Hydraulic pump residual life prediction method and device based on time sequence characteristics and medium

Abstract

The invention relates to the technical field of computers, in particular to a hydraulic pump residual life prediction method and device based on time sequence characteristics and a medium. The method comprises the following steps: acquiring original operation data of a hydraulic pump; preprocessing the original operation data to obtain preprocessed operation data; performing feature extraction and dimension reduction on the preprocessed operation data to obtain dimension reduction features with time domain features and frequency domain features; and determining the residual life of the hydraulic pump according to the dimension reduction features and a pre-constructed prediction model. The problem of poor prediction precision caused by adopting a single index for prediction in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, device and medium for predicting the remaining life of a hydraulic pump based on time series characteristics. Background Art

[0002] The hydraulic pump is the heart of the hydraulic system. Once it fails, the entire hydraulic system will be paralyzed. Therefore, it is particularly important to accurately predict the remaining life of the hydraulic pump, which is of great significance to ensure the reliability and safety of the hydraulic pump. In related technologies, sensor information is mainly used to monitor its degradation process, which has the problem of low prediction accuracy; or a single indicator (such as return oil volume) is used for prediction, which has the problem of poor prediction accuracy due to the single reference indicator. Summary of the invention

[0003] The present application provides a method, device, electronic device and medium for predicting the remaining life of a hydraulic pump based on time series characteristics, which can solve the problem of poor precision and low accuracy in predicting the remaining life of a hydraulic pump in the prior art. The technical solution is as follows:

[0004] In a first aspect, a method for predicting the remaining life of a hydraulic pump based on time series characteristics is provided, the method comprising:

[0005] Obtain the original operating data of the hydraulic pump;

[0006] Preprocessing the original operation data to obtain preprocessed operation data;

[0007] Extract features and reduce dimension of preprocessed running data to obtain reduced dimension features with time domain features and frequency domain features;

[0008] Determine the remaining life of a hydraulic pump based on reduced dimensionality features and a pre-built prediction model.

[0009] In a second aspect, a device for predicting the remaining life of a hydraulic pump based on time series characteristics is provided, the device comprising:

[0010] A data acquisition module, used to acquire original operation data of the hydraulic pump;

[0011] A data preprocessing module is used to preprocess the original operation data to obtain preprocessed operation data;

[0012] A feature extraction module is used to extract features from the preprocessed running data and perform dimensionality reduction to obtain dimensionality reduction features with time domain features and frequency domain features;

[0013] The predictive analysis module is used to determine the remaining life of the hydraulic pump based on the reduced dimension features and the pre-built predictive model.

[0014] In a third aspect of an embodiment of the present application, an electronic device is disclosed. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0015] In a fourth aspect of the embodiments of the present application, a computer-readable storage medium is disclosed, which stores a computer program. When the computer program is executed by a processor, the steps of the above method are implemented.

[0016] The embodiment of the present application pre-processes the original operating data of the hydraulic pump to initially avoid the problem of affecting the prediction accuracy due to abnormal values ​​through data filtering, and then extracts time domain features and frequency domain features to predict the remaining life of the hydraulic pump through a prediction model. The prediction task is performed by combining the time domain features and the frequency domain features, taking into account the operating performance and frequency response characteristics that affect the life of the hydraulic pump. Compared with the technical solution that uses a single indicator for prediction, the prediction accuracy is significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.

[0018] Figure 1 A flowchart of a method for predicting the remaining life of a hydraulic pump based on time series characteristics according to an embodiment of the present application;

[0019] Figure 2 This is a structural diagram of a hydraulic pump remaining life prediction device based on time series characteristics according to an embodiment of the present application. DETAILED DESCRIPTION

[0020] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0021] The present application embodiment provides a method for predicting the remaining life of a hydraulic pump based on time series characteristics, such as Figure 1 As shown, the method includes: step S101 to step S104.

[0022] Step S101: Acquire original operation data of the hydraulic pump.

[0023] In the embodiment of the present application, the original operating data is used to characterize the parameters of the hydraulic pump performance, such as flow, pressure, speed, displacement, efficiency, etc.

[0024] Specifically, the original operation data can be pre-stored or obtained from the hydraulic pump when the preset prediction conditions are met. For example, the operation log of the hydraulic pump can be read from a preset database to obtain the original operation data; or when the preset prediction conditions are met (such as every 1 month or 1 quarter), a data acquisition instruction is sent to the hydraulic pump, so that the original operation data fed back by the hydraulic pump according to the data acquisition instruction is obtained, thereby saving storage space.

