CN116776502B - Intelligent prediction method and system for service life of spring hydraulic mounting machine - Google Patents
Intelligent prediction method and system for service life of spring hydraulic mounting machine Download PDFInfo
- Publication number
- CN116776502B CN116776502B CN202311047692.0A CN202311047692A CN116776502B CN 116776502 B CN116776502 B CN 116776502B CN 202311047692 A CN202311047692 A CN 202311047692A CN 116776502 B CN116776502 B CN 116776502B
- Authority
- CN
- China
- Prior art keywords
- data
- life
- spring
- data set
- parameter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 64
- 239000011159 matrix material Substances 0.000 claims abstract description 70
- 238000004088 simulation Methods 0.000 claims abstract description 45
- 239000013598 vector Substances 0.000 claims abstract description 38
- 238000004364 calculation method Methods 0.000 claims abstract description 9
- 238000003062 neural network model Methods 0.000 claims abstract description 8
- 230000002159 abnormal effect Effects 0.000 claims description 16
- 230000007613 environmental effect Effects 0.000 claims description 13
- 238000010606 normalization Methods 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 230000005856 abnormality Effects 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to the field of data processing, and provides an intelligent prediction method and system for the service life of a spring hydraulic mounting machine, wherein the intelligent prediction method comprises the following steps: simulating springs in the spring hydraulic mounting machine by using a spring simulation model to obtain time sequence data of each parameter of the springs in the simulation process, thereby obtaining a prediction matrix; determining a life characteristic vector corresponding to each moment in the simulation process based on the prediction matrix; and predicting the service life of the spring at the current moment based on the service life characteristic vector corresponding to each moment in the simulation process by using the neural network model. According to the scheme, influences on the service life of the spring from multiple aspects and angles are considered, so that the service life of the spring under different use environments is predicted more accurately, and the calculation is simple.
Description
Technical Field
The application relates to the field of data processing, in particular to an intelligent prediction method and system for the service life of a spring hydraulic mounting machine.
Background
At present, with the appearance of various springs, the service lives of various springs in the market are different. Because the spring has good torsion force characteristics, the spring is widely applied to the mechanical field and becomes an irreplaceable part in mechanical equipment. However, for a spring with a poor service life, the use of the machine equipment is indirectly affected, and even serious potential safety hazards are brought.
Along with the rapid development of the data processing field, the service life data of the spring are often visualized, and the service life of the spring is predicted and analyzed by using a prediction model. For example, a common weighted moving average algorithm is often used for time sequence prediction, and can eliminate the influence of periodic fluctuation and random fluctuation of a sequence on prediction to a certain extent, so as to obtain a good prediction result. However, the selection of the weights in the algorithm is complex, the adaptive weights are required to be carried out according to the characteristics of the time sequence, so that a better prediction result can be obtained, and the calculation is complex.
Disclosure of Invention
The application provides an intelligent prediction method and system for the service life of a spring hydraulic mounting machine, which are used for considering the influence of multiple aspects and angles on the service life of the spring, so that the prediction service life of the spring under different use environments is more accurate and the calculation is simple.
In a first aspect, the application provides an intelligent prediction method for the service life of a spring hydraulic mounting machine, which comprises the following steps:
simulating springs in the spring hydraulic mounting machine by using a spring simulation model to obtain time sequence data of each parameter of the springs in the simulation process, thereby obtaining a prediction matrix;
determining a life characteristic vector corresponding to each moment in the simulation process based on the prediction matrix;
and predicting the service life of the spring at the current moment based on the service life characteristic vector corresponding to each moment in the simulation process by using the neural network model.
In an embodiment, determining a lifetime feature vector corresponding to each time in the simulation process based on the prediction matrix includes:
grouping the time series data in the prediction matrix to obtain a plurality of data sets corresponding to each time series data, wherein each data set represents different use environments;
determining an ideal life index and life influence degree under the corresponding use environment of each data set;
predicting the predicted life of each data set under the corresponding use environment based on the ideal life index and the life influence degree of each data set under the corresponding use environment;
and constructing a life characteristic vector corresponding to each moment in the simulation process based on the predicted life and the prediction matrix.
In one embodiment, determining the ideal life index and the life impact level in the corresponding use environment for each data set includes:
calculating an ideal life index under the use environment corresponding to each data group based on the environment suitability corresponding to each data group and the spring loss rate corresponding to each data group;
and determining the service life influence degree corresponding to each data group based on the abnormal data points in each data group.
In an embodiment, the method further comprises:
calculating the corresponding environment suitability degree of each data group based on the variation coefficient of each data group and the difference value of the maximum value and the minimum value in each data group;
the computing mode of the environment suitability degree corresponding to each data group is as follows:
wherein ,indicating the environmental suitability of the jth data set in the time-series data of the ith parameter,/for the jth data set>For normalization function-> and />Maximum value and minimum value of j-th data group in time series data respectively representing i-th parameter,/->A coefficient of variation of the j-th data set in the time-series data representing the i-th parameter,/->Is an error term.
