted first outer loop data is less than the total data length, continue the loop; if the length of the intercepted first outer loop data is equal to the total data length, output the processed first outer loop data and a section of data before the first outer loop.

[0024] Furthermore, in step S4, the specific steps are:

[0025] When the cumulative length of the signal is less than the gradient judgment window, the sliding average algorithm is used to calculate the warning number, alarm number, warning line and alarm line for each set of data in the acceleration vector and impact vector;

[0026] When the cumulative length of the signal is greater than or equal to the gradient judgment window, the warning lines of each group of data are combined according to the working conditions to form a warning curve diagram, and the alarm lines of each group of data are combined to form an alarm curve diagram.

[0027] Furthermore, in step S4, the specific steps of the sliding average algorithm are:

[0028] S41: Set the sliding average data length, attention threshold, warning threshold, early warning judgment window, and number of successful early warning judgments; if the signal length under the current working condition is less than or equal to the sliding average data length, execute the next step; if the signal length under the current working condition is greater than the sliding average data length, execute the second outer loop;

[0029] S42: Perform median filtering on the signal data and calculate the mean of the index data length of each number in the filtered data, and then multiply it by the warning threshold or attention threshold respectively to obtain the early warning line or alarm line accordingly;

[0030] S43: If the length of the data in the early warning judgment window is continuous and the number of successful early warning judgments is greater than the warning line, an early warning is issued; if the length of the data in the early warning judgment window is continuous and the number of successful early warning judgments is greater than the alarm line, an alarm is issued.

[0031] Furthermore, in step S4, the specific steps of the second outer loop are:

[0032] S44: setting the range of the second outer loop to [sliding average data length + 1, signal length under the current working condition + 1]; intercepting new data in the range of [starting point of the second outer loop - sliding average data length - 1, starting point of the second outer loop - 1];

[0033] S45: Perform median filtering on the new data and calculate the filtered mean, and then multiply it by the warning threshold or the attention threshold to obtain the early warning line or the alarm line respectively;

[0034] S46: If the number of data with a continuous length of the warning judgment window and the number of successful warning judgments is greater than the mean multiplied by the warning threshold, a warning is issued once; if the number of data with a continuous length of the warning judgment window and the number of successful warning judgments is greater than the mean multiplied by the attention threshold, an alarm is issued once.

[0035] Furthermore, the step S4 further includes the following steps:

[0036] S47: Calculate the average value of the index data length of each number in the data of a certain length, and then multiply it by the warning threshold or the attention threshold respectively to obtain the early warning line or the alarm line accordingly;

[0037] S48: Output the number of warnings, the number of alarms, the warning line and the alarm line.

[0038] An equipment failure trend warning system based on MVMD and gradient judgment,

[0039] The data acquisition submodule is used to acquire the vibration signal of the equipment and perform MVMD decomposition to reconstruct the signal according to the correlation;

[0040] A data reconstruction submodule, used to respectively form an acceleration vector and an impact vector according to the reconstructed signal;

[0041] The working condition segmentation submodule is used to call the gradient judgment algorithm according to the cumulative length of the signal and segment the working condition according to the severity of the change of the acceleration vector and the impact vector;

[0042] The trend warning submodule is used to calculate the warning line and alarm line that follow the data trend through sliding average under each working condition, and to perform trend warning and alarm.

[0043] A computer memory stores a computer program executable by a computer processor, wherein the computer program executes an equipment failure trend early warning method based on MVMD and gradient judgment.

[0044] The beneficial effects of the present invention are:

[0045] 1. The present invention provides an equipment failure trend warning method and system based on MVMD and gradient judgment. By adopting a gradient judgment algorithm to segment the working conditions, median filtering is performed for different working conditions to remove sudden jumps, and finally a sliding average algorithm is used to track trends and issue warnings, thereby achieving the function of accurately warning equipment failures.

[0046] 2. To address the situation where equipment early warnings have false alarms or missed alarms, the present invention adopts the MVMD decomposition algorithm to continuously extract all IMFs without knowing the number of modes. The computational complexity is low, and the noise interference in the signal is effectively removed, so that the early warning line changes in a timely manner, reducing the situation of false alarms and missed alarms.

[0047] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 is a flow chart of an embodiment of the present invention.

[0050] Figure 2 4 is a flow chart of an analysis algorithm according to an embodiment of the present invention.