[0025] Step S102: pre-processing the original operation data to obtain pre-processed operation data.

[0026] Step S103: extracting features from the preprocessed operation data and performing dimensionality reduction to obtain dimensionality reduction features having time domain features and frequency domain features.

[0027] Specifically, a preset algorithm may be used to extract features, or a feature extraction model may be used to extract features. More specifically, different feature extraction models may be used to extract features of different dimensions, or the same model may be used to extract features.

[0028] Specifically, feature extraction can be performed on at least one of the flow parameters, pressure parameters, efficiency parameters, and oscillation parameters of the hydraulic pump to obtain time domain features and frequency domain features, where the time domain features include peak value, mean value, variance, mean square value, root mean square value, etc., and the frequency domain features may include transfer matrix, amplitude-frequency characteristics, etc.

[0029] Step S104: Determine the remaining life of the hydraulic pump based on the dimensionality reduction features and the pre-built prediction model.

[0030] Specifically, the prediction model may use a classifier, which is used for the classifier to output classification identifiers corresponding to different dimensionality reduction features, so as to determine different remaining lifespans of the hydraulic pump through the classification identifiers. For example, the classification identifier Tag1 indicates that the remaining lifespan of the hydraulic pump is N1 years.

[0031] In the embodiment of the present application, by preprocessing the original operating data of the hydraulic pump, the problem of affecting the prediction accuracy due to abnormal values ​​is initially avoided through data filtering, and then the time domain characteristics and frequency domain characteristics are extracted to predict the remaining life of the hydraulic pump through the prediction model. The prediction task is performed by combining the time domain characteristics and the frequency domain characteristics, taking into account the operating performance and frequency response characteristics that affect the life of the hydraulic pump. Compared with the technical solution that uses a single indicator for prediction, the prediction accuracy is significantly improved.

[0032] In some embodiments, step S102 further includes at least one of the following steps:

[0033] If the mean value of any variable parameter in the original running data is greater than twice the standard deviation of the variable parameter, the variable parameter is determined to be an abnormal point and a removal operation is performed;

[0034] If the data value of any variable parameter at any time point in the original operation data is zero, the data of any variable parameter at any time point is determined to be an abnormal point, and a removal operation is performed;

[0035] The original operation data is filtered so that the frequency of the original operation data is within a preset time frequency range.

[0036] The embodiment of the present application compares the mean with two times the standard deviation. When the mean is greater than two times the standard deviation, it indicates that the data dispersion is too high, thereby identifying abnormal data to avoid the influence of abnormal data on the recognition accuracy during the prediction process.

[0037] In the embodiment of the present application, the variable parameters are used to characterize the index parameters that affect the working state or performance of the hydraulic pump, such as flow, pressure, speed, displacement, efficiency, etc.

[0038] Specifically, the original operation data can be filtered by using a preset high-pass filter and a low-pass filter. In specific implementation, the original operation data can be sequentially input into a high-pass filter and a low-pass filter to eliminate the influence of low-frequency noise and high-frequency noise on the original operation data.

[0039] In some embodiments, step S103 further includes:

[0040] Step S1031: determining various variable parameters of preprocessing operation data;

[0041] Step S1032: Input each variable parameter into the preset time domain feature extraction model and frequency domain feature extraction model respectively to obtain the time domain features and frequency domain features corresponding to each variable parameter respectively, the time domain features at least include mean, maximum, minimum, kurtosis and standard deviation, and the frequency domain features at least include dominant frequency, second-order dominant frequency, third-order dominant frequency and third-order wavelet packet decomposition energy.

[0042] Specifically, the mean can be determined by referring to the following formula:

[0043] , where A is the mean, n is the signal length, is a signal, i is an integer from 1 to n.

[0044] Specifically, the standard deviation can be determined by referring to the following formula:

[0045] , where n is the signal length, is the signal, i is an integer from 1 to n, and A is the signal mean.

[0046] Specifically, the kurtosis can be determined by referring to the following formula:

[0047] , where K is the kurtosis, n is the signal length, is the signal, A is the signal mean, and σ is the signal standard deviation.