In an embodiment, the method further comprises:
determining the similarity between each data group in the time series data of the elastic force parameters and each data group in the time series of the stress parameters, determining the sum of differences between two adjacent data groups in each data group in the time series data of the stiffness parameters, and calculating the spring loss rate corresponding to each data group based on the calculated sum of the similarity and the differences;
the calculation mode of the spring loss rate corresponding to each data set is as follows:
wherein ,spring loss rate representing the j-th data set in the prediction matrix,/>Representation->Distance function->J-th data set of time-series data representing elasticity parameters in the prediction matrix,/th data set of time-series data representing elasticity parameters in the prediction matrix>J-th data set in time series representing stress parameters in the prediction matrix,/th data set in time series representing stress parameters in the prediction matrix>Similarity between the jth data set in the time series data representing the spring force parameter and the jth data set in the time series of the stress parameter, m representing the number of data in each data set,/> and />Adjacent (q-1) th data in the j-th data group of the time-series data respectively representing the stiffness parameters in the prediction matrix,/>Jth in time series data representing stiffness parametersThe sum of the differences between two adjacent data in each data set.
In one embodiment, calculating an ideal life index for each data set in a use environment based on the environmental fitness corresponding to each data set and the spring rate corresponding to each data set comprises:
calculating an ideal life index under the corresponding use environment of each data group by using the following formula:
wherein ,indicating the ideal life index in the use environment corresponding to the jth data set,/for the data set>Spring loss rate representing the j-th data set in the prediction matrix,/>The environmental suitability of the j-th data group in the time-series data of the i-th parameter is represented, and k represents the number of parameters.
In one embodiment, determining a degree of life impact for each data set based on outlier data points in each data set includes:
performing anomaly detection on the time series data of each parameter, and determining an anomaly data point in the time series data of each parameter;
if the j-th data group in the time series data of the current parameter has abnormal data points, determining a life impact index of the j-th data group in the time series data of the current parameter based on the sum of the difference value between each data in the j-th data group in the time series data of the current parameter and each abnormal data point in the j-th data group in the time series data of the current parameter;
and adding the life influence indexes of the j-th data set in the time sequence data of all the parameters to obtain the life influence degree corresponding to the j-th data set, and calculating the life influence degree corresponding to each data set.
In one embodiment, predicting a predicted lifetime in a use environment corresponding to each data set based on an ideal lifetime index and a lifetime impact degree in the use environment corresponding to each data set includes:
calculating the ratio between the ideal life index of the jth data set and the life influence degree of the jth data set, and normalizing the calculated ratio;
and calculating the product between the standard service life and the normalization result, so as to obtain the predicted service life of the j-th data set under the corresponding service environment.
In an embodiment, constructing a lifetime feature vector corresponding to each time in the simulation process based on the predicted lifetime and the prediction matrix includes:
and constructing a life characteristic vector corresponding to each moment in the simulation process based on the elastic parameter corresponding to each moment in the prediction matrix, the rigidity parameter corresponding to each moment and the stress parameter corresponding to each moment.
In a second aspect, the present application provides an intelligent predicting system for the service life of a spring hydraulic mounting machine, comprising:
the parameter acquisition module is used for simulating the springs in the spring hydraulic mounting machine by using a spring simulation model, and acquiring time sequence data of each parameter of the springs in the simulation process so as to obtain a prediction matrix;
the vector determining module is used for determining a life characteristic vector corresponding to each moment in the simulation process based on the prediction matrix;
and the life prediction module is used for predicting the service life of the spring at the current moment based on the life characteristic vector corresponding to each moment in the simulation process by utilizing the neural network model.
The intelligent prediction method and system for the service life of the spring hydraulic mounting machine have the beneficial effects that the intelligent prediction method and system are different from the prior art, and comprise the following steps: simulating springs in the spring hydraulic mounting machine by using a spring simulation model to obtain time sequence data of each parameter of the springs in the simulation process, thereby obtaining a prediction matrix; determining a life characteristic vector corresponding to each moment in the simulation process based on the prediction matrix; and predicting the service life of the spring at the current moment based on the service life characteristic vector corresponding to each moment in the simulation process by using the neural network model. According to the scheme, influences on the service life of the spring from multiple aspects and angles are considered, so that the service life of the spring under different use environments is predicted more accurately, and the calculation is simple.