[0051] Figure 3 4 is a flow chart of a gradient algorithm according to an embodiment of the present invention.

[0052] Figure 4 4 is a flow chart of a sliding average algorithm according to an embodiment of the present invention.

[0053] Figure 5 1 is a warning curve diagram of the effective value of acceleration according to an embodiment of the present invention.

[0054] Figure 6 3 is a warning curve diagram of the impact characteristic value of the embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0056] Example 1

[0057] See also Figure 1 ,The specific steps of a device failure trend early warning method based on MVMD and gradient judgment are as follows:

[0058] S1: Obtain the vibration signal of the equipment and perform MVMD decomposition to reconstruct the signal according to the correlation;

[0059] S2: compose the acceleration vector and impact vector respectively according to the reconstructed signal;

[0060] S3: Call the gradient judgment algorithm according to the cumulative length of the signal, and divide the working conditions according to the severity of the changes in the acceleration vector and the impact vector;

[0061] S4: Under each working condition, the early warning line and alarm line following the data trend are calculated through sliding average, and trend early warning and alarm are performed.

[0062] Furthermore, in step S1, the specific steps are:

[0063] S11: Obtain vibration signals for a period of time and perform normalization;

[0064] S12: Perform MVMD decomposition on the vibration signal to obtain the basic modal components and then normalize them;

[0065] S13: Calculate the correlation coefficient between each normalized vibration signal and the corresponding normalized fundamental modal component, add the fundamental modal components whose correlation coefficients are greater than a preset value, and obtain a reconstructed signal.

[0066] In step S2, the specific steps are:

[0067] S21: Calculate the effective value of acceleration and impact characteristic value of each reconstructed signal;

[0068] S22: forming an acceleration vector and an impact vector according to the effective value of acceleration and the impact characteristic value respectively.

[0069] In step S3, the specific steps are:

[0070] S31: Setting the gradient judgment window, the number of successful gradient judgments, and the gradient judgment threshold;

[0071] S32: Set loop: When the cumulative length of the signal is greater than or equal to the gradient judgment window, execute the next step; otherwise, output the data directly;

[0072] S33: Set the first outer loop, with the range of [gradient judgment window - 1, total data length - gradient judgment window]; if the ratio of adjacent data is greater than the gradient judgment threshold or less than the inverse of the gradient judgment threshold, execute the next step;

[0073] S34: Set the first inner loop, with the range of [0, number of successful gradient judgments], and if the ratio of the first and last data is greater than the gradient judgment threshold or less than the inverse of the gradient judgment threshold, record the last data;

[0074] S35: If the length of the tail data is greater than or equal to the number of successful gradient judgments, intercept the data according to the starting point of the first outer loop, and store the data before the first outer loop as a working condition;

[0075] S36: If the length of the intercepted first outer loop data is less than the total data length, continue the loop; if the length of the intercepted first outer loop data is equal to the total data length, output the processed first outer loop data and a section of data before the first outer loop.

[0076] Furthermore, in step S4, the specific steps are:

[0077] When the cumulative length of the signal is less than the gradient judgment window, the sliding average algorithm is used to calculate the warning number, alarm number, warning line and alarm line for each set of data in the acceleration vector and impact vector;

[0078] When the cumulative length of the signal is greater than or equal to the gradient judgment window, the warning lines of each group of data are combined to form a warning curve diagram, and the alarm lines of each group of data are combined to form an alarm curve diagram.

[0079] Furthermore, in step S4, the specific steps of the sliding average algorithm are:

[0080] S41: Set the sliding average data length, attention threshold, warning threshold, early warning judgment window, and number of successful early warning judgments; if the signal length under the current working condition is less than or equal to the sliding average data length, execute the next step; if the signal length under the current working condition is greater than the sliding average data length, execute the second outer loop;

[0081] S42: Perform median filtering on the signal data and calculate the mean of the index data length of each number in the filtered data, and then multiply it by the warning threshold or attention threshold respectively to obtain the early warning line or alarm line accordingly;

[0082] S43: If the length of the data in the early warning judgment window is continuous and the number of successful early warning judgments is greater than the warning line, an early warning is issued; if the length of the data in the early warning judgment window is continuous and the number of successful early warning judgments is greater than the alarm line, an alarm is issued.

