CN116123349A - Intelligent upper-mounted axial flow valve - Google Patents

Intelligent upper-mounted axial flow valve Download PDF

Info

Publication number
CN116123349A
CN116123349A CN202211643344.5A CN202211643344A CN116123349A CN 116123349 A CN116123349 A CN 116123349A CN 202211643344 A CN202211643344 A CN 202211643344A CN 116123349 A CN116123349 A CN 116123349A
Authority
CN
China
Prior art keywords
flow velocity
feature
matrix
scale
vector
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.)
Withdrawn
Application number
CN202211643344.5A
Other languages
Chinese (zh)
Inventor
周道亮
胡宝
杨景程
杨选辰
张金扬
赵曾然
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang Diantai Valve Industry Co ltd
Original Assignee
Zhejiang Diantai Valve Industry Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhejiang Diantai Valve Industry Co ltd filed Critical Zhejiang Diantai Valve Industry Co ltd
Priority to CN202211643344.5A priority Critical patent/CN116123349A/en
Publication of CN116123349A publication Critical patent/CN116123349A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M13/00—Testing of machine parts
    • G01M13/003—Machine valves
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16K—VALVES; TAPS; COCKS; ACTUATING-FLOATS; DEVICES FOR VENTING OR AERATING
    • F16K37/00—Special means in or on valves or other cut-off apparatus for indicating or recording operation thereof, or for enabling an alarm to be given
    • F16K37/0025—Electrical or magnetic means
    • F16K37/005—Electrical or magnetic means for measuring fluid parameters

Landscapes

  • Engineering & Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Mechanical Engineering (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)

Abstract

The utility model relates to an intelligent monitoring field, it specifically discloses an intelligent top-loading axial flow valve, it is through adopting the artificial intelligence detection technique based on degree of depth study to utilize the transition matrix between the velocity of flow characteristic of entry and export to simulate the reasonable operating mode of top-loading axial flow valve in high dimension space, and vibration characteristic then represents the actual operating mode of top-loading axial flow valve, thereby represent the differentiation expression between actual operating mode and the reasonable operating mode with the difference between the two, and obtain the testing result of the operating condition of top-loading axial flow valve through the classifier. Therefore, the working state of the upper-mounted axial flow valve can be accurately detected based on actual conditions, so that an abnormal early warning prompt is generated when the working state is abnormal, and the normal operation of the upper-mounted axial flow valve is ensured.