Drawings
FIG. 1 is a flow chart of an embodiment of an intelligent spring life prediction method for a spring hydraulic mounting machine according to the present application;
FIG. 2 is a flowchart illustrating an embodiment of the step S12 in FIG. 1;
FIG. 3 is a flowchart illustrating an embodiment of the step S22 in FIG. 2;
FIG. 4 is a schematic diagram of an intelligent predicting system for life of springs of a hydraulic spring mounting machine according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
The present application will be described in detail with reference to the accompanying drawings and examples.
Referring to fig. 1, fig. 1 is a schematic flow chart of an embodiment of an intelligent predicting method for a service life of a spring hydraulic mounting machine, which specifically includes:
step S11: and simulating the springs in the spring hydraulic mounting machine by using a spring simulation model to obtain time sequence data of each parameter of the springs in the simulation process, thereby obtaining a prediction matrix.
When abnormal conditions occur in the working process of the spring, the service life of the spring is more likely to be influenced. Therefore, in order to predict the service life of the spring of the hydraulic mounting machine, the spring in the hydraulic mounting machine is simulated by using a MakeReal3D spring simulation model, time series data of spring force, spring stiffness and spring stress in the simulation model are obtained, and the three groups of time series data are used as a basic data sequence of the spring service life prediction. In the application, the data parameters are collected at intervals of t, the length of the time sequence is recorded as n, n=500 and t=2s are set in the application, and an implementer can define the time interval and the sequence length according to the needs, so that the implementation is reasonable.
Here, based on the acquired three-dimensional time-series data, the time-series data of each dimension is normalized in order to avoid the influence of the difference in dimension between the different dimensions on the prediction. Constructing a prediction matrix from the aboveThe method comprises the following steps:
in the formula (I), the total number of the components,for a prediction matrix, for better differentiation of the data sequences, note +.>Time-series data representing the ith parameter, wherein +.>I.e. prediction matrix->The first line in (a) represents time series data of spring force parameters, respectively>I.e. prediction matrix->Time series data of the spring rate parameter are shown in the second row,/->I.e. prediction matrix->The third row in (c) represents time series data of spring stress parameters.
In addition, it should be noted that, the time sequence of the spring force parameters is an incremental sequence, that is, the spring force of the spring increases with the lapse of time, the spring force change of the spring can be increased by 0.1N every 1s, and the practitioner can select the appropriate spring force change by himself. The spring stress and the spring stiffness also change correspondingly with the change of the spring force. Thus, a prediction matrix for predicting the service life of the spring can be obtained.
Step S12: and determining a life characteristic vector corresponding to each moment in the simulation process based on the prediction matrix.
Specifically, referring to fig. 2, step S12 specifically includes:
step S21: grouping the time series data in the prediction matrix to obtain a plurality of data sets corresponding to each time series data, wherein each data set represents different use environments.
The present application aims to predict the life of a spring from acquired data so as to predict the life of the spring in different use environments. Based on the constructed prediction matrix, the method analyzes sequence data from different dimensions and prepares for predicting the service life of the spring.
In order to predict time-series data of ith row in matrixAs examples, namely:
due to the different circumstances of use of the springs,which are presented in different sequence ranges over the data sequence. In order to simulate different usage environments of the springs, it is necessary to perform a sequence sliding window on the data sequence, with the sequence sliding window dividing the time sequence data into different data sets, wherein each set characterizes each usage environment of the springs. The size of the sequence sliding window is 15, namely, each data point is taken as a center point, the center point data, 7 data on the left side of the center point and 7 data on the right side of the center point are recorded as one group, the average value is adopted to replace the data loss, the time series data is divided into different data groups in a continuous sliding window mode, namely, each data group contains 15 data, so that the number of groups divided into each row of time series is n according to the acquisition time, and the j data group in the i row of time series data is recorded as the j data group。
Based on the preparation work, each data set characterizes different use environments of the springs, and the data characteristics of each data set can reflect whether the use environments of the springs are suitable to a certain extent. The more suitable the environment of use of the spring, the longer the spring life under that environment of use should be.
Specifically, the prediction matrix of the present applicationThe method specifically comprises 3 rows of time sequence data, namely time sequence data of the elastic parameters of the 1 st row, and n data groups are obtained after grouping; time series data of the second behavior spring stiffness parameter are grouped to obtain n data groups; time series data of the third behavior spring stress parameter, which are grouped, also obtain n data sets. It will be appreciated that the data set after the 3-line time series data packet represents a parameter change in the use environment for the same time period.
Step S22: an ideal life index and a life impact degree under the corresponding use environment of each data set are determined.
In one embodiment, referring to fig. 3, step S22 includes:
step S31: an ideal life index for each data set in the corresponding use environment is calculated based on the environmental fitness corresponding to each data set and the spring rate corresponding to each data set.