[0083] Furthermore, in step S4, the specific steps of the second outer loop are:

[0084] S44: setting the range of the second outer loop to [sliding average data length + 1, signal length under the current working condition + 1]; intercepting new data in the range of [starting point of the second outer loop - sliding average data length - 1, starting point of the second outer loop - 1];

[0085] S45: Perform median filtering on the new data and calculate the filtered mean, and then multiply it by the warning threshold or attention threshold to obtain the early warning line or alarm line respectively;

[0086] S46: If the number of data with a continuous length of the warning judgment window and the number of successful warning judgments is greater than the mean multiplied by the warning threshold, a warning is issued once; if the number of data with a continuous length of the warning judgment window and the number of successful warning judgments is greater than the mean multiplied by the attention threshold, an alarm is issued once.

[0087] Furthermore, step S4 further includes the following steps:

[0088] S47: Calculate the average value of the index data length of each number in the data of a certain length, and then multiply it by the warning threshold or the attention threshold respectively to obtain the early warning line or the alarm line accordingly;

[0089] S48: Output the number of warnings, the number of alarms, the warning line and the alarm line.

[0090] This embodiment uses a gradient judgment algorithm to divide the working conditions, performs median filtering for different working conditions, removes sudden jumps, and finally uses a sliding average algorithm to track trends and issue early warnings, thereby achieving the function of accurately warning of equipment failures.

[0091] Example 2

[0092] The steps of this embodiment are the same as those of embodiment 1, except that each step is applied to a specific example. Figure 2 , specifically including the following steps:

[0093] S1: Obtain the vibration signal of the equipment and perform MVMD decomposition to reconstruct the signal according to the correlation. The specific steps are as follows:

[0094] S11: Obtain one year's raw acceleration vibration signal data x1, x2, ..., xn; normalize each signal X to obtain X_rec;

[0095] S12: Perform MVMD decomposition on the signal X to obtain the basic modal components IMF; normalize each basic modal component to obtain IMF_rec;

[0096] S13: Calculate the correlation coefficient between X_rec and each normalized fundamental modal component IMF_rec, add the fundamental modal components IMF with a correlation coefficient greater than 0.3 to obtain the reconstructed signal Y; obtain each reconstructed signal y1, y2, ..., yn;

[0097] S2: Form the acceleration vector and impact vector based on the reconstructed signal; the specific steps are:

[0098] S21: Calculate the effective value of acceleration and the impact characteristic value of each reconstructed signal y1, y2, ..., yn;

[0099] S22: forming an acceleration vector acc_rms and an impact vector imp_rms according to the acceleration effective value and the impact characteristic value respectively;

[0100] S3: Call the gradient judgment algorithm according to the cumulative length of the signal, and divide the working conditions according to the severity of the changes in the acceleration vector and the impact vector;

[0101] That is, when the data length is greater than TR, the gradient judgment algorithm is called on the vectors acc_rms and imp_rms to obtain the judged data acc_rms_new, imp_rms_new and acc_rms_rest, imp_rms_rest;

[0102] Figure 3 This is the gradient algorithm flow chart of this embodiment, and the specific steps are:

[0103] S31: Setting the gradient judgment window TR, the number of successful gradient judgments TL and the gradient judgment threshold TH;

[0104] S32: Set a While loop. When the length of the trend model calculation data pool is greater than or equal to the gradient judgment window TR, proceed to the next step. Otherwise, directly output the data.

[0105] S33: Set the outer loop, the outer loop i range is [TR-1, N-TR]; if the ratio of data[i+1] to data[i] is greater than TH or less than 1 / TH, proceed to the next step;

[0106] S34: Set the inner loop, the inner loop j range is [0, TL], when the ratio of data[i+j+1] to data[ij] is greater than TH or less than 1 / TH, record data[i+j+1] into the vector Thr;

[0107] S35: When the length of vector Thr is greater than or equal to TL, intercept the data data[0:i+1] and data[i+1:N], and store data[0:i+1] in data_other;

[0108] S36: When the length of the processed data data[i+1:N] is less than N, jump back to the While loop; when the length of the processed data is equal to N, return the processed data data and data_other.

[0109] S4: Under each working condition, the early warning line and alarm line following the data trend are calculated by sliding average, and trend early warning and alarm are performed; the specific steps are as follows:

[0110] When the data length is less than TR, the trend warning algorithm is called to obtain the number of warnings and alarms, and the warning lines and alarm lines of the data acc_rms and imp_rms are drawn.