Description

Intelligent upper-mounted axial flow valve
Technical Field
The present application relates to the field of intelligent monitoring, and more particularly, to an intelligent top-loading axial flow valve.
Background
The axial flow type regulating valve is widely applied to the whole process of oil gas production: the method is suitable for production, treatment, transportation, storage and distribution of fluid medium natural gas, liquefied gas, petroleum and liquefied chemical products. The control principle of the axial flow regulator valve is non-mechanical. The rubber collar serves as a control element, extending the auxiliary collar. It relies on differential pressure or a constrictive reaction acting on both ends to achieve the purpose of controlling pressure or flow. The axial flow type regulating valve adopts a non-mechanical control mode, so that the axial flow valve is not mechanically connected with the control element. The axial flow valve has compact structure, light weight, unique clamp design, easy installation and convenient use.
The cavity of the axial flow type regulating valve body is designed into a streamline cone coaxial with the pipeline. As the fluid passes, it disperses and bypasses the cone, creating a concentric flow through the valve body and then through the sleeve throttle to the downstream conduit. The valve core moves coaxially in the sleeve to regulate the flow. Due to the flow passage characteristic of the axial flow type regulating valve, the valve cavity is uniformly stressed, and vibration caused by uneven force is controlled. And the medium flow direction of the axial flow type regulating valve is single, and the inlet and outlet directions cannot be changed. The valve core is provided with a pressure balance hole by adopting soft and hard bidirectional sealing, and a plurality of evenly distributed orifices are arranged on the sleeve, which is similar to a traditional sleeve valve and a low noise valve.
In the working process of the existing axial flow type regulating valve, a vortex phenomenon is generated in an upper cavity when a medium flows through a valve body, so that radial vibration of a valve core is caused, cavitation phenomenon is generated, and flushing is caused on valve internal parts, thereby influencing the service life of the valve and the regulating performance of the valve.
Therefore, an intelligent top-loading axial flow valve is desired, which can perform self-monitoring on the performance of the axial flow valve so as to send out an early warning prompt when the abnormal performance is detected, thereby ensuring the service life and the performance of the valve.
Disclosure of Invention
The present application has been made in order to solve the above technical problems. The embodiment of the application provides an intelligent top-loading axial flow valve, which simulates reasonable working conditions of the top-loading axial flow valve by adopting an artificial intelligent detection technology based on deep learning and utilizing a transfer matrix between flow velocity characteristics of an inlet and an outlet in a high-dimensional space, wherein vibration characteristics represent actual working conditions of the top-loading axial flow valve, so that differential expression between the actual working conditions and the reasonable working conditions is represented by the difference between the two, and a detection result of the working state of the top-loading axial flow valve is obtained through a classifier. Therefore, the working state of the upper-mounted axial flow valve can be accurately detected based on actual conditions, so that an abnormal early warning prompt is generated when the working state is abnormal, and the normal operation of the upper-mounted axial flow valve is ensured.
According to one aspect of the present application, there is provided an intelligent top-loading axial flow valve comprising: the sensor monitoring module is used for acquiring fluid flow velocity values of inlets at a plurality of preset time points in a preset time period, fluid flow velocity values of outlets of the preset time period and vibration signals of the to-be-monitored upper-mounted axial flow valve in the preset time period; a sensor data structuring module, configured to arrange the fluid flow rate values of the inlet and the fluid flow rate values of the outlet at the plurality of predetermined time points into an inlet flow rate input vector and an outlet flow rate input vector according to a time dimension, respectively; the flow velocity feature extraction module is used for respectively passing the inlet flow velocity input vector and the outlet flow velocity input vector through the multi-scale neighborhood feature extraction module to obtain an inlet flow velocity feature vector and an outlet flow velocity feature vector; a global transfer module for calculating a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector; the vibration signal characteristic extraction module is used for obtaining a vibration characteristic matrix through a convolution neural network model using a spatial attention mechanism according to a waveform diagram of the vibration signal of the to-be-monitored upper-loading axial flow valve in the preset time period; the differential module is used for calculating a differential characteristic matrix between the vibration characteristic matrix and the transfer matrix; the small-scale feature optimization module is used for correcting the differential feature matrix based on a small-scale feature association mode between the vibration feature matrix and the transfer matrix to obtain a corrected differential feature matrix; and the intelligent monitoring result generation module is used for enabling the corrected differential feature matrix to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the working state of the upper-mounted axial flow valve to be monitored is normal or not.
In the above-mentioned intelligent up-loading axial flow valve, the multiscale neighborhood feature extraction module includes: the device comprises a first convolution layer, a second convolution layer parallel to the first convolution layer and a cascade layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first scale, and the second convolution kernel uses a one-dimensional convolution kernel with a second scale.
In the above-mentioned intelligent facial make-up axial flow valve, the velocity of flow characteristic draws the module, includes: a first scale feature extraction unit, configured to input the inlet flow velocity input vector and the outlet flow velocity input vector into a first convolution layer of the multi-scale neighborhood feature extraction module, respectively, to obtain a first neighborhood scale inlet flow velocity feature vector and a first neighborhood scale outlet flow velocity feature vector, where the first convolution layer has a first one-dimensional convolution kernel with a first length; a second scale feature extraction unit, configured to input the inlet flow velocity input vector and the outlet flow velocity input vector into a second convolution layer of the multi-scale neighborhood feature extraction module, respectively, to obtain a second neighborhood scale inlet flow velocity feature vector and a second neighborhood scale outlet flow velocity feature vector, where the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and the multiscale fusion unit is used for cascading the first neighborhood scale inlet flow velocity characteristic vector and the first neighborhood scale outlet flow velocity characteristic vector with the second neighborhood scale inlet flow velocity characteristic vector and the second neighborhood scale outlet flow velocity characteristic vector respectively to obtain the inlet flow velocity characteristic vector and the outlet flow velocity characteristic vector.
In the above-mentioned intelligent top-loading axial flow valve, the first scale feature extraction unit is further configured to: performing one-dimensional convolution encoding on the inlet flow velocity input vector and the outlet flow velocity input vector by using a first convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a first neighborhood scale inlet flow velocity feature vector and a first neighborhood scale outlet flow velocity feature vector; wherein, the formula is:
Figure BDA0004008669160000031
wherein a is the width of the first convolution kernel in the X direction, F (a) is a first convolution kernel parameter vector, G (X-a) is a local vector matrix calculated by a convolution kernel function, w is the size of the first convolution kernel, and X represents the inlet flow velocity input vector and the outlet flow velocity input vector; the second scale feature extraction unit is further configured to: performing one-dimensional convolution encoding on the inlet flow velocity input vector and the outlet flow velocity input vector by using a second convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a second neighborhood scale inlet flow velocity feature vector and a second neighborhood scale outlet flow velocity feature vector; wherein, the formula is:
Figure BDA0004008669160000032
wherein b is the width of the second convolution kernel in the X direction, F (b) is a second convolution kernel parameter vector, G (X-b) is a local vector matrix that operates with a convolution kernel function, m is the size of the second convolution kernel, and X represents the inlet flow velocity input vector and the outlet flow velocity input vector.
In the above intelligent top-loading axial flow valve, the global transfer module is further configured to: calculating a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector with the following formula; wherein, the formula is:
Figure BDA0004008669160000033
wherein V is c Representing the inlet flow velocity characteristic vector, V representing the outlet flow velocity characteristic vector, M 2 Representing the transfer matrix.
In the above-mentioned intelligent upper-mounting axial flow valve, the vibration signal feature extraction module is further configured to: respectively carrying out convolution processing, pooling processing along a channel dimension and nonlinear activation processing on a waveform diagram of a vibration signal of the to-be-monitored upper-loading axial flow valve in the preset time period in forward transmission of layers by using each layer of the convolutional neural network model so as to output a plurality of initial vibration feature matrixes by the last layer of the convolutional neural network model; and inputting the plurality of initial vibration feature matrices into a spatial attention layer of the convolutional neural network model to obtain the vibration feature matrix.
In the above-mentioned intelligent top-loading axial flow valve, the differential module is further configured to: calculating a differential feature matrix between the vibration feature matrix and the transfer matrix according to the following formula; wherein, the formula is:
Figure BDA0004008669160000041
Wherein M is 1 Representing the vibration characteristic matrix, M 2 Representing the transfer matrix, M c Representing the differential feature matrix,>
Figure BDA0004008669160000042
representing the difference by location.
In the above-mentioned intelligent top-loading axial flow valve, the small scale feature optimization module includes: the small-scale feature association mode extraction unit is used for calculating a small-scale local derivative matrix of the vibration feature matrix and the transfer matrix as a weighted feature matrix, wherein the formula is as follows:
Figure BDA0004008669160000043
wherein the method comprises the steps of
Figure BDA0004008669160000044
And->
Figure BDA0004008669160000045
The vibration characteristic matrix, the transfer matrix and the vibration characteristic matrix are respectivelyVector values for the (i, j) th position of the small-scale local derivative matrix; and the weighting action unit is used for carrying out dot multiplication on the differential feature matrix by taking the small-scale local derivative matrix as a weighting matrix to carry out feature value weighting so as to obtain a corrected differential feature matrix.
In the above-mentioned intelligent upper-mounting axial flow valve, the intelligent monitoring result generation module includes: the expansion unit is used for expanding the corrected differential feature matrix into classification feature vectors based on row vectors or column vectors; the full-connection coding unit is used for carrying out full-connection coding on the classification characteristic vectors by using a plurality of full-connection layers of the classifier so as to obtain coded classification characteristic vectors; and the classification result generating unit is used for enabling the coding classification feature vector to pass through a Softmax classification function of the classifier to obtain the classification result.
According to another aspect of the present application, there is provided a method of using an intelligent top-loading axial flow valve, comprising: acquiring fluid flow velocity values of inlets at a plurality of preset time points in a preset time period, fluid flow velocity values of outlets of the preset time period, and vibration signals of the to-be-monitored upper-mounted axial flow valve in the preset time period; arranging the fluid flow rate values of the inlets and the fluid flow rate values of the outlets at a plurality of preset time points into an inlet flow rate input vector and an outlet flow rate input vector according to the time dimension respectively; respectively passing the inlet flow velocity input vector and the outlet flow velocity input vector through a multi-scale neighborhood feature extraction module to obtain an inlet flow velocity feature vector and an outlet flow velocity feature vector; calculating a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector; the waveform diagram of the vibration signal of the to-be-monitored upper-mounted axial flow valve in the preset time period is processed through a convolutional neural network model using a spatial attention mechanism to obtain a vibration characteristic matrix; calculating a differential feature matrix between the vibration feature matrix and the transfer matrix; correcting the differential feature matrix based on a small-scale feature association mode between the vibration feature matrix and the transfer matrix to obtain a corrected differential feature matrix; and passing the corrected differential feature matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the working state of the upper-mounted axial flow valve to be monitored is normal or not.