Specifically, in this embodiment, the environmental suitability corresponding to each data set is calculated based on the coefficient of variation of each data set and the difference between the maximum value and the minimum value in each data set. In one embodiment, the environmental fitness corresponding to each data set is calculated by:
wherein ,indicating the environmental suitability of the jth data set in the time-series data of the ith parameter,/for the jth data set>For normalization function-> and />Maximum value and minimum value of j-th data group in time series data respectively representing i-th parameter,/->A coefficient of variation of the j-th data set in the time-series data representing the i-th parameter,/->The absolute value is prevented from being 0, and the empirical value is 1. Specifically, the elastic parameter in the prediction matrix is recorded as the 1 st parameter, the stiffness parameter is recorded as the 2 nd parameter, and the stress parameter is recorded as the 3 rd parameter.
Absolute difference in data setThe smaller the size, the description of the environment of useThe smaller the fluctuation of the spring data, i.e. the smaller the abrasion and the smaller the damage to the materials caused by the use of the spring in the environment, the greater the environment suitability, i.e. the longer the service life of the spring in the use environment. At the same time, the coefficient of variation in the data set>The smaller the usage environment, the smaller the degree of dispersion of the data, i.e. the smaller the abrasion and the less destructive the material of the usage spring in the environment, the greater the environment suitability, i.e. the longer the service life of the spring in the usage environment. The calculation of the coefficient of variation is known in the art and will not be described in detail herein.
In addition, from the correlation between time series data, in the normal use process of the spring, certain regular changes exist among data parameters. For example, in normal use, there is a strong correlation between spring force and spring stress, as these two parameters are related to the deformation of the spring, i.e. the spring force changes, and the spring deformation changes, and the spring stress changes. However, once the use environment of the spring is worse, the correlation between the two is more likely to be broken, and the change of the data generates an abnormal phenomenon, so that the service life of the spring is greatly influenced. With respect to the spring rate, when the use condition of the spring is good, the spring rate tends to a fixed value, and when the use condition of the spring is bad, the spring rate may be abnormally changed.
Based on the method, the similarity between each data set in the time series data of the elastic force parameter and each data set in the time series of the stress parameter is determined, the sum of the differences between two adjacent data sets in each data set in the time series data of the stiffness parameter is determined, and the spring loss rate corresponding to each data set is calculated based on the calculated sum of the similarity and the difference. In one embodiment, the spring loss rate for each data set is calculated by:
wherein ,spring loss rate representing the j-th data set in the prediction matrix,/>Representation->Distance function->J-th data set of time-series data representing elasticity parameters in the prediction matrix,/th data set of time-series data representing elasticity parameters in the prediction matrix>J-th data set in time series representing stress parameters in the prediction matrix,/th data set in time series representing stress parameters in the prediction matrix>Similarity between the jth data set in the time series data representing the spring force parameter and the jth data set in the time series of the stress parameter, m representing the number of data in each data set,/> and />Adjacent (q-1) th data in the j-th data group of the time-series data respectively representing the stiffness parameters in the prediction matrix,/>And a sum of differences between adjacent two data in the j-th data set in the time-series data representing the stiffness parameter.
As the spring force changes and the spring deformation changes, the spring stress changes, namely, the smaller the distance dtw between the jth data group in the time sequence data of the 1 st parameter, namely the spring force parameter, and the jth data group in the time sequence data of the 3 rd parameter, namely the stress parameter, the greater the similarity between the sequences, namely, the better the spring running condition, the smaller the spring loss rate, and the longer the spring life in the use environment. Meanwhile, when the use condition of the spring is good, the spring stiffness tends to be a fixed value, namely the smaller the difference value between two adjacent data, such as the q-th data and the (q-1) -th data, in the j-th data group in the time sequence of the 2 nd parameter, namely the stiffness parameter, in the prediction matrix, is, the stronger the stability of the spring stiffness data sequence at the moment is, the better the use condition is, the smaller the spring loss rate is, namely the longer the spring life in the use environment is.
So far, based on analysis among different dimensions of the prediction matrix, the environment suitability degree and the spring loss rate are obtained. The service life of the spring can be reflected to a certain extent by the environment suitability and the spring loss rate, but the influence of the service environment on the spring is considered at the moment, and the abnormal phenomenon existing in the spring is not considered, so that the service life of the spring reflected in the service environment can be recorded as an ideal service life index at the moment.