[0111] When the data length is greater than or equal to TR, the trend warning algorithm is called to obtain the number of warnings and alarms, and the warning lines and alarm lines of each group of data in acc_rms and imp_rms are drawn, and the warning lines and alarm lines of each group of data are combined;

[0112] Figure 4 This is a flow chart of the sliding average algorithm of this embodiment, which specifically includes the following steps:

[0113] S41: Set the sliding average data length L, attention threshold H, warning threshold HH to 2.5, early warning judgment window NN, number of early warning judgment successes AN, and rising edge VH; when the length of the data entering the trend early warning algorithm is less than or equal to L, execute step S42; when the length of the data entering the trend early warning algorithm is greater than L, execute step S44;

[0114] S42: Perform median filtering on the data data, with a fixed median coefficient of 3; calculate the average value of the index data length of each number in the filtered data, and multiply it by HH and H to obtain the warning line and alarm line. For example, for data x1, x2, ..., xn, the nth warning value is [(x1+x2+...+xn) / n]*HH;

[0115] S43: When the number of AN in the consecutive NN numbers is greater than the warning point, a warning is issued once; when the number of AN in the consecutive NN numbers is greater than the alarm point, an alarm is issued once;

[0116] S44: Set an outer loop, the outer loop i range is [L+1, len(data)+1], intercept the data data_new, the interception range is [iL-1, i-1], and obtain the data data1;

[0117] S45: Perform median filtering on the data data1 with a filter coefficient of 19, and calculate the average of the filtered data to obtain avg; multiply avg by HH to obtain the warning line; multiply avg by H to obtain the alarm line;

[0118] S46: Accumulate NN data. When AN of the consecutive NN data are greater than avg multiplied by HH, issue an alarm. When AN of the consecutive NN data are greater than avg multiplied by H, issue an alarm.

[0119] S47: Calculate the average value of the index data length of each number within 1000 data points and multiply it by HH and H to obtain the early warning line and the alarm line;

[0120] S48: Returns the number of warnings and alarms, warning lines and alarm lines.

[0121] Example 3

[0122] The steps of this embodiment are the same as those of embodiment 1, except that each step is applied to a specific instance.

[0123] Each acceleration signal is normalized according to step S11 .

[0124] Set the maximum balance parameter , the double ascent step size is 0, and the tolerance of the convergence criterion is 10 -6 , according to step S12, each acceleration signal is decomposed by MVMD to obtain a series of basic modal components IMF; and then the basic modal components IMF are normalized.

[0125] According to step 13, the correlation coefficient between the normalized acceleration signal and the normalized fundamental modal components is calculated, and the fundamental modal components with correlation coefficients greater than 0.3 are added together to obtain a reconstructed signal.

[0126] According to step S2 , the acceleration effective value and the impact characteristic value of each group of reconstructed signals are calculated and formed into vectors acc_rms and imp_rms.

[0127] According to step S3, the accumulation times TR is set to 48. When the data length of acc_rms and imp_rms is greater than 48, the gradient judgment algorithm is called to divide acc_rms and imp_rms into data of different working conditions.

[0128] According to step S4, the sliding average trend warning algorithm is called to draw warning lines for data of different working conditions, such as Figure 5 The effective value of acceleration and Figure 6 As shown in the impact characteristic value; when the working conditions change, the early warning line changes accordingly; when the equipment fails, the alarm line will sound an alarm.

[0129] In this embodiment, in order to solve the problem of false alarms and missed alarms in equipment warnings, the MVMD decomposition algorithm is used to continuously extract all IMFs without knowing the number of modes. The computational complexity is low, and the noise interference in the signal is effectively removed, so that the warning line changes in time, reducing the situation of false alarms and missed alarms.

[0130] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0131] Example 4

[0132] This embodiment is used to implement the principles of the above method embodiment to construct an equipment failure trend warning system based on MVMD and gradient judgment, including a data acquisition submodule, a data reconstruction submodule, a working condition segmentation submodule and a trend warning submodule.

[0133] The data acquisition submodule is used to acquire the vibration signal of the equipment and perform MVMD decomposition to reconstruct the signal according to the correlation;

[0134] A data reconstruction submodule, used to respectively form an acceleration vector and an impact vector according to the reconstructed signal;

[0135] The working condition segmentation submodule is used to call the gradient judgment algorithm according to the cumulative length of the signal and segment the working condition according to the severity of the change of the acceleration vector and the impact vector;

[0136] The trend warning submodule is used to calculate the warning line and alarm line that follow the data trend through sliding average under each working condition, and to perform trend warning and alarm.