[0048] Specifically, the time domain feature extraction mode is generally a long short-term memory neural network LSTM model. The long short-term memory neural network LSTM is a special RNN, which is mainly implemented through three gated logics: forget gate, input gate and output gate. Among them, the forget gate is used to determine how much of the input data of the network at the current moment needs to be saved to the unit state; the input gate is used to determine how much of the unit state at the previous moment needs to be retained to the current moment; the output gate is used to control how much of the current unit state needs to be output to the current output value.

[0049] The forget gate formula is as follows:

[0050] ;

[0051] in, is the output of the forget gate, is the sigmoid activation function, and are weights and biases, is the hidden state of the previous time step, is the current input.

[0052] The input gate formula is as follows:

[0053] ;

[0054] ;

[0055] in, is the output of the input gate, is a candidate value, and is the weight, and is bias.

[0056] The output gate formula is as follows:

[0057] ;

[0058] in, is the current state of the memory cell, is the memory cell state at the previous time step.

[0059] In some embodiments, before step S1032, the method further includes:

[0060] Based on the sliding window, the standardized operating data in each sliding window is respectively input into the preset time domain feature extraction model and frequency domain feature extraction model to obtain the window time domain features and window frequency domain features of each sliding window.

[0061] Specifically, the sliding window size can be set as needed.

[0062] In this embodiment, the sliding window is fixed, that is, the length of data obtained after each sliding is the same, thereby achieving the purpose of data segmentation.

[0063] In some embodiments, step S104 further includes:

[0064] Input the dimension reduction features into the prediction model to obtain the prediction results;

[0065] Based on the prediction results, determine the remaining life of the hydraulic pump.

[0066] Specifically, the prediction result is generally a data identifier. The data identifier output by the prediction model is queried in a preset data identifier comparison table. The data identifier comparison table includes data identifiers corresponding to different remaining lifespans. Therefore, the corresponding remaining lifespan can be determined by query.

[0067] In some embodiments, before step S104, the method further includes:

[0068] Get training data set and test data set;

[0069] The training data set and the test data set are preprocessed and then feature extracted to obtain the time-frequency features of each variable parameter corresponding to each historical operation record in the training data set and the time-frequency features of each variable parameter corresponding to each historical operation record in the test data set. The time-frequency features include time domain features and frequency domain features.

[0070] According to the predefined correspondence between the operating time, different operating environment parameters and the remaining life of the hydraulic pump, the time-frequency features of the training data set and the time-frequency features of the test data set are marked with health status;

[0071] The pre-built initialization model is trained using the time-frequency features of the labeled training data set, and the trained initialization model is verified using the time-frequency features of the test data set until the evaluation accuracy of the trained prediction model reaches the preset evaluation parameters.

[0072] Specifically, the life span of the hydraulic pump can be defined as needed. For example, referring to Table 1, 125 is used as the life span, and the remaining life span greater than 125 is corrected to 125. The operating time is converted into a remaining life label; the operating time is matched to the full life cycle, and the failure is defined as 0 life span, and the stable operation is defined as 125.

[0073] Table 1

[0074]

[0075] When applied, the maximum life can be adjusted according to the different operating environments and different operating durations of the product.

[0076] In some embodiments, step S104 further includes:

[0077] Evaluate the health assessment results of hydraulic pumps based on their remaining life;

[0078] If the health assessment result is a fault, a device maintenance suggestion is generated and the device maintenance suggestion is output and displayed.

[0079] In the embodiment of the present application, the health assessment result is used to characterize the state of the hydraulic pump equipment, which may include the health, sub-health, and failure states of the equipment operation. Among them, health means that the product is operating normally, which can be defined as the outlet pressure and outlet flow rate being within 95% of the design conditions; sub-health means that the product can still operate normally, which can be defined as the outlet pressure and outlet flow rate being within 85% to 95% of the design conditions; and failure means that the product is faulty, which can be defined as the outlet pressure and outlet flow rate being outside 85% of the design conditions.

[0080] Another embodiment of the present application provides a device for predicting the remaining life of a hydraulic pump based on time series characteristics, such as Figure 2 As shown, the device 20 includes: a data acquisition module 201, a data preprocessing module 202, a feature extraction module 203 and a prediction analysis module 204.

[0081] The data acquisition module 201 is used to acquire the original operation data of the hydraulic pump;

[0082] The data preprocessing module 202 is used to preprocess the original operation data to obtain preprocessed operation data;

[0083] The feature extraction module 203 is used to extract features from the pre-processed operation data and perform dimension reduction to obtain dimension reduction features with time domain features and frequency domain features;

[0084] The prediction analysis module 204 is used to determine the remaining life of the hydraulic pump based on the dimension reduction features and the pre-built prediction model.