Thus, the ideal life index under the use environment corresponding to each data set is calculated based on the above calculated environment suitability and spring loss rate by using the following formula:
wherein ,indicating the ideal life index in the use environment corresponding to the jth data set,/for the data set>Representing the spring loss rate of the jth data set (in particular the jth data set corresponding to the 3 parameters of spring force, stiffness and stress) in the prediction matrix,/, for the (c)>Indicating the environmental suitability of the j-th data group in the time-series data of the i-th parameter,k represents the number of parameters, which in the present application include spring force parameters, stiffness parameters and stress parameters, i.e. k=3. />And the sum of the environmental suitability of the j-th data set corresponding to the elasticity parameter, the rigidity parameter and the stress parameter is represented, and the calculated ideal life index is also the ideal life index in the use environment corresponding to the j-th data set corresponding to the elasticity parameter, the rigidity parameter and the stress parameter.
The smaller the spring loss rate of the jth data set in the prediction matrix, which indicates that the smaller the damage degree of the spring in the use environment is, the ideal life index of the spring in the use environment isThe larger. Meanwhile, the environmental suitability degree of the j-th data group in the time-series data of the i-th parameter +.>The larger the service environment is, the more suitable for the use of the spring is, the less possibility is provided for reducing the service life of the spring, and the ideal life index of the spring in the service environment is +.>The larger.
Step S32: and determining the service life influence degree corresponding to each data group based on the abnormal data points in each data group.
Specifically, abnormality detection is performed on the time-series data of each parameter, and abnormal data points in the time-series data of each parameter are determined. In one embodiment, the anomalies in the use of the spring itself may be reflected to some extent in view of anomalies in the data that may affect the life of the spring. Therefore, the time series data of each parameter is subjected to anomaly detection by using an LOF anomaly detection algorithm, the LOF value of each data point is calculated through the k 'neighborhood, the empirical value of k' is 10, and all the anomaly data points of the time series data of each parameter are obtained, so that the anomaly data points in each data group can be determined.
If there are abnormal data points in the j-th data group in the time-series data of the current parameter, a life impact index of the j-th data group in the time-series data of the current parameter is determined based on a sum of differences between each data in the j-th data group in the time-series data of the current parameter and each abnormal data point in the j-th data group in the time-series data of the current parameter.
Specifically, the life impact index is calculated by:
in the formula (I), the total number of the components,life impact index,/for the j-th data set in the time series data representing the i-th parameter in the prediction matrix>Representing a criterion when there are outlier data points in the j-th data set of the time series data of the i-th parameter in the prediction matrix +.>The value is 1; on the contrary, let(s)>The value is 0. That is, when there is an abnormal data point in the j-th data group in the time-series data of the i-th parameter,/-th>, wherein ,/>Representing the sum of the differences between each data in the j-th data set in the time-series data of the i-th parameter and each outlier data point in the j-th data set in the time-series data of the i-th parameter, wherein +.>The (q) th data in the (j) th data group in the time-series data representing the (i) th parameter in the prediction matrix>An r-th outlier data point in a j-th data set in the time-series data representing an i-th parameter in the prediction matrix. m represents the number of data in each data group, < >>Representing the number of outlier data points within each data set. Difference between each data point and the outlier in the data set +.>The larger and the number of outlier data points within the data set +.>The larger the spring is, the greater the influence degree of the abnormal phenomenon of the spring on the use of the spring is, and the greater the service life influence degree is; conversely, the difference between each data point in the data set and the outlier +.>The smaller and the number of outlier data points within the data set +.>The less, the less the life impact, even if no outlier data points are contained in the j-th group in the i-th row in the prediction matrix.
It will be appreciated that if there are no outlier data points in the j-th data set in the time series data for the i-th parameter,taking a value of 0, then the lifetime effect index of the j-th data set in the time series data of the i-th parameter,/->Is 0.
And adding the life influence indexes of the j-th data set in the time sequence data of all the parameters to obtain the life influence degree corresponding to the j-th data set, and calculating the life influence degree corresponding to each data set. Specifically, the calculation mode of the life impact degree corresponding to the jth data set is as follows:
wherein ,indicating the extent of the influence of the lifetime corresponding to the jth data set,/->For normalizing the function, k represents the number of parameters, the value of the application is 3,/for>The lifetime effect index of the j-th data set representing the spring force parameter, the stiffness parameter and the stress parameter is added.
Step S23: the predicted life in the use environment corresponding to each data set is predicted based on the ideal life index and the life influence degree in the use environment corresponding to each data set.
Calculating the ratio of the ideal life index of the jth data set to the life influence degree of the jth data setNormalizing the calculated ratio; and calculating the product between the standard service life and the normalization result, so as to obtain the predicted service life of the j-th data set under the corresponding service environment. In one embodiment, the predicted lifetime +.>The calculation mode of (a) is as follows:
in the formula (I), the total number of the components,representing the predicted lifetime of the j-th data set in the prediction matrix, +.>Indicating the standard service life of the springs, in general, each spring will have a standard service life, but due to the influence of the environment of use, the service life of the springs will be reduced, the empirical value of M being +.>And twice. />Indicating the extent of life effect in the use environment characterized by the jth data set in the prediction matrix,/for the prediction matrix>Representing the ideal life index, +_f, of the environment in which the j-th data set in the prediction matrix is characterized in use>For normalization function->As an error term, the denominator value was avoided to be 0, and the empirical value was 0.0001.