[0137] Each sub-module is mainly used to implement each step of the method embodiment, which will not be described in detail here.

[0138] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0139] This embodiment also includes a processor, a communication interface, a memory, and a communication bus; wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; the memory stores a computer program, and when the program is executed by the processor, the processor executes the steps of a device failure trend warning method based on MVMD and gradient judgment.

[0140] This embodiment further provides a computer-readable storage medium having executable instructions stored thereon. When the instructions are executed by a processor, the processor implements a device failure trend warning method based on MVMD and gradient judgment.

[0141] 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 take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware.

[0142] Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0143] This application is described with reference to the flowcharts of the method and computer program product according to Example 1 of the application and the block diagram of the device (system) according to Example 3. It should be understood that each process or block in the flowchart or block diagram, as well as combinations of processes or blocks in the flowchart or block diagram, can be implemented by computer program instructions.

[0144] 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 produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce the instructions for implementing the process Figure 1 a process or multiple processes or boxes Figure 1 An equipment failure trend early warning system based on MVMD and gradient judgment that specifies functions in one or more boxes.

[0145] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work 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 or boxes Figure 1 The function specified in one or more boxes.

[0146] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes or boxes Figure 1 The steps of a device failure trend early warning method based on MVMD and gradient judgment are specified in one or more boxes.

[0147] The above embodiments are intended only to illustrate the design concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. The scope of protection of the present invention is not limited to the above embodiments. Therefore, any equivalent changes or modifications made based on the principles and design concepts disclosed in the present invention are within the scope of protection of the present invention.

Claims

1. A device failure trend early warning method based on MVMD and gradient judgment, characterized by: The following steps are involved: S1: Obtain the vibration signal of the equipment and perform MVMD decomposition to reconstruct the signal according to the correlation; S2: compose the acceleration vector and impact vector respectively according to the reconstructed signal; S3: Call the gradient judgment algorithm according to the cumulative length of the signal, and divide the working conditions according to the severity of the changes in the acceleration vector and the impact vector; the specific steps are: S31: Setting the gradient judgment window, the number of successful gradient judgments, and the gradient judgment threshold; S32: Set loop: When the cumulative length of the signal is greater than or equal to the gradient judgment window, execute the next step; otherwise, output the data directly; S33: Set the first outer loop, with the range of [gradient judgment window - 1, total data length - gradient judgment window]; if the ratio of adjacent data is greater than the gradient judgment threshold or less than the inverse of the gradient judgment threshold, execute the next step; S34: Set the first inner loop, with the range of [0, number of successful gradient judgments], and if the ratio of the first and last data is greater than the gradient judgment threshold or less than the inverse of the gradient judgment threshold, record the last data; S35: If the length of the tail data is greater than or equal to the number of successful gradient judgments, intercept the data according to the starting point of the first outer loop, and store the data before the first outer loop as a working condition; S36: If the length of the intercepted first outer loop data is less than the total length of the data, continue the loop; if the length of the intercepted first outer loop data is equal to the total length of the data, output the processed first outer loop data and a section of data before the first outer loop; S4: Under each working condition, the early warning line and alarm line following the data trend are calculated through sliding average, and trend early warning and alarm are performed.

2. The equipment failure trend early warning method based on MVMD and gradient judgment according to claim 1 is characterized by: In the step S1, the specific steps are: S11: Obtain vibration signals for a period of time and perform normalization; S12: Perform MVMD decomposition on the vibration signal to obtain the basic modal components and then normalize them; S13: Calculate the correlation coefficient between each normalized vibration signal and the corresponding normalized fundamental modal component, add the fundamental modal components whose correlation coefficients are greater than a preset value, and obtain a reconstructed signal.

3. The equipment failure trend early warning method based on MVMD and gradient judgment according to claim 1 is characterized by: In the step S2, the specific steps are: S21: Calculate the effective value of acceleration and impact characteristic value of each reconstructed signal; S22: forming an acceleration vector and an impact vector according to the effective value of acceleration and the impact characteristic value respectively.