Compared with the prior art, the intelligent top-loading axial flow valve provided by the application has the advantages that the artificial intelligent detection technology based on deep learning is adopted, so that the reasonable working condition of the top-loading axial flow valve is simulated by utilizing the transfer matrix between the flow velocity characteristics of the inlet and the outlet in a high-dimensional space, the vibration characteristics represent the actual working condition of the top-loading axial flow valve, the difference between the two is used for representing the differential expression between the actual working condition and the reasonable working condition, and the detection result of the working state of the top-loading axial flow valve is obtained through the classifier. Therefore, the working state of the upper-mounted axial flow valve can be accurately detected based on actual conditions, so that an abnormal early warning prompt is generated when the working state is abnormal, and the normal operation of the upper-mounted axial flow valve is ensured.
Drawings
The foregoing and other objects, features and advantages of the present application will become more apparent from the following more particular description of embodiments of the present application, as illustrated in the accompanying drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.
FIG. 1 is an application scenario diagram of an intelligent top-loading axial flow valve according to an embodiment of the present application;
FIG. 2 is a block diagram of an intelligent top-loading axial flow valve according to an embodiment of the present application;
FIG. 3 is a system architecture diagram of an intelligent top-loading axial flow valve according to an embodiment of the present application;
FIG. 4 is a block diagram of a flow rate feature extraction module in an intelligent, top-loading axial flow valve according to an embodiment of the present application;
FIG. 5 is a flow chart of vibration signal feature extraction in an intelligent, top-loading axial flow valve according to an embodiment of the present application;
FIG. 6 is a block diagram of a small-scale feature optimization module in an intelligent top-loading axial flow valve according to an embodiment of the present application;
FIG. 7 is a block diagram of an intelligent monitoring result generation module in an intelligent top-loading axial flow valve according to an embodiment of the present application;
fig. 8 is a flow chart of a method of using an intelligent top-loading axial flow valve according to an embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application and not all of the embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Summary of the application
As described in the background art, in the working process of the existing axial flow type regulating valve, a vortex phenomenon is generated in an upper cavity when a medium flows through a valve body, so that radial vibration of a valve core is caused, cavitation phenomenon is generated, and flushing is caused on valve internal parts, thereby influencing the service life of the valve and the regulating performance of the valve. Therefore, an intelligent top-loading axial flow valve is desired, which can perform self-monitoring on the performance of the axial flow valve so as to send out an early warning prompt when the abnormal performance is detected, thereby ensuring the service life and the performance of the valve.
Accordingly, considering that the valve chamber is subjected to uniform force due to the flow passage characteristics of the axial flow type regulating valve, vibration caused by the non-uniform force is controlled. Therefore, in the technical scheme of the application, in the process of actually monitoring the working state of the upper-loading axial flow valve, the reasonable working condition of the upper-loading axial flow valve can be simulated through the correlation information between the fluid flow velocity value of the inlet and the fluid flow velocity value of the outlet, the actual working condition of the upper-loading axial flow valve is simulated based on the vibration information of the upper-loading axial flow valve, the performance of the upper-loading axial flow valve is represented by the difference between the two conditions, and the working state of the upper-loading axial flow valve is detected. However, in the actual detection process, it is found that since the vibration signal of the upper-loading axial flow valve is easily interfered by other vibration signals, it is difficult to perform collection and separation, and in the process of detecting the operating state of the upper-loading axial flow valve, there are difficulties in how to establish the association relationship between the flow rate of the inlet fluid and the flow rate of the outlet fluid, and how to perform the operating state detection of the upper-loading axial flow valve based on the difference information between the reasonable operating condition and the actual operating condition.
In recent years, deep learning and neural networks have been widely used in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks have also shown levels approaching and even exceeding humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
Deep learning and the development of neural networks provide new solutions and solutions for mining complex mappings between the flow rate of inlet fluid and the flow rate of outlet fluid. Those of ordinary skill in the art will appreciate that deep learning based deep neural network models may be adapted with appropriate training strategies, such as by gradient descent back-propagation algorithms, to adjust parameters of the deep neural network model to enable modeling of complex nonlinear correlations between things, which is obviously suitable for modeling and establishing complex mappings between inlet fluid flow rates and outlet fluid flow rates.
Specifically, in the technical scheme of the application, an artificial intelligent detection technology based on deep learning is adopted to simulate the reasonable working condition of the upper-loading axial flow valve by utilizing a transfer matrix between flow velocity characteristics of an inlet and an outlet in a high-dimensional space, and the vibration characteristics represent the actual working condition of the upper-loading axial flow valve, so that the difference between the two is used for representing the differential expression between the actual working condition and the reasonable working condition, and the detection result of the working state of the upper-loading axial flow valve is obtained through a classifier. Therefore, the working state of the upper-mounted axial flow valve can be accurately detected based on actual conditions, so that an abnormal early warning prompt is generated when the working state is abnormal, and the normal operation of the upper-mounted axial flow valve is ensured.
More specifically, in the technical solution of the present application, first, fluid flow velocity values of inlets at a plurality of predetermined time points within a predetermined period of time and fluid flow velocity values of outlets of the predetermined period of time are obtained, and vibration signals of the upper-loading type axial flow valve to be monitored in the predetermined period of time are obtained. Then, in order to extract reasonable working condition characteristics of the upper-mounted axial flow valve, correlation characteristic distribution information between the inlet flow rate and the outlet flow rate needs to be extracted. That is, first, it is necessary to arrange the fluid flow rate values of the inlet and the fluid flow rate values of the outlet at the plurality of predetermined time points into an inlet flow rate input vector and an outlet flow rate input vector, respectively, in a time dimension to integrate distribution information of the fluid flow rate values of the inlet and the fluid flow rate values of the outlet, respectively, in time sequence.
Then, considering that the fluid flow velocity value of the inlet and the fluid flow velocity value of the outlet have volatility and uncertainty in the time dimension, in order to explore the time sequence dynamic change characteristic information, the inlet flow velocity input vector and the outlet flow velocity input vector are further subjected to characteristic mining through a multi-scale neighborhood characteristic extraction module respectively so as to extract dynamic multi-scale neighborhood associated characteristics of the fluid flow velocity of the inlet and the fluid flow velocity of the outlet under different time spans in the preset time period respectively, and thus an inlet flow velocity characteristic vector and an outlet flow velocity characteristic vector are obtained.
Further, a transfer matrix of the inlet flow velocity feature vector relative to the outlet flow velocity feature vector is recalculated to represent a correlation feature between the fluid flow velocity dynamic feature of the inlet and the fluid flow velocity dynamic feature of the outlet. That is, the transfer matrix is used to simulate the reasonable working condition of the upper-mounted axial flow valve, so that the dynamic characteristic information of the reasonable working condition of the upper-mounted axial flow valve is dug.
It should be understood that, since the vibration signal of the to-be-monitored upper-loading type axial flow valve has a waveform pattern in a time domain, feature mining of the waveform pattern of the vibration signal of the to-be-monitored upper-loading type axial flow valve is performed using a convolutional neural network model having excellent performance in terms of implicit feature extraction of images. In particular, in the process of detecting the working state of the upper-mounted axial flow valve, the detection accuracy of the working state of the upper-mounted axial flow valve is low due to the fact that more interference of other external environment factors and other internal vibration factors exists. Therefore, in the technical scheme of the application, when the feature extraction of the vibration signal is performed, the vibration signal feature information of the upper-mounted axial flow valve needs to be focused, that is, specifically, the waveform diagram of the vibration signal of the upper-mounted axial flow valve to be monitored in the preset time period is processed in a convolutional neural network model using a spatial attention mechanism, so that the vibration feature distribution information of the vibration signal, which is focused on the upper-mounted axial flow valve to be monitored in space in the waveform diagram of the vibration signal, is extracted, and the actual working condition feature of the upper-mounted axial flow valve is represented, so that the vibration feature matrix is obtained.
Then, the reasonable working condition of the upper-mounted axial flow valve is simulated by taking into consideration a transfer matrix between the dynamic characteristic of the fluid flow rate of the inlet and the dynamic characteristic of the fluid flow rate of the outlet, and the vibration characteristic of the upper-mounted axial flow valve represents the actual working condition of the upper-mounted axial flow valve, so in the technical scheme of the application, the differential characteristic matrix between the vibration characteristic matrix and the transfer matrix is further calculated, differential expression between the actual working condition and the reasonable working condition of the upper-mounted axial flow valve, namely differential characteristic distribution between the two is dug, and classification processing is carried out in a classifier by taking the differential characteristic distribution as a classification characteristic matrix, so that a classification result for representing whether the working condition of the upper-mounted axial flow valve to be monitored is normal or not is obtained. Therefore, the working state of the upper-mounted axial flow valve can be detected based on actual conditions, and further, an abnormality early warning prompt is generated when the working state is abnormal.
In particular, in the technical solution of the present application, when calculating the differential feature matrix between the vibration feature matrix and the transfer matrix, the convolution neural network model using the spatial attention mechanism can strengthen the local feature expression inside the vibration feature matrix, and the transfer matrix is focused on expressing the global distribution transfer feature of the inlet flow velocity feature vector relative to the outlet flow velocity feature vector, so that the small-scale feature association expression of the vibration feature matrix and the transfer matrix needs to be improved, and then the expression effect of the differential feature matrix on the position-by-position association differential feature between the vibration feature matrix and the transfer matrix is improved.
Thus, the vibration characteristic matrix is calculated, e.g. denoted as M 1 And the transfer matrix, e.g. denoted as M 2 As a weighted feature matrix, expressed as:
Figure BDA0004008669160000081
wherein the method comprises the steps of
Figure BDA0004008669160000082
And->
Figure BDA0004008669160000083