Degree of influence of lifeThe smaller the spring is, the more the ideal life index is, the more normal the spring operates, and the predicted life of the spring in the use environment is more similar to the standard service life; conversely, when the degree of influence of life is +.>The larger the ideal life index is, the smaller the predicted life of the spring in the use environment is, namely the loss of the spring is, and the spring operates abnormallyThe larger. Thus, the predicted service life of each data group in the prediction matrix under the corresponding use environment is obtained.
Step S24: and constructing a life characteristic vector corresponding to each moment in the simulation process based on the predicted life and the prediction matrix.
And constructing a life characteristic vector corresponding to each moment in the simulation process based on the elastic parameter corresponding to each moment in the prediction matrix, the rigidity parameter corresponding to each moment and the stress parameter corresponding to each moment.
Because all data sets have the same length as the time series data, namely, a predicted life is obtained at each moment, and a life characteristic vector at each moment is constructed according to the obtained predicted life and by combining a prediction matrixThe method comprises the following steps:
in the formula (I), the total number of the components,a lifetime feature vector representing the time at time f, +.>Indicating the predicted lifetime at time f, +.>The spring force, spring rate and spring stress acquired at time f are shown, respectively.
Step S13: and predicting the service life of the spring at the current moment based on the service life characteristic vector corresponding to each moment in the simulation process by using the neural network model.
By the method, the life characteristic vector corresponding to n-1 moments in the simulation process can be obtained, and the neural network is utilized to predict the life in the life characteristic vector at the nth moment. The neural network is an LSTM neural network, an Adam algorithm is used as an optimization algorithm, a cross entropy function is used as a loss function, the input of the neural network is a life characteristic vector of the previous (n-1) moment, and the output of the neural network is the first parameter in the life characteristic vector of the nth moment, namely the predicted life of the nth moment. Therefore, the service life of the spring at the current moment can be predicted.
According to the intelligent prediction method for the service life of the spring hydraulic mounting machine, the time series data are grouped according to the obtained time series data, so that different use environments of the spring are simulated, and the environment suitability is calculated according to the data characteristics of each group sequence in each dimension time series. Meanwhile, the ideal life index is obtained by combining the similarity between the dimension sequences in normal use and abnormal use of the spring and the change rule of the spring stiffness, and the beneficial effect of the method is that the ideal life of the spring under normal conditions is measured. In order to make the completeness of the life prediction of the spring stronger, the influence of data abnormality on the life of the spring is considered, and then the life prediction is carried out, the effective effect is that the influence of multiple aspects and angles on the life of the spring is considered, the completeness of the life prediction is made stronger, and the life prediction of the spring under different use environments is made more accurate.
Referring to fig. 4, fig. 4 is a schematic structural diagram of an embodiment of an intelligent predicting system for life of a spring hydraulic mounting machine, where the system of the embodiment can implement the method described in any of the foregoing embodiments, and the system of the embodiment specifically includes: a parameter acquisition module 41, a vector determination module 42 and a lifetime prediction module 43.
The parameter acquisition module 41 is configured to simulate a spring in the spring hydraulic mounting machine by using a spring simulation model, and obtain time series data of each parameter of the spring in the simulation process, thereby obtaining a prediction matrix.
The vector determining module 42 is configured to determine a lifetime feature vector corresponding to each time in the simulation process based on the prediction matrix.
The life prediction module 43 is configured to predict a life of the spring at a current time based on a life feature vector corresponding to each time in the simulation process by using the neural network model.
The foregoing is only the embodiments of the present application, and therefore, the patent scope of the application is not limited thereto, and all equivalent structures or equivalent processes using the descriptions of the present application and the accompanying drawings, or direct or indirect application in other related technical fields, are included in the scope of the application.