4. The equipment failure trend early warning method based on MVMD and gradient judgment according to claim 1 is characterized by: In the step S4, the specific steps are: When the cumulative length of the signal is less than the gradient judgment window, the sliding average algorithm is used to calculate the warning number, alarm number, warning line and alarm line for each set of data in the acceleration vector and impact vector; When the cumulative length of the signal is greater than or equal to the gradient judgment window, the warning lines of each group of data are combined according to the working conditions to form a warning curve diagram, and the alarm lines of each group of data are combined to form an alarm curve diagram.

5. The equipment failure trend early warning method based on MVMD and gradient judgment according to claim 4 is characterized by: In step S4, the specific steps of the sliding average algorithm are: S41: Set the sliding average data length, attention threshold, warning threshold, early warning judgment window, and number of successful early warning judgments; if the signal length under the current working condition is less than or equal to the sliding average data length, execute the next step; if the signal length under the current working condition is greater than the sliding average data length, execute the second outer loop; S42: Perform median filtering on the signal data and calculate the mean of the index data length of each number in the filtered data, and then multiply it by the warning threshold or attention threshold respectively to obtain the early warning line or alarm line accordingly; S43: If the length of the data in the early warning judgment window is continuous and the number of successful early warning judgments is greater than the warning line, an early warning is issued; if the length of the data in the early warning judgment window is continuous and the number of successful early warning judgments is greater than the alarm line, an alarm is issued.

6. The equipment failure trend early warning method based on MVMD and gradient judgment according to claim 5 is characterized by: In step S4, the specific steps of the second outer loop are: S44: setting the range of the second outer loop to [sliding average data length + 1, signal length under the current working condition + 1]; intercepting new data in the range of [starting point of the second outer loop - sliding average data length - 1, starting point of the second outer loop - 1]; S45: Perform median filtering on the new data and calculate the filtered mean, and then multiply it by the warning threshold or the attention threshold to obtain the early warning line or the alarm line respectively; S46: If the number of data with a continuous length of the warning judgment window and the number of successful warning judgments is greater than the mean multiplied by the warning threshold, a warning is issued once; if the number of data with a continuous length of the warning judgment window and the number of successful warning judgments is greater than the mean multiplied by the attention threshold, an alarm is issued once.

7. The equipment failure trend early warning method based on MVMD and gradient judgment according to claim 6 is characterized by: The step S4 further includes the following steps: S47: Calculate the average value of the index data length of each number in the data of a certain length, and then multiply it by the warning threshold or the attention threshold respectively to obtain the early warning line or the alarm line accordingly; S48: Output the number of warnings, the number of alarms, the warning line and the alarm line.

8. An equipment failure trend warning system based on MVMD and gradient judgment, characterized by: The data acquisition submodule is used to acquire the vibration signal of the equipment and perform MVMD decomposition to reconstruct the signal according to the correlation; A data reconstruction submodule, used to respectively form an acceleration vector and an impact vector according to the reconstructed signal; The working condition segmentation submodule is used to call the gradient judgment algorithm according to the cumulative length of the signal and segment the working condition according to the severity of the change of the acceleration vector and the impact vector; specifically: Set the gradient judgment window, the number of successful gradient judgments, and the gradient judgment threshold; Set up a loop: When the cumulative length of the signal is greater than or equal to the gradient judgment window, execute the next step; otherwise, output the data directly; Set the first outer loop with a range of [gradient judgment window - 1, total data length - gradient judgment window]; if the ratio of adjacent data is greater than the gradient judgment threshold or less than the inverse of the gradient judgment threshold, execute the next step; Set the first inner loop to [0, number of successful gradient judgments]. If the ratio of the first to the last data is greater than the gradient judgment threshold or less than the inverse of the gradient judgment threshold, record the last data. If the length of the tail data is greater than or equal to the number of successful gradient judgments, the data is intercepted according to the starting point of the first outer loop, and the data before the first outer loop is stored as a working condition; If the length of the intercepted first outer loop data is less than the total length of the data, the loop continues; if the length of the intercepted first outer loop data is equal to the total length of the data, the processed first outer loop data and the data before the first outer loop are output; The trend warning submodule is used to calculate the warning line and alarm line that follow the data trend through sliding average under each working condition, and to perform trend warning and alarm.

9. A computer memory, characterized in that: A computer program executable by a computer processor is stored therein, and the computer program executes the equipment failure trend warning method based on MVMD and gradient judgment as described in any one of claims 1 to 7.

Images