[0085] In the embodiment of the present application, by preprocessing the original operating data of the hydraulic pump, the problem of affecting the prediction accuracy due to abnormal values ​​is initially avoided through data filtering, and then the time domain characteristics and frequency domain characteristics are extracted to predict the remaining life of the hydraulic pump through the prediction model. The prediction task is performed by combining the time domain characteristics and the frequency domain characteristics, taking into account the operating performance and frequency response characteristics that affect the life of the hydraulic pump. Compared with the technical solution that uses a single indicator for prediction, the prediction accuracy is significantly improved.

[0086] Furthermore, the data preprocessing module includes at least one of the following submodules:

[0087] The outlier elimination submodule is used to determine that any variable parameter in the original operation data is an outlier and perform an elimination operation if the mean value of any variable parameter in the original operation data is greater than twice the standard deviation of the variable parameter;

[0088] The null value point elimination submodule is used to determine that any variable parameter in the original operation data is an abnormal point if the data value of the variable parameter is zero, and perform a elimination operation;

[0089] The filtering and denoising processing submodule is used to filter the original operation data so that the frequency of the original operation data is within a preset time frequency range.

[0090] Furthermore, the feature extraction module includes:

[0091] The variable determination submodule is used to determine the various variable parameters of the preprocessing operation data;

[0092] The feature extraction submodule is used to input each variable parameter into the preset time domain feature extraction model and frequency domain feature extraction model respectively, and obtain the time domain features and frequency domain features corresponding to each variable parameter respectively. The time domain features at least include mean, maximum, minimum, kurtosis and standard deviation, and the frequency domain features at least include main frequency, second-order main frequency, third-order main frequency, and third-order wavelet packet decomposition energy.

[0093] Furthermore, the feature extraction submodule:

[0094] The sliding window processing unit is used to input the standardized operating data in each sliding window into a preset time domain feature extraction model and a frequency domain feature extraction model based on the sliding window, so as to obtain the window time domain features and window frequency domain features of the sliding window of each sliding.

[0095] Furthermore, the prediction analysis module includes:

[0096] The prediction result determination submodule is used to input the dimension reduction features into the prediction model to obtain the prediction result;

[0097] The remaining life determination submodule is used to determine the remaining life of the hydraulic pump based on the prediction results.

[0098] Furthermore, before the prediction analysis module, the device also includes:

[0099] The sample data acquisition submodule is used to obtain the training data set and the test data set;

[0100] The sample feature extraction submodule is used to extract features from the training data set and the test data set after preprocessing, so as to obtain the time-frequency features of the training data set and the time-frequency features of the test data set, wherein the time-frequency features include time domain features and frequency domain features;

[0101] The sample marking processing submodule is used to mark the health status of the time-frequency features of the training data set and the time-frequency features of the test data set according to the corresponding relationship between the predefined operating time of the hydraulic pump, different operating environment parameters and the remaining life;

[0102] The model training processing submodule is used to train the pre-built initialization model using the time-frequency characteristics of the labeled training data set, and to verify the trained initialization model using the time-frequency characteristics of the test data set until the evaluation accuracy of the trained prediction model reaches the preset evaluation parameters.

[0103] Furthermore, the remaining life determination submodule also includes:

[0104] A health assessment unit, used to evaluate the health assessment result of the hydraulic pump according to the remaining life;

[0105] The evaluation and suggestion unit is used to generate equipment maintenance suggestions and output the equipment maintenance suggestions if the health evaluation result is a fault.

[0106] The hydraulic pump remaining life prediction device based on time series characteristics of this embodiment can execute the hydraulic pump remaining life prediction method based on time series characteristics shown in the embodiment of the present application. The implementation principle is similar and will not be repeated here.

[0107] Another embodiment of the present application provides a terminal, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0108] Specifically, the processor may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA or other programmable logic device, a transistor logic device, a hardware component or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0109] Specifically, the processor is connected to the memory via a bus, and the bus may include a path for transmitting information. The bus may be a PCI bus or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc.

[0110] The memory may be a ROM or other type of static storage device that can store static information and instructions, a RAM or other type of dynamic storage device that can store information and instructions, or an EEPROM, CD-ROM or other optical disk storage, optical disk storage (including compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0111] Optionally, the memory is used to store the code of the computer program for executing the solution of the present application, and the execution is controlled by the processor. The processor is used to execute the application program code stored in the memory to implement the actions of the device provided in the above embodiment.