Vector values at the (i, j) th positions of the vibration feature matrix, the transfer matrix, and the small-scale local derivative matrix, respectively. Here, by calculating the vibration characteristic matrix M 1 And the transfer matrix M 2 Small-scale local derivative features in between can be based on the vibration feature matrix M 1 And the transfer matrix M 2 The geometrical approximation of the corresponding positions in between mimics the physics of the mutual expression between data sequences, thereby enhancing the local nonlinear dependence of the cross-feature-domain positions with a point-by-point regression of the feature matrices by position. Thus, by locally deriving the matrix M at said small scale w And performing point multiplication on the differential feature matrix as a weighting matrix to perform feature value weighting, so that the expression effect of the differential feature matrix on the vibration feature matrix and the small-scale position-by-position correlation differential feature of the transfer matrix can be improved, and the accuracy of the classification result of the differential feature matrix is improved. Therefore, the working state of the upper-mounted axial flow valve can be accurately detected based on actual conditions, so that an abnormal early warning prompt is generated when the working state is abnormal, and the normal operation of the upper-mounted axial flow valve is ensured.
Based on this, this application proposes an intelligent facial make-up formula axial flow valve, it includes: the sensor monitoring module is used for acquiring fluid flow velocity values of inlets at a plurality of preset time points in a preset time period, fluid flow velocity values of outlets of the preset time period and vibration signals of the to-be-monitored upper-mounted axial flow valve in the preset time period; a sensor data structuring module, configured to arrange the fluid flow rate values of the inlet and the fluid flow rate values of the outlet at the plurality of predetermined time points into an inlet flow rate input vector and an outlet flow rate input vector according to a time dimension, respectively; the flow velocity feature extraction module is used for respectively passing the inlet flow velocity input vector and the outlet flow velocity input vector through the multi-scale neighborhood feature extraction module to obtain an inlet flow velocity feature vector and an outlet flow velocity feature vector; a global transfer module for calculating a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector; the vibration signal characteristic extraction module is used for obtaining a vibration characteristic matrix through a convolution neural network model using a spatial attention mechanism according to a waveform diagram of the vibration signal of the to-be-monitored upper-loading axial flow valve in the preset time period; the differential module is used for calculating a differential characteristic matrix between the vibration characteristic matrix and the transfer matrix; the small-scale feature optimization module is used for correcting the differential feature matrix based on a small-scale feature association mode between the vibration feature matrix and the transfer matrix to obtain a corrected differential feature matrix; and the intelligent monitoring result generation module is used for enabling the corrected differential feature matrix to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the working state of the upper-mounted axial flow valve to be monitored is normal or not.
Fig. 1 is an application scenario diagram of an intelligent top-loading axial flow valve according to an embodiment of the present application. As shown in fig. 1, in this application scenario, fluid flow rate values (e.g., D1 as shown in fig. 1) of inlets at a plurality of predetermined time points within a predetermined period of time and fluid flow rate values (e.g., D2 as shown in fig. 1) of outlets of the predetermined period of time are acquired by a flow rate sensor (e.g., S1 as shown in fig. 1), and vibration signals of the on-package axial flow valve to be monitored of the predetermined period of time are acquired by a vibration signal sensor (e.g., V as shown in fig. 1). The information is then input to a server (e.g., S2 in fig. 1) deployed with an intelligent on-board axial flow valve algorithm, where the server is capable of processing the input information with the intelligent on-board axial flow valve algorithm to generate a classification result that indicates whether the operating state of the on-board axial flow valve to be monitored is normal.
Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.
Exemplary System
Fig. 2 is a block diagram of an intelligent top-loading axial flow valve according to an embodiment of the present application. As shown in fig. 2, an intelligent top-loading axial flow valve 300 according to an embodiment of the present application includes: a sensor monitoring module 310; a sensor data structuring module 320; a flow rate feature extraction module 330; a global transfer module 340; a vibration signal feature extraction module 350; a differential module 360; a small scale feature optimization module 370; and an intelligent monitoring result generation module 380.
The sensor monitoring module 310 is configured to obtain fluid flow rate values of inlets at a plurality of predetermined time points within a predetermined time period, fluid flow rate values of outlets of the predetermined time period, and vibration signals of the to-be-monitored top-loading axial flow valve within the predetermined time period; the sensor data structuring module 320 is configured to arrange the fluid flow rate values of the inlet and the fluid flow rate values of the outlet at the plurality of predetermined time points into an inlet flow rate input vector and an outlet flow rate input vector according to a time dimension, respectively; the flow velocity feature extraction module 330 is configured to pass the inlet flow velocity input vector and the outlet flow velocity input vector through a multi-scale neighborhood feature extraction module to obtain an inlet flow velocity feature vector and an outlet flow velocity feature vector; the global transfer module 340 is configured to calculate a transfer matrix of the inlet flow velocity feature vector relative to the outlet flow velocity feature vector; the vibration signal feature extraction module 350 is configured to obtain a vibration feature matrix from a waveform diagram of a vibration signal of the to-be-monitored upper-loading axial flow valve in the predetermined period of time by using a convolutional neural network model of a spatial attention mechanism; the difference module 360 is configured to calculate a difference feature matrix between the vibration feature matrix and the transfer matrix; the small-scale feature optimization module 370 is configured to correct the differential feature matrix based on a small-scale feature association pattern between the vibration feature matrix and the transfer matrix to obtain a corrected differential feature matrix; and the intelligent monitoring result generating module 380 is configured to pass the corrected differential feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate whether the working state of the upper-loading axial flow valve to be monitored is normal.
Fig. 3 is a system architecture diagram of an intelligent top-loading axial flow valve according to an embodiment of the present application. As shown in fig. 3, in the network architecture, first, the sensor monitoring module 310 obtains fluid flow velocity values of inlets at a plurality of preset time points in a preset time period and fluid flow velocity values of outlets at the preset time period, and vibration signals of the to-be-monitored top-loading axial flow valve in the preset time period; the sensor data structuring module 320 arranges the fluid flow rate values of the inlet and the fluid flow rate values of the outlet at a plurality of predetermined time points acquired by the sensor monitoring module 310 into an inlet flow rate input vector and an outlet flow rate input vector according to a time dimension, respectively; next, the flow velocity feature extraction module 330 respectively passes the inlet flow velocity input vector and the outlet flow velocity input vector obtained by the sensor data structuring module 320 through a multi-scale neighborhood feature extraction module to obtain an inlet flow velocity feature vector and an outlet flow velocity feature vector; the global transfer module 340 calculates a transfer matrix of the inlet flow velocity feature vector obtained by the flow velocity feature extraction module 330 with respect to the outlet flow velocity feature vector; then, the vibration signal feature extraction module 350 obtains a vibration feature matrix from the waveform of the vibration signal of the to-be-monitored upper-loading axial flow valve in the predetermined period acquired by the sensor monitoring module 310 through a convolutional neural network model using a spatial attention mechanism; the difference module 360 calculates a difference feature matrix between the vibration feature matrix obtained by the vibration signal feature extraction module 350 and the transfer matrix obtained by the global transfer module 340; the small-scale feature optimization module 370 corrects the differential feature matrix based on a small-scale feature association pattern between the vibration feature matrix and the transfer matrix to obtain a corrected differential feature matrix; furthermore, the intelligent monitoring result generating module 380 passes the corrected differential feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate whether the working state of the upper-loading axial flow valve to be monitored is normal.
Specifically, during operation of the intelligent on-board axial flow valve 300, the sensor monitoring module 310 is configured to obtain fluid flow rate values of the inlet at a plurality of predetermined time points within a predetermined time period and fluid flow rate values of the outlet at the predetermined time period, and a vibration signal of the on-board axial flow valve to be monitored within the predetermined time period. The valve cavity is uniformly stressed to control vibration caused by uneven force in consideration of the flow passage characteristic of the axial flow type regulating valve. Therefore, in the technical scheme of the application, in the process of actually monitoring the working state of the upper-loading axial flow valve, the reasonable working condition of the upper-loading axial flow valve can be simulated through the correlation information between the fluid flow velocity value of the inlet and the fluid flow velocity value of the outlet, the actual working condition of the upper-loading axial flow valve is simulated based on the vibration information of the upper-loading axial flow valve, the performance of the upper-loading axial flow valve is represented by the difference between the two conditions, and the working state of the upper-loading axial flow valve is detected. Therefore, in the technical scheme of the application, an artificial intelligent detection technology based on deep learning is adopted to simulate the reasonable working condition of the upper-mounted axial flow valve by utilizing a transfer matrix between flow velocity characteristics of an inlet and an outlet in a high-dimensional space, and the vibration characteristics represent the actual working condition of the upper-mounted axial flow valve, so that the difference between the two is used for representing the differential expression between the actual working condition and the reasonable working condition, and the detection result of the working condition of the upper-mounted axial flow valve is obtained through a classifier. In a specific example of the present application, the fluid flow rate values of the inlet of a plurality of predetermined time points within a predetermined period of time and the fluid flow rate values of the outlet of the predetermined period of time may be acquired by the flow rate sensor, and the vibration signal of the on-package axial flow valve to be monitored for the predetermined period of time may be acquired by the vibration signal sensor.
Specifically, during operation of the intelligent on-board axial flow valve 300, the sensor data structuring module 320 is configured to arrange the fluid flow rate values of the inlet and the fluid flow rate values of the outlet at the plurality of predetermined time points into an inlet flow rate input vector and an outlet flow rate input vector according to a time dimension, respectively. In order to extract reasonable working condition characteristics of the upper-mounted axial flow valve, correlation characteristic distribution information between the inlet flow rate and the outlet flow rate needs to be extracted. That is, first, it is necessary to arrange the fluid flow rate values of the inlet and the fluid flow rate values of the outlet at the plurality of predetermined time points into an inlet flow rate input vector and an outlet flow rate input vector, respectively, in a time dimension to integrate distribution information of the fluid flow rate values of the inlet and the fluid flow rate values of the outlet, respectively, in time sequence.
Specifically, during the operation of the intelligent top-loading axial flow valve 300, the flow velocity feature extraction module 330 is configured to pass the inlet flow velocity input vector and the outlet flow velocity input vector through a multi-scale neighborhood feature extraction module to obtain an inlet flow velocity feature vector and an outlet flow velocity feature vector. In view of the fact that the fluid flow velocity value of the inlet and the fluid flow velocity value of the outlet have volatility and uncertainty in the time dimension, in order to explore the time sequence dynamic change characteristic information, the inlet flow velocity input vector and the outlet flow velocity input vector are further subjected to characteristic mining through a multi-scale neighborhood characteristic extraction module respectively, so that dynamic multi-scale neighborhood associated characteristics of the fluid flow velocity of the inlet and the fluid flow velocity of the outlet in different time spans in the preset time period are extracted respectively, and an inlet flow velocity characteristic vector and an outlet flow velocity characteristic vector are obtained. Wherein, the multiscale neighborhood feature extraction module comprises: the device comprises a first convolution layer, a second convolution layer parallel to the first convolution layer and a cascade layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first scale, and the second convolution kernel uses a one-dimensional convolution kernel with a second scale.