Claims (2)
1. An intelligent prediction method for the service life of a spring hydraulic mounting machine is characterized by comprising the following steps:
simulating springs in the spring hydraulic mounting machine by using a spring simulation model to obtain time sequence data of each parameter of the springs in the simulation process, thereby obtaining a prediction matrix;
determining a life characteristic vector corresponding to each moment in the simulation process based on the prediction matrix;
predicting the service life of the spring at the current moment based on the service life characteristic vector corresponding to each moment in the simulation process by utilizing the neural network model;
determining a life characteristic vector corresponding to each moment in the simulation process based on the prediction matrix, wherein the life characteristic vector comprises:
grouping the time series data in the prediction matrix to obtain a plurality of data sets corresponding to each time series data, wherein each data set represents different use environments;
determining an ideal life index and life influence degree under the corresponding use environment of each data set;
predicting the predicted life of each data set under the corresponding use environment based on the ideal life index and the life influence degree of each data set under the corresponding use environment;
constructing life characteristic vectors corresponding to each moment in the simulation process based on the predicted life and the prediction matrix;
determining an ideal life index and a life impact degree under the corresponding use environment of each data set comprises:
calculating an ideal life index under the use environment corresponding to each data group based on the environment suitability corresponding to each data group and the spring loss rate corresponding to each data group;
determining the service life influence degree corresponding to each data group based on the abnormal data points in each data group;
the method further comprises the steps of:
calculating the corresponding environment suitability degree of each data group based on the variation coefficient of each data group and the difference value of the maximum value and the minimum value in each data group;
the computing mode of the environment suitability degree corresponding to each data group is as follows:
wherein ,indicating the environmental suitability of the jth data set in the time-series data of the ith parameter,/for the jth data set>For normalization function-> and />Maximum value and minimum value of j-th data group in time series data respectively representing i-th parameter,/->A coefficient of variation of the j-th data set in the time-series data representing the i-th parameter,/->Is an error term;
the method further comprises the steps of:
determining the similarity between each data group in the time series data of the elastic force parameters and each data group in the time series of the stress parameters, determining the sum of differences between two adjacent data groups in each data group in the time series data of the stiffness parameters, and calculating the spring loss rate corresponding to each data group based on the calculated sum of the similarity and the differences;
the calculation mode of the spring loss rate corresponding to each data set is as follows:
wherein ,spring loss rate representing the j-th data set in the prediction matrix,/>Representation->Distance function->J-th data set of time-series data representing elasticity parameters in the prediction matrix,/th data set of time-series data representing elasticity parameters in the prediction matrix>J-th data set in time series representing stress parameters in the prediction matrix,/th data set in time series representing stress parameters in the prediction matrix>Similarity between the jth data set in the time series data representing the spring force parameter and the jth data set in the time series of the stress parameter, m representing the number of data in each data set,/> and />Time series respectively representing stiffness parameters in prediction matrixThe adjacent (q-1) th data in the j-th data group in the data,a sum of differences between adjacent two data in a j-th data set in the time-series data representing the stiffness parameter;
calculating an ideal life index for each data set in a corresponding use environment based on the environmental fitness corresponding to each data set and the spring loss rate corresponding to each data set, comprising:
calculating an ideal life index under the corresponding use environment of each data group by using the following formula:
wherein ,indicating the ideal life index in the use environment corresponding to the jth data set,/for the data set>Spring loss rate representing the j-th data set in the prediction matrix,/>Indicating the environmental suitability of the j-th data group in the time-series data of the i-th parameter, k indicating the number of parameters;
determining a corresponding lifetime impact level for each data set based on the outlier data points in each data set, comprising:
performing anomaly detection on the time series data of each parameter, and determining an anomaly data point in the time series data of each parameter;
if the j-th data group in the time series data of the current parameter has abnormal data points, determining a life impact index of the j-th data group in the time series data of the current parameter based on the sum of the difference value between each data in the j-th data group in the time series data of the current parameter and each abnormal data point in the j-th data group in the time series data of the current parameter;
adding the life influence indexes of the j-th data set in the time sequence data of all the parameters to obtain the life influence degree corresponding to the j-th data set, and calculating the life influence degree corresponding to each data set according to the life influence degree;
predicting a predicted lifetime in the use environment corresponding to each data set based on the ideal lifetime index and the lifetime influence degree in the use environment corresponding to each data set, comprising:
calculating the ratio between the ideal life index of the jth data set and the life influence degree of the jth data set, and normalizing the calculated ratio;
calculating the product between the standard service life and the normalization result, so as to obtain the predicted service life of the j-th data set under the corresponding service environment;
based on the predicted life and the prediction matrix, constructing life characteristic vectors corresponding to each moment in the simulation process, wherein the life characteristic vectors comprise:
and constructing a life characteristic vector corresponding to each moment in the simulation process based on the elastic parameter corresponding to each moment in the prediction matrix, the rigidity parameter corresponding to each moment and the stress parameter corresponding to each moment.