[0112] Another embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which implements the above method when executed by a processor.

[0113] The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0114] It will be appreciated by those skilled in the art that all or some of the steps and systems in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0115] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions without violating the spirit of the present application. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. A method for predicting the remaining life of a hydraulic pump based on time series characteristics, characterized in that: The method comprises: Obtain the original operating data of the hydraulic pump; Preprocessing the original operation data to obtain preprocessed operation data; Extracting features and reducing the dimension of the preprocessed operation data to obtain reduced-dimensional features with time domain features and frequency domain features; The remaining life of the hydraulic pump is determined based on the dimensionality reduction features and the pre-built prediction model.

2. The method according to claim 1, characterized in that The preprocessing of the original operation data to obtain preprocessed operation data includes at least one of the following: If the mean value of any variable parameter in the original operation data is greater than twice the standard deviation of the variable parameter, the variable parameter is determined to be an abnormal point, and a removal operation is performed; If the data value of any variable parameter in the original operation data is zero, the variable parameter is determined to be an abnormal point, and a removal operation is performed; The original operation data is filtered so that the frequency of the original operation data is within a preset time frequency range.

3. The method according to claim 1, characterized in that The extracting features and reducing the dimension of the pre-processed operation data to obtain reduced dimension features having time domain features and frequency domain features includes: Determining various variable parameters of the preprocessing operation data; The variable parameters are respectively input into the preset time domain feature extraction model and frequency domain feature extraction model to obtain the time domain features and frequency domain features corresponding to the variable parameters, wherein the time domain features include at least mean, maximum, minimum, kurtosis and standard deviation, and the frequency domain features include at least dominant frequency, second-order dominant frequency, third-order dominant frequency and third-order wavelet packet decomposition energy.

4. The method according to claim 3, characterized in that The step of inputting the variable parameters into a preset time domain feature extraction model and a frequency domain feature extraction model respectively includes: Based on the sliding window, the standardized operating data in the sliding window of each sliding is respectively input into the preset time domain feature extraction model and the frequency domain feature extraction model to obtain the window time domain features and window frequency domain features of the sliding window of each sliding.

5. The method according to claim 1, characterized in that Determining the remaining life of the hydraulic pump based on the dimensionality reduction feature and the pre-built prediction model includes: Inputting the dimension reduction features into the prediction model to obtain a prediction result; The remaining life of the hydraulic pump is determined based on the prediction result.

6. The method according to claim 5, characterized in that Before inputting the dimension reduction feature into the pre-built prediction model to obtain the prediction output result, the method further includes: Get training data set and test data set; Preprocessing the training data set and the test data set respectively and then extracting features to obtain time-frequency features of the training data set and time-frequency features of the test data set, wherein the time-frequency features include time domain features and frequency domain features; Marking the time-frequency features of the training data set and the time-frequency features of the test data set for health status according to the predefined correspondence between the operating time, different operating environment parameters and the remaining life of the hydraulic pump; The pre-built initialization model is trained using the time-frequency features of the labeled training data set, and the trained initialization model is verified using the time-frequency features of the test data set until the evaluation accuracy of the trained prediction model reaches the preset evaluation parameters.

7. The method according to claim 5, characterized in that The determining the remaining life of the hydraulic pump based on the dimensionality reduction feature and the pre-built prediction model further includes: Evaluating a health assessment result of the hydraulic pump according to the remaining life; If the health assessment result is a fault, a device maintenance suggestion is generated and the device maintenance suggestion is output and displayed.

8. A device for predicting the remaining life of a hydraulic pump based on time series characteristics, characterized in that: include: A data acquisition module, used to acquire original operation data of the hydraulic pump; A data preprocessing module, used for preprocessing the original operation data to obtain preprocessed operation data; A feature extraction module is used to extract features and reduce the dimension of the pre-processed operation data to obtain reduced dimension features with time domain features and frequency domain features; The prediction analysis module is used to determine the remaining life of the hydraulic pump based on the dimensionality reduction features and the pre-built prediction model.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-readable instructions, and the processor is used to execute the computer-readable instructions, wherein the computer-readable instructions execute the method according to any one of claims 1 to 7 when executed.

10. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the method according to any one of claims 1 to 7.

Images