Fig. 4 is a block diagram of a flow rate feature extraction module in an intelligent, top-loading axial flow valve according to an embodiment of the present application. As shown in fig. 4, the flow rate feature extraction module 330 includes: a first scale feature extraction unit 331, configured to input the inlet flow velocity input vector and the outlet flow velocity input vector into a first convolution layer of the multi-scale neighborhood feature extraction module, respectively, to obtain a first neighborhood scale inlet flow velocity feature vector and a first neighborhood scale outlet flow velocity feature vector, where the first convolution layer has a first one-dimensional convolution kernel with a first length; a second scale feature extraction unit 332, configured to input the inlet flow velocity input vector and the outlet flow velocity input vector into a second convolution layer of the multi-scale neighborhood feature extraction module, respectively, to obtain a second neighborhood scale inlet flow velocity feature vector and a second neighborhood scale outlet flow velocity feature vector, where the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and a multi-scale fusion unit 333, configured to concatenate the first neighborhood scale inlet flow velocity feature vector and the first neighborhood scale outlet flow velocity feature vector with the second neighborhood scale inlet flow velocity feature vector and the second neighborhood scale outlet flow velocity feature vector, respectively, to obtain the inlet flow velocity feature vector and the outlet flow velocity feature vector. More specifically, the first scale-feature extraction unit 331 is further configured to: performing one-dimensional convolution encoding on the inlet flow velocity input vector and the outlet flow velocity input vector by using a first convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a first neighborhood scale inlet flow velocity feature vector and a first neighborhood scale outlet flow velocity feature vector; wherein, the formula is:
Figure BDA0004008669160000131
Wherein a is the width of the first convolution kernel in the X direction, F (a) is a first convolution kernel parameter vector, G (X-a) is a local vector matrix calculated by a convolution kernel function, w is the size of the first convolution kernel, and X represents the inlet flow velocity input vector and the outlet flow velocity input vector; the second scale feature extraction unit 332 is further configured to: performing one-dimensional convolution encoding on the inlet flow velocity input vector and the outlet flow velocity input vector by using a second convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a second neighborhood scale inlet flow velocity feature vector and a second neighborhood scale outlet flow velocity feature vector; wherein, the formula is:
Figure BDA0004008669160000132
wherein b is the width of the second convolution kernel in the X direction, F (b) is a second convolution kernel parameter vector, G (X-b) is a local vector matrix that operates with a convolution kernel function, m is the size of the second convolution kernel, and X represents the inlet flow velocity input vector and the outlet flow velocity input vector.
Specifically, during operation of the intelligent on-board axial flow valve 300, the global transfer module 340 is configured to calculate a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector. And calculating a transfer matrix of the inlet flow velocity characteristic vector relative to the outlet flow velocity characteristic vector to represent the correlation characteristic between the fluid flow velocity dynamic characteristic of the inlet and the fluid flow velocity dynamic characteristic of the outlet. That is, the transfer matrix is used to simulate the reasonable working condition of the upper-mounted axial flow valve, so that the dynamic characteristic information of the reasonable working condition of the upper-mounted axial flow valve is dug. In one specific example of the present application, a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector is calculated with the following formula; wherein, the formula is:
Figure BDA0004008669160000133
Wherein V is c Representing the inlet flow velocity characteristic vector, V representing the outlet flow velocity characteristic vector, M 2 Representing the transfer matrix.
Specifically, during the operation of the intelligent on-board axial flow valve 300, the vibration signal feature extraction module 350 is configured to obtain a vibration feature matrix from a waveform diagram of the vibration signal of the on-board axial flow valve to be monitored for the predetermined period of time by using a convolutional neural network model of a spatial attention mechanism. It should be understood that, since the vibration signal of the to-be-monitored upper-loading type axial flow valve has a waveform pattern in a time domain, feature mining of the waveform pattern of the vibration signal of the to-be-monitored upper-loading type axial flow valve is performed using a convolutional neural network model having excellent performance in terms of implicit feature extraction of images. In particular, in the process of detecting the working state of the upper-mounted axial flow valve, the detection accuracy of the working state of the upper-mounted axial flow valve is low due to the fact that more interference of other external environment factors and other internal vibration factors exists. Therefore, in the technical scheme of the application, when the feature extraction of the vibration signal is performed, the vibration signal feature information of the upper-mounted axial flow valve needs to be focused, that is, specifically, the waveform diagram of the vibration signal of the upper-mounted axial flow valve to be monitored in the preset time period is processed in a convolutional neural network model using a spatial attention mechanism, so that the vibration feature distribution information of the vibration signal, which is focused on the upper-mounted axial flow valve to be monitored in space in the waveform diagram of the vibration signal, is extracted, and the actual working condition feature of the upper-mounted axial flow valve is represented, so that the vibration feature matrix is obtained.
Fig. 5 is a flow chart of vibration signal feature extraction in an intelligent, top-loading axial flow valve according to an embodiment of the present application. As shown in fig. 5, in the vibration signal feature extraction process, it includes: s210, performing convolution processing, pooling processing along a channel dimension and nonlinear activation processing on a waveform diagram of a vibration signal of the to-be-monitored upper-loading axial flow valve in the preset time period in forward transmission of layers by using each layer of the convolutional neural network model so as to output a plurality of initial vibration feature matrixes by the last layer of the convolutional neural network model; and S220, inputting the plurality of initial vibration feature matrices into a spatial attention layer of the convolutional neural network model to obtain the vibration feature matrix.
Specifically, during operation of the intelligent, self-contained axial flow valve 300, the difference module 360 is configured to calculate a difference signature matrix between the vibration signature matrix and the transfer matrix. The reasonable working condition of the upper-mounted axial flow valve is simulated by taking the transfer matrix between the dynamic characteristic of the fluid flow rate of the inlet and the dynamic characteristic of the fluid flow rate of the outlet into consideration, and the vibration characteristic of the upper-mounted axial flow valve represents the actual working condition of the upper-mounted axial flow valve, so in the technical scheme of the application, the differential characteristic matrix between the vibration characteristic matrix and the transfer matrix is further calculated, and differential expression between the actual working condition and the reasonable working condition of the upper-mounted axial flow valve, namely differential characteristic distribution between the two is dug. In a specific example of the present application, the calculating the differential feature matrix between the vibration feature matrix and the transfer matrix includes: calculating a differential feature matrix between the vibration feature matrix and the transfer matrix according to the following formula; wherein, the formula is:
Figure BDA0004008669160000141
Wherein M is 1 Representing the vibration characteristic matrix, M 2 Representing the transfer matrix, M c Representing the matrix of the differential features in question,
Figure BDA0004008669160000151
representing the difference by location. />
Specifically, during the operation of the intelligent on-package axial flow valve 300, the small-scale feature optimization module 370 is configured to correct the differential feature matrix based on a small-scale feature correlation pattern between the vibration feature matrix and the transfer matrix to obtain a corrected differential feature matrix. In particular, in the technical solution of the present application, when calculating the differential feature matrix between the vibration feature matrix and the transfer matrix, the convolution neural network model using the spatial attention mechanism can strengthen the local feature expression inside the vibration feature matrix, and the transfer matrix is focused on expressing the global distribution transfer feature of the inlet flow velocity feature vector relative to the outlet flow velocity feature vector, so that the small-scale feature association expression of the vibration feature matrix and the transfer matrix needs to be improved, and then the expression effect of the differential feature matrix on the position-by-position association differential feature between the vibration feature matrix and the transfer matrix is improved. Therefore, in the technical scheme of the application, the small-scale local derivative matrix of the vibration characteristic matrix and the transfer matrix is calculated as a weighted characteristic matrix, wherein the formula is as follows:
Figure BDA0004008669160000152
Wherein the method comprises the steps of
Figure BDA0004008669160000153
And->
Figure BDA0004008669160000154
Vector values at the (i, j) th positions of the vibration feature matrix, the transfer matrix, and the small-scale local derivative matrix, respectively; and taking the small-scale local derivative matrix as a weighting matrix to carry out dot multiplication on the differential characteristic matrix to carry out characteristic value weighting so as to obtain a corrected differential characteristic matrix. Here, by calculating the vibration characteristic matrix M 1 And the transfer matrix M 2 Small-scale local derivative features in between can be based on the vibration feature matrix M 1 And the transfer matrix M 2 The geometrical approximation of the corresponding positions in between mimics the physics of the mutual expression between data sequences, thereby enhancing the local nonlinear dependence of the cross-feature-domain positions with a point-by-point regression of the feature matrices by position. Thus, by locally deriving the matrix M at said small scale w The differential feature matrix is multiplied by the point as a weighting matrix to weight the feature value, so that the differential feature matrix can be improvedAnd the vibration characteristic matrix and the small-scale position-by-position correlation differential characteristic of the transfer matrix express effect, so that the accuracy of the classification result of the differential characteristic matrix is improved. Therefore, the working state of the upper-mounted axial flow valve can be accurately detected based on actual conditions, so that an abnormal early warning prompt is generated when the working state is abnormal, and the normal operation of the upper-mounted axial flow valve is ensured.
FIG. 6 is a block diagram of a small-scale feature optimization module in an intelligent top-loading axial flow valve according to an embodiment of the present application. As shown in fig. 6, the small-scale feature optimization module 370 includes: the small-scale feature association mode extraction unit 371 is configured to calculate a small-scale local derivative matrix of the vibration feature matrix and the transfer matrix as a weighted feature matrix, where the formula is as follows:
Figure BDA0004008669160000161
wherein the method comprises the steps of
Figure BDA0004008669160000162
And->
Figure BDA0004008669160000163
Vector values at the (i, j) th positions of the vibration feature matrix, the transfer matrix, and the small-scale local derivative matrix, respectively; and a weighting unit 372, configured to perform dot multiplication on the differential feature matrix with the small-scale local derivative matrix as a weighting matrix to perform feature value weighting to obtain a corrected differential feature matrix.
Specifically, during the operation of the intelligent top-loading axial flow valve 300, the intelligent monitoring result generating module 380 is configured to pass the corrected differential feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate whether the working state of the top-loading axial flow valve to be monitored is normal. That is, the corrected differential feature matrix is used as a classification feature matrix to be classified in a classifier, so as to obtain the upper-loading axial flow to be monitored A classification of whether the valve operating condition is normal. Therefore, the working state of the upper-mounted axial flow valve can be detected based on actual conditions, and further, an abnormality early warning prompt is generated when the working state is abnormal. More specifically, the corrected differential feature matrix is processed using the classifier to obtain a classification result with the following formula: o=softmax { (W) n ,B n ):…:(W 1 ,B 1 ) Project (F), where Project (F) represents projecting the corrected differential feature matrix as a vector, W 1 To W n Weight matrix for all the connection layers of each layer, B 1 To B n Representing the bias vector for each fully connected layer. Specifically, the classifier includes a plurality of fully connected layers and a Softmax layer cascaded with a last fully connected layer of the plurality of fully connected layers. In the classification process of the classifier, the corrected differential feature matrix is first projected as a vector, for example, in a specific example, the corrected differential feature matrix is expanded along a row vector or a column vector to be a classification feature vector; then, performing multiple full-connection coding on the classification feature vectors by using multiple full-connection layers of the classifier to obtain coded classification feature vectors; further, inputting the coding classification feature vector into a Softmax layer of the classifier, namely, classifying the coding classification feature vector by using the Softmax classification function to obtain a first probability value that the coding classification feature vector belongs to the working state of the upper-mounted axial flow valve to be monitored normally and a second probability value that the coding classification feature vector belongs to the working state of the upper-mounted axial flow valve to be monitored abnormally; and then, determining the label corresponding to the larger one of the first probability value and the second probability value as the classification result, namely, if the first probability value is larger than the second probability value, the classification result is that the working state of the upper-mounted axial flow valve to be monitored is normal, and if the second probability value is larger than the first probability value, the classification result is that the working state of the upper-mounted axial flow valve to be monitored is abnormal.
Fig. 7 is a block diagram of an intelligent monitoring result generating module in an intelligent uploading axial flow valve according to an embodiment of the present application, as shown in fig. 7, the intelligent monitoring result generating module 380 includes: an expanding unit 381 for expanding the corrected differential feature matrix into classified feature vectors based on row vectors or column vectors; a full-connection encoding unit 382, configured to perform full-connection encoding on the classification feature vector by using multiple full-connection layers of the classifier to obtain an encoded classification feature vector; and a classification result generating unit 383, configured to pass the encoded classification feature vector through a Softmax classification function of the classifier to obtain the classification result.
In summary, the intelligent up-loading axial flow valve 300 according to the embodiment of the present application is illustrated, by adopting an artificial intelligent detection technology based on deep learning, to simulate a reasonable working condition of the up-loading axial flow valve by using a transfer matrix between flow velocity characteristics of an inlet and an outlet in a high-dimensional space, and the vibration characteristics represent an actual working condition of the up-loading axial flow valve, so that a difference between the two represents a differential expression between the actual working condition and the reasonable working condition, and a detection result of a working state of the up-loading axial flow valve is obtained through a classifier. Therefore, the working state of the upper-mounted axial flow valve can be accurately detected based on actual conditions, so that an abnormal early warning prompt is generated when the working state is abnormal, and the normal operation of the upper-mounted axial flow valve is ensured.
Exemplary method
Fig. 8 is a flow chart of a method of using an intelligent top-loading axial flow valve according to an embodiment of the present application. As shown in fig. 8, a method for using an intelligent top-loading axial flow valve according to an embodiment of the present application includes the steps of: s110, acquiring fluid flow velocity values of inlets at a plurality of preset time points in a preset time period, fluid flow velocity values of outlets of the preset time period, and vibration signals of the to-be-monitored upper-mounted axial flow valve in the preset time period; s120, arranging the fluid flow velocity values of the inlets and the fluid flow velocity values of the outlets of the plurality of preset time points into an inlet flow velocity input vector and an outlet flow velocity input vector according to the time dimension respectively; s130, respectively passing the inlet flow velocity input vector and the outlet flow velocity input vector through a multi-scale neighborhood feature extraction module to obtain an inlet flow velocity feature vector and an outlet flow velocity feature vector; s140, calculating a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector; s150, obtaining a vibration characteristic matrix through a convolution neural network model using a spatial attention mechanism according to a waveform diagram of a vibration signal of the to-be-monitored upper-loading axial flow valve in the preset time period; s160, calculating a differential feature matrix between the vibration feature matrix and the transfer matrix; s170, correcting the differential feature matrix based on a small-scale feature association mode between the vibration feature matrix and the transfer matrix to obtain a corrected differential feature matrix; and S180, passing the corrected differential feature matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the working state of the upper-mounted axial flow valve to be monitored is normal or not.
In one example, in the method for using an intelligent top-loading axial flow valve, the step S130 includes: respectively inputting the inlet flow velocity input vector and the outlet flow velocity input vector into a first convolution layer of the multi-scale neighborhood feature extraction module to obtain a first neighborhood scale inlet flow velocity feature vector and a first neighborhood scale outlet flow velocity feature vector, wherein the first convolution layer is provided with a first one-dimensional convolution kernel with a first length; respectively inputting the inlet flow velocity input vector and the outlet flow velocity input vector into a second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second neighborhood scale inlet flow velocity feature vector and a second neighborhood scale outlet flow velocity feature vector, wherein the second convolution layer is provided with a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and cascading the first neighborhood scale inlet flow velocity feature vector and the first neighborhood scale outlet flow velocity feature vector with the second neighborhood scale inlet flow velocity feature vector and the second neighborhood scale outlet flow velocity feature vector respectively to obtain the inlet flow velocity feature vector and the outlet flow velocity feature vector. More specifically, the inputting the inlet flow velocity input vector and the outlet flow velocity input vector into the first convolution layer of the multi-scale neighborhood feature extraction module to obtain a first neighborhood scale inlet flow velocity feature vector and a first neighborhood scale outlet flow velocity feature vector, respectively, includes: performing one-dimensional convolution encoding on the inlet flow velocity input vector and the outlet flow velocity input vector by using a first convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a first neighborhood scale inlet flow velocity feature vector and a first neighborhood scale outlet flow velocity feature vector; wherein, the formula is:
Figure BDA0004008669160000181
Wherein a is the width of the first convolution kernel in the X direction, F (a) is a first convolution kernel parameter vector, G (X-a) is a local vector matrix calculated by a convolution kernel function, w is the size of the first convolution kernel, and X represents the inlet flow velocity input vector and the outlet flow velocity input vector; and inputting the inlet flow velocity input vector and the outlet flow velocity input vector into a second convolution layer of the multi-scale neighborhood feature extraction module respectively to obtain a second neighborhood scale inlet flow velocity feature vector and a second neighborhood scale outlet flow velocity feature vector, including: performing one-dimensional convolution encoding on the inlet flow velocity input vector and the outlet flow velocity input vector by using a second convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a second neighborhood scale inlet flow velocity feature vector and a second neighborhood scale outlet flow velocity feature vector; wherein, the formula is:
Figure BDA0004008669160000191
wherein b is the width of the second convolution kernel in the X direction, F (b) is a second convolution kernel parameter vector, G (X-b) is a local vector matrix that operates with a convolution kernel function, m is the size of the second convolution kernel, and X represents the inlet flow velocity input vector and the outlet flow velocity input vector. Wherein, the multiscale neighborhood feature extraction module comprises: the device comprises a first convolution layer, a second convolution layer parallel to the first convolution layer and a cascade layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first scale, and the second convolution kernel uses a one-dimensional convolution kernel with a second scale.
In one example, in the method for using an intelligent top-loading axial flow valve, the step S140 includes: calculating a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector with the following formula; wherein, the formula is:
Figure BDA0004008669160000192
wherein V is c Representing the inlet flow velocity characteristic vector, V representing the outlet flow velocity characteristic vector, M 2 Representing the transfer matrix.
In one example, in the method for using an intelligent top-loading axial flow valve, the step S150 includes: respectively carrying out convolution processing, pooling processing along a channel dimension and nonlinear activation processing on a waveform diagram of a vibration signal of the to-be-monitored upper-loading axial flow valve in the preset time period in forward transmission of layers by using each layer of the convolutional neural network model so as to output a plurality of initial vibration feature matrixes by the last layer of the convolutional neural network model; and inputting the plurality of initial vibration feature matrices into a spatial attention layer of the convolutional neural network model to obtain the vibration feature matrix.
In one example, in the method for using an intelligent top-loading axial flow valve, the step S160 includes: calculating a differential feature matrix between the vibration feature matrix and the transfer matrix according to the following formula; wherein, the formula is:
Figure BDA0004008669160000193
Wherein M is 1 Representing the vibration characteristic matrix, M 2 Representing the transfer matrix, M c Representing the differenceThe characteristic matrix is divided into a plurality of characteristic matrices,
Figure BDA0004008669160000194
representing the difference by location.
In one example, in the method for using an intelligent top-loading axial flow valve, the step S170 includes: calculating a small-scale local derivative matrix of the vibration characteristic matrix and the transfer matrix as a weighted characteristic matrix, wherein the formula is as follows:
Figure BDA0004008669160000201
wherein the method comprises the steps of
Figure BDA0004008669160000202
And->
Figure BDA0004008669160000203
Vector values at the (i, j) th positions of the vibration feature matrix, the transfer matrix, and the small-scale local derivative matrix, respectively; and taking the small-scale local derivative matrix as a weighting matrix to carry out dot multiplication on the differential characteristic matrix to carry out characteristic value weighting so as to obtain a corrected differential characteristic matrix.
In one example, in the method for using an intelligent top-loading axial flow valve, the step S180 includes: expanding the corrected differential feature matrix into classified feature vectors based on row vectors or column vectors; performing full-connection coding on the classification feature vectors by using a plurality of full-connection layers of the classifier to obtain coded classification feature vectors; and passing the coding classification feature vector through a Softmax classification function of the classifier to obtain the classification result.
In summary, the use method of the intelligent top-loading axial flow valve according to the embodiment of the application is clarified, by adopting an artificial intelligent detection technology based on deep learning, reasonable working conditions of the top-loading axial flow valve are simulated by utilizing a transfer matrix between flow velocity characteristics of an inlet and an outlet in a high-dimensional space, and vibration characteristics represent actual working conditions of the top-loading axial flow valve, so that differential expression between the actual working conditions and the reasonable working conditions is represented by the difference between the two, and a detection result of the working state of the top-loading axial flow valve is obtained through a classifier. Therefore, the working state of the upper-mounted axial flow valve can be accurately detected based on actual conditions, so that an abnormal early warning prompt is generated when the working state is abnormal, and the normal operation of the upper-mounted axial flow valve is ensured.

Claims (9)

1. An intelligent top-loading axial flow valve, comprising: the sensor monitoring module is used for acquiring fluid flow velocity values of inlets at a plurality of preset time points in a preset time period, fluid flow velocity values of outlets of the preset time period and vibration signals of the to-be-monitored upper-mounted axial flow valve in the preset time period; a sensor data structuring module, configured to arrange the fluid flow rate values of the inlet and the fluid flow rate values of the outlet at the plurality of predetermined time points into an inlet flow rate input vector and an outlet flow rate input vector according to a time dimension, respectively; the flow velocity feature extraction module is used for respectively passing the inlet flow velocity input vector and the outlet flow velocity input vector through the multi-scale neighborhood feature extraction module to obtain an inlet flow velocity feature vector and an outlet flow velocity feature vector; a global transfer module for calculating a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector; the vibration signal characteristic extraction module is used for obtaining a vibration characteristic matrix through a convolution neural network model using a spatial attention mechanism according to a waveform diagram of the vibration signal of the to-be-monitored upper-loading axial flow valve in the preset time period; the differential module is used for calculating a differential characteristic matrix between the vibration characteristic matrix and the transfer matrix; the small-scale feature optimization module is used for correcting the differential feature matrix based on a small-scale feature association mode between the vibration feature matrix and the transfer matrix to obtain a corrected differential feature matrix; and the intelligent monitoring result generation module is used for enabling the corrected differential feature matrix to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the working state of the upper-mounted axial flow valve to be monitored is normal or not.
2. The intelligent, top-loading axial flow valve of claim 1, wherein the multi-scale neighborhood feature extraction module comprises: the device comprises a first convolution layer, a second convolution layer parallel to the first convolution layer and a cascade layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first scale, and the second convolution kernel uses a one-dimensional convolution kernel with a second scale.
3. The intelligent, top-loading axial flow valve of claim 2, wherein the flow rate feature extraction module comprises: a first scale feature extraction unit, configured to input the inlet flow velocity input vector and the outlet flow velocity input vector into a first convolution layer of the multi-scale neighborhood feature extraction module, respectively, to obtain a first neighborhood scale inlet flow velocity feature vector and a first neighborhood scale outlet flow velocity feature vector, where the first convolution layer has a first one-dimensional convolution kernel with a first length; a second scale feature extraction unit, configured to input the inlet flow velocity input vector and the outlet flow velocity input vector into a second convolution layer of the multi-scale neighborhood feature extraction module, respectively, to obtain a second neighborhood scale inlet flow velocity feature vector and a second neighborhood scale outlet flow velocity feature vector, where the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and the multiscale fusion unit is used for cascading the first neighborhood scale inlet flow velocity characteristic vector and the first neighborhood scale outlet flow velocity characteristic vector with the second neighborhood scale inlet flow velocity characteristic vector and the second neighborhood scale outlet flow velocity characteristic vector respectively to obtain the inlet flow velocity characteristic vector and the outlet flow velocity characteristic vector.
4. The intelligent, top-loading axial flow valve of claim 3, wherein the first scale feature extraction unit is further configured to: performing one-dimensional convolution encoding on the inlet flow velocity input vector and the outlet flow velocity input vector by using a first convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a first neighborhood scale inlet flow velocity feature vector and a first neighborhood scale outlet flow velocity feature vector; wherein, the formula is:
Figure FDA0004008669150000021
wherein a is the width of the first convolution kernel in the X direction, F (a) is a first convolution kernel parameter vector, G (X-a) is a local vector matrix calculated by a convolution kernel function, w is the size of the first convolution kernel, and X represents the inlet flow velocity input vector and the outlet flow velocity input vector; the second scale feature extraction unit is further configured to: performing one-dimensional convolution encoding on the inlet flow velocity input vector and the outlet flow velocity input vector by using a second convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a second neighborhood scale inlet flow velocity feature vector and a second neighborhood scale outlet flow velocity feature vector; wherein, the formula is:
Figure FDA0004008669150000022
Wherein b is the width of the second convolution kernel in the X direction, F (b) is a second convolution kernel parameter vector, G (X-b) is a local vector matrix that operates with a convolution kernel function, m is the size of the second convolution kernel, and X represents the inlet flow velocity input vector and the outlet flow velocity input vector.
5. The intelligent, on-board axial flow valve of claim 4, wherein the global transfer module is further configured to: calculating a transfer matrix of the inlet flow velocity eigenvector relative to the outlet flow velocity eigenvector with the following formula; wherein, the formula is:
Figure FDA0004008669150000023
wherein V is c Representing the inlet flow velocity characteristic vector, V representing the outlet flow velocity characteristic vector, M 2 Representing the transfer matrix.
6. The intelligent, top-loading axial flow valve of claim 5, wherein the vibration signal feature extraction module is further configured to: respectively carrying out convolution processing, pooling processing along a channel dimension and nonlinear activation processing on a waveform diagram of a vibration signal of the to-be-monitored upper-loading axial flow valve in the preset time period in forward transmission of layers by using each layer of the convolutional neural network model so as to output a plurality of initial vibration feature matrixes by the last layer of the convolutional neural network model; and inputting the plurality of initial vibration feature matrices into a spatial attention layer of the convolutional neural network model to obtain the vibration feature matrix.
7. The intelligent, top-loading axial flow valve of claim 6, wherein the differential module is further configured to: calculating a differential feature matrix between the vibration feature matrix and the transfer matrix according to the following formula; wherein, the formula is:
Figure FDA0004008669150000031
wherein M is 1 Representing the vibration characteristic matrix, M 2 Representing the transfer matrix, M c Representing the matrix of the differential features in question,
Figure FDA0004008669150000032
representing the difference by location.
8. The intelligent, top-loading axial flow valve of claim 7, wherein the small scale feature optimization module comprises: the small-scale feature association mode extraction unit is used for calculating a small-scale local derivative matrix of the vibration feature matrix and the transfer matrix as a weighted feature matrix, wherein the formula is as follows:
Figure FDA0004008669150000033
wherein the method comprises the steps of
Figure FDA0004008669150000034
And->
Figure FDA0004008669150000035
Vector values at the (i, j) th positions of the vibration feature matrix, the transfer matrix, and the small-scale local derivative matrix, respectively; and the weighting action unit is used for carrying out dot multiplication on the differential feature matrix by taking the small-scale local derivative matrix as a weighting matrix to carry out feature value weighting so as to obtain a corrected differential feature matrix.
9. The intelligent top-loading axial flow valve of claim 8, wherein the intelligent monitoring result generation module comprises: the expansion unit is used for expanding the corrected differential feature matrix into classification feature vectors based on row vectors or column vectors; the full-connection coding unit is used for carrying out full-connection coding on the classification characteristic vectors by using a plurality of full-connection layers of the classifier so as to obtain coded classification characteristic vectors; and the classification result generating unit is used for passing the coding classification feature vector through a Softmax classification function of the classifier to obtain the classification result.
CN202211643344.5A 2022-12-20 2022-12-20 Intelligent upper-mounted axial flow valve Withdrawn CN116123349A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211643344.5A CN116123349A (en) 2022-12-20 2022-12-20 Intelligent upper-mounted axial flow valve

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211643344.5A CN116123349A (en) 2022-12-20 2022-12-20 Intelligent upper-mounted axial flow valve

Publications (1)

Publication Number Publication Date
CN116123349A true CN116123349A (en) 2023-05-16

Family

ID=86298405

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211643344.5A Withdrawn CN116123349A (en) 2022-12-20 2022-12-20 Intelligent upper-mounted axial flow valve

Country Status (1)

Country Link
CN (1) CN116123349A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116736895A (en) * 2023-05-25 2023-09-12 潮州深能城市燃气发展有限公司 Gas flow control system based on sonic nozzle method
CN116992226A (en) * 2023-06-16 2023-11-03 青岛西格流体技术有限公司 Water pump motor fault detection method and system

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116736895A (en) * 2023-05-25 2023-09-12 潮州深能城市燃气发展有限公司 Gas flow control system based on sonic nozzle method
CN116736895B (en) * 2023-05-25 2024-05-03 潮州深能城市燃气发展有限公司 Gas flow control system based on sonic nozzle method
CN116992226A (en) * 2023-06-16 2023-11-03 青岛西格流体技术有限公司 Water pump motor fault detection method and system

Similar Documents

Publication Publication Date Title
CN116123349A (en) Intelligent upper-mounted axial flow valve
Park et al. Machine-learning-based feedback control for drag reduction in a turbulent channel flow
Che et al. Hybrid multimodal fusion with deep learning for rolling bearing fault diagnosis
Ayodeji et al. Causal augmented ConvNet: A temporal memory dilated convolution model for long-sequence time series prediction
CN116048031B (en) Control system and method for petroleum auxiliary production
US7177787B2 (en) Method for modelling hydrodynamic characteristics of multiphase flows using neuronal networks
Jose et al. Early detection and classification of internal leakage in boom actuator of mobile hydraulic machines using SVM
CN116086790A (en) Performance detection method and system for high-pressure valve of hydrogen fuel cell
EP4081453A1 (en) Control system
CN116451567A (en) Leakage assessment and intelligent disposal method for gas negative pressure extraction pipeline
Isaza et al. Situation prediction based on fuzzy clustering for industrial complex processes
KR102404498B1 (en) Industrial gearbox failure diagnosis apparatus and method using convolutional neural network based on adaptive time-frequency representation
Li et al. Structural health monitoring data anomaly detection by transformer enhanced densely connected neural networks
Tang et al. A light deep adaptive framework toward fault diagnosis of a hydraulic piston pump
KR102266279B1 (en) Method for Building Reduced Order Model for Implementing Transient State
Esfahani et al. A new hybrid model for RUL prediction through machine learning
CN116449881A (en) Intelligent cylinder and control method thereof
Fu et al. High imbalance fault diagnosis of aviation hydraulic pump based on data augmentation via local wavelet similarity fusion
CN116597635A (en) Wireless communication intelligent gas meter controller and control method thereof
Zheng et al. Normalized recurrent dynamic adaption network: A new framework with dynamic alignment for intelligent fault diagnosis
CN116025765A (en) Axial flow type regulating valve and control method thereof
CN116124448A (en) Fault diagnosis system and method for wind power gear box
CN115164115A (en) Gas pipeline leakage rate prediction method based on physical model driven machine learning
Dube et al. Machine Learning Approach to Predict Aerodynamic Performance of Underhood and Underbody Drag Enablers
CN106773703B (en) Online prediction technique is loaded based on fuzzy reasoning and the forging press of Taylor expansion

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
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20230516