2. An intelligent predicting system for the life of a spring of a hydraulic spring mounting machine, the system being configured to implement the method of claim 1, comprising:
the parameter acquisition module is used for simulating the springs in the spring hydraulic mounting machine by using a spring simulation model, and acquiring time sequence data of each parameter of the springs in the simulation process so as to obtain a prediction matrix;
the vector determining module is used for determining a life characteristic vector corresponding to each moment in the simulation process based on the prediction matrix;
and the life prediction module is used for predicting the service life of the spring at the current moment based on the life characteristic vector corresponding to each moment in the simulation process by utilizing the neural network model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311047692.0A CN116776502B (en) | 2023-08-21 | 2023-08-21 | Intelligent prediction method and system for service life of spring hydraulic mounting machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311047692.0A CN116776502B (en) | 2023-08-21 | 2023-08-21 | Intelligent prediction method and system for service life of spring hydraulic mounting machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116776502A CN116776502A (en) | 2023-09-19 |
| CN116776502B true CN116776502B (en) | 2023-10-24 |
Family
ID=87993379
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311047692.0A Active CN116776502B (en) | 2023-08-21 | 2023-08-21 | Intelligent prediction method and system for service life of spring hydraulic mounting machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116776502B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110726542A (en) * | 2019-10-28 | 2020-01-24 | 山东泰开高压开关有限公司 | Analysis method for fatigue life of spring |
| CN110889248A (en) * | 2019-11-06 | 2020-03-17 | 江苏科技大学 | Air spring fatigue life prediction platform and prediction method thereof |
| JP2022043906A (en) * | 2020-09-04 | 2022-03-16 | いすゞ自動車株式会社 | Leaf-spring life prediction device for vehicle |
| CN115203832A (en) * | 2022-05-13 | 2022-10-18 | 武汉理工大学 | Method, device, equipment and medium for predicting fatigue life of suspension spring |
| CN116090111A (en) * | 2023-04-10 | 2023-05-09 | 华东交通大学 | Automobile leaf spring fatigue life prediction method based on deep learning model |
-
2023
- 2023-08-21 CN CN202311047692.0A patent/CN116776502B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110726542A (en) * | 2019-10-28 | 2020-01-24 | 山东泰开高压开关有限公司 | Analysis method for fatigue life of spring |
| CN110889248A (en) * | 2019-11-06 | 2020-03-17 | 江苏科技大学 | Air spring fatigue life prediction platform and prediction method thereof |
| JP2022043906A (en) * | 2020-09-04 | 2022-03-16 | いすゞ自動車株式会社 | Leaf-spring life prediction device for vehicle |
| CN115203832A (en) * | 2022-05-13 | 2022-10-18 | 武汉理工大学 | Method, device, equipment and medium for predicting fatigue life of suspension spring |
| CN116090111A (en) * | 2023-04-10 | 2023-05-09 | 华东交通大学 | Automobile leaf spring fatigue life prediction method based on deep learning model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116776502A (en) | 2023-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111222290B (en) | Multi-parameter feature fusion-based method for predicting residual service life of large-scale equipment | |
| KR101919076B1 (en) | Time-series data predicting system | |
| CN108022058B (en) | Wind turbine state reliability assessment method | |
| CN108956111B (en) | Abnormal state detection method and detection system for mechanical part | |
| CN109784528A (en) | Water quality prediction method and device based on time series and support vector regression | |
| WO2016147657A1 (en) | Information processing device, information processing method, and recording medium | |
| CN112668105B (en) | Helicopter transmission shaft abnormity judgment method based on SAE and Mahalanobis distance | |
| CN114266289A (en) | Complex equipment health state assessment method | |
| CN113900894A (en) | Method, system, equipment and medium for predicting large-scale cluster state | |
| CN115358137A (en) | Bearing metal temperature prediction method based on CNN-GRU algorithm | |
| CN117330963B (en) | Energy storage power station fault detection method, system and equipment | |
| CN117112371B (en) | Observable full-link log tracking method and system | |
| CN116776502B (en) | Intelligent prediction method and system for service life of spring hydraulic mounting machine | |
| CN113589172A (en) | Service life estimation method for power grid components | |
| CN107346301B (en) | Water quality monitoring noise data real-time detection method based on double-time-window verification | |
| CN107437112B (en) | A kind of mixing RVM model prediction methods based on the multiple dimensioned kernel function of improvement | |
| CN115795999A (en) | Performance abnormity early warning method for long-term service pumped storage unit | |
| CN115935814A (en) | Transformer fault prediction method based on ARIMA-SVM model | |
| CN108053093A (en) | A kind of k- neighbour's method for diagnosing faults based on the conversion of average influence Value Data | |
| CN110597729B (en) | Pressure testing method, device and system based on dimension | |
| CN114266013A (en) | Deep learning virtual perception network-based transmission system vibration decoupling method | |
| CN112949951A (en) | Data prediction method, data prediction device, electronic equipment and storage medium | |
| CN118091527B (en) | Voltage transformer error assessment method based on interpretive deep learning | |
| CN118445625B (en) | Building structure fault prediction method | |
| CN118312796B (en) | High-precision FFT test optimization method, system and storage medium of tester |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |