Electromagnetic valve detection system based on magnetic field sensor and model training method thereof

Abstract

The invention relates to the technical field of electromagnetic valve detection, in particular to an electromagnetic valve detection system based on a magnetic field sensor and a model training method of the electromagnetic valve detection system based on the magnetic field sensor. At least one detection site for placing an electromagnetic valve and at least two first mounting sites for placing a magnetic field sensor are arranged on the detection table; the upper computer system comprises a learning module and a data processing module; the magnetic field sensors are arranged on the first installation site, the number of the magnetic field sensors is at least two, the magnetic field sensors are connected with the learning module and the data processing module, and the magnetic field sensors are used for transmitting electromagnetic changes in the working process of the electromagnetic valve to the learning module; and the learning module forms a database by using the obtained data, and a data model is obtained by training through a machine learning method.

Description

Technical Field

[0001] The present disclosure relates to the technical field of solenoid valve detection, and in particular, to a solenoid valve detection system based on a magnetic field sensor and a model training method thereof. Background Art

[0002] A solenoid valve is a valve body containing one or several holes, and its working principle is based on the combination of electromagnetic force and mechanical movement. The opening and closing of the valve are achieved by controlling the on-off of the current of the electromagnet. The solenoid valve consists of an iron core, an inductance coil, a valve body, a seal, etc. Among them, the iron core is installed in the valve body, and the valve is composed of a conical valve core and sealing parts, etc., and the coil is mainly made of copper wire. When the coil is energized and excited, the magnetic force can cause the iron core to move, thereby cutting off or energizing the fluid in the pipeline to adjust the pressure or flow rate. Nowadays, solenoid valves have been widely used in the fields of hydraulics, pneumatics, machine tools, textiles, metallurgy, chemical engineering, boilers, etc.

[0003] The dynamic characteristics such as the displacement, response time, and electromagnetic force of a solenoid valve are the core indicators for evaluating the performance of a solenoid valve. Usually, they are measured by means of laser displacement detection, force rod conduction, etc., which have the advantage of high accuracy. However, due to the limitation of the product's own structure, it is difficult to directly detect the dynamic characteristics of solenoid valves by the direct method (except for a few solenoid valves with the valve core outside). In addition, the detection equipment related to the direct method is expensive and the detection time is long, and it is impossible to realize the rapid detection of a large number of products. At present, the method adopted is to indirectly detect the dynamic characteristics of solenoid valves. The existing method is to detect the inflection points of the changes in data such as current and resistance over time, and then process the detection response time through a mathematical model. However, since the electromagnetic environment is very complex when the solenoid valve is actually working, there is currently no mathematical model that can describe this process through data such as current, vibration, and resistance, which makes it impossible to complete the detection of displacement and electromagnetic force. Summary of the Invention

[0004] In view of the deficiencies or problems existing in the prior art, the present disclosure provides a solenoid valve detection system based on a magnetic field sensor. By detecting the spatial magnetic field data during the working process of the solenoid valve and combining machine learning methods, the dynamic characteristic data of the displacement, response time, and electromagnetic force of the solenoid valve can be directly measured.

[0005] The technical solution adopted by the present disclosure to solve the above technical problems is as follows: A solenoid valve detection system based on a magnetic field sensor, comprising:

[0006] A detection table, on which at least one detection site for placing a solenoid valve and at least two first installation sites for placing magnetic field sensors are provided;

[0007] Host computer system, the host computer system includes a learning module and a data processing module;

[0008] Magnetic field sensors, the magnetic field sensors are arranged at the first installation sites, at least two magnetic field sensors, the magnetic field sensors are respectively connected to the learning module and the data processing module, the magnetic field sensors are used to transmit the electromagnetic changes during the operation of the solenoid valve to the learning module, the learning module forms a database with the obtained data, and obtains a data model through machine learning methods, and the data processing module compares the data of the solenoid valve to be tested with the data model in the learning module to determine whether the solenoid valve is qualified.

[0009] As a preferred embodiment, it further includes a displacement sensor, the displacement sensor is used to measure the amount of movement during the movement of the solenoid valve, and the displacement sensor is respectively connected to the host computer system and the solenoid valve.

[0010] As a preferred embodiment, it further includes an externally connected force sensor, the force sensor is used to measure the electromagnetic force received by the magnetic core during the movement of the solenoid valve, and the force sensor is connected to the host computer system and the solenoid valve.

[0011] As a preferred embodiment, it further includes a current sensor and / or a voltage sensor, and the current sensor and / or the voltage sensor are respectively electrically connected to the host computer system and the solenoid valve.

[0012] As a preferred embodiment, the electromagnetic change of the solenoid valve is the dynamic characteristic data of the solenoid valve, including displacement, response time and electromagnetic force.

[0013] As a preferred embodiment, when there are more than three magnetic field sensors at the first installation site, at least one magnetic field sensor is not at the same height as the remaining magnetic field sensors.

[0014] As a preferred embodiment, the detection site is a groove.

[0015] As a preferred embodiment, the magnetic field sensor is an amorphous wire GMI magnetic sensor or a GMR sensor or a TMR sensor or a Hall sensor.

[0016] As a preferred embodiment, the learning module is a convolutional neural network model or a BP neural network model.

[0017] As a preferred embodiment, the detection system includes the following detection method: detecting the solenoid valve to be tested and comparing it with the data in the database.

[0018] As a preferred embodiment, the detection of the solenoid valve to be tested includes the detection of the dynamic characteristic data of the solenoid valve to be tested. The detection of the dynamic characteristic data of the solenoid valve to be tested includes the following steps: Place the solenoid valve to be tested at the detection site, energize the solenoid valve, the magnetic field sensor collects the relevant data information of the solenoid valve, and transmits this information to the data processing module. Through the trained learning module, the electromagnetic force data, displacement data of the target product, and the corresponding solenoid valve are output, and the dynamic characteristic data of the solenoid valve to be tested and whether the target product is qualified are obtained.

[0019] As a preferred embodiment, the detection of the solenoid valve to be tested further includes the fault analysis of the unqualified solenoid valve. The fault analysis of the unqualified solenoid valve includes the following steps: If the result output by the data processing module is an unqualified product, the host computer system analyzes whether the electromagnetic change of the solenoid valve to be tested is qualified. If the electromagnetic change is qualified, the solenoid valve to be tested is a qualified product; if the electromagnetic change is unqualified, it is necessary to detect whether the current of the solenoid valve to be tested is equal to the rated current. If the detected current of the solenoid valve to be tested is not equal to the rated current, a fault occurs in the solenoid valve winding; if the detected current of the solenoid valve to be tested is equal to the rated current, then it is judged whether the magnetic field is equal to the rated magnetic field. If the magnetic field is not equal to the rated magnetic field, a fault occurs in the valve core; if the magnetic field is equal to the rated magnetic field, it is judged whether the response time is equal to the rated time. If the response time is not equal to the rated time, the valve core is stuck.

[0020] Another object of the present application also provides a model training method for a solenoid valve detection system based on a magnetic field sensor, which is characterized in that the model training includes the following steps: Using the magnetic field sensor data, current sensor data and / or voltage sensor data, and time data as input data, and the electromagnetic force data, displacement data, qualified samples and unqualified samples as result data for data training and verification, and establishing a database.

[0021] As a preferred embodiment, establishing the database includes the following steps: Place the target solenoid valve at the detection site, make the solenoid valve work through the host computer system, the magnetic field sensor, displacement sensor, current sensor and / or voltage sensor collect the relevant data of the target solenoid valve in chronological order, and transmit this data to the learning module. The data processing module sorts the magnetic field sensor data, current sensor data and / or voltage sensor data, time data, electromagnetic force data, displacement data, and the corresponding solenoid valve in chronological order to form a database and store it in the host computer system; Repeat the above steps for target solenoid valves of different models.

[0022] As a preferred embodiment, when there are two or more detection sites, the acquisition method is as follows: First, place the solenoid valves separately at the detection sites for data acquisition, and then place the solenoid valves at all detection sites by means of permutation and combination and collect the corresponding data.

[0023] As a preferred embodiment, the current sensor data and / or voltage sensor data time data are used as input data, and the electromagnetic force data, displacement data, and / or response time are collected by using existing detection devices and then input into the learning module.

[0024] Compared with the prior art, by arranging a magnetic field sensor on the detection table, after the solenoid valve is electrified, the magnetic field sensor transmits the electromagnetic change of the solenoid valve to the learning module, and the data processing module forms a database with the data obtained by the learning module, and compares the data of the solenoid valve to be measured with the database to measure the dynamic characteristic data such as the displacement, response time, and electromagnetic force of the solenoid valve to be measured; this detection system can directly measure the dynamic characteristic data of the solenoid valve and perform fault analysis, which is of great significance for quickly evaluating the performance of solenoid valves in batches. Description of the Drawings

[0025] The following will further describe the present application in detail with reference to the drawings and preferred embodiments. However, those skilled in the art will understand that these drawings are only drawn for the purpose of explaining the preferred embodiments and should not be construed as limiting the scope of the present application. In addition, unless otherwise specified, the drawings only schematically show the composition or structure of the described object and may include exaggerated displays, and the drawings are not necessarily drawn to scale.

[0026] Figure 1 is a schematic structural diagram of a solenoid valve detection system based on a magnetic field sensor according to the present disclosure;

[0027] Figure 2 is a flow chart for quickly detecting the quality of a solenoid valve to be measured according to the present disclosure;

[0028] Figure 3 is a flow chart of the learning steps for the dynamic characteristic data of the solenoid valve according to the present disclosure.

[0029] Description of the Reference Numerals:

[0030] 1. Detection table; 2. Detection site; 3. First installation site. Detailed Embodiments

[0031] As Figure 1As shown in the figure, the present application provides a solenoid valve detection system based on a magnetic field sensor, including: a detection table 1, on which there are at least one detection site 2 for placing a solenoid valve and at least two first installation sites 3 for placing magnetic field sensors; an upper computer system, which includes a learning module and a data processing module; a magnetic field sensor, which is arranged on the first installation site 3, and there are at least two magnetic field sensors, and the magnetic field sensors are electrically connected to the learning module and the data processing module respectively. The magnetic field sensor is used to transmit the electromagnetic changes during the operation of the solenoid valve to the learning module. The learning module forms a database with the obtained data and trains a data model through machine learning methods. The data processing module compares the data of the solenoid valve to be tested with the data model in the learning module to determine whether the solenoid valve is qualified. Among them, the electromagnetic changes during the operation of the solenoid valve are the dynamic characteristic data of the solenoid valve, including displacement, response time, and electromagnetic force. The detection site 2 is preferably a groove.

[0032] Specifically, the magnetic field sensor is an amorphous wire GMI magnetic sensor or a GMR sensor or a TMR sensor or a Hall sensor; the learning module is a convolutional neural network model or a BP neural network model. The machine learning method is a decision tree algorithm or a naive Bayes algorithm or a K-nearest neighbor algorithm or an AdaBoost algorithm or a PageRank algorithm or an EM algorithm (expectation maximization algorithm) or an Apriori algorithm or an SVM algorithm or a K-means clustering algorithm or a linear regression algorithm.

[0033] It also includes an external displacement sensor, which is used to measure the amount of movement of the solenoid valve during movement, and the displacement sensor is electrically connected to the upper computer system and the solenoid valve respectively.

[0034] It also includes an external force sensor, which is used to measure the electromagnetic force received by the magnetic core of the solenoid valve during movement, and the force sensor is electrically connected to the upper computer system and the solenoid valve.

[0035] It also includes a set current sensor and / or a voltage sensor, and the current sensor and / or the voltage sensor are electrically connected to the upper computer system and the solenoid valve respectively. Specifically, there are a second installation site and a third installation site on the detection table, the current sensor is arranged at the second installation site, and the voltage sensor is arranged at the third installation site.

[0036] It should be noted that when there are more than three magnetic field sensors at the first installation site 3, at least one magnetic field sensor is not at the same height as the remaining magnetic field sensors. Such a setting enables the system to measure the magnetic field in three-dimensional space.

[0037] As Figure 2As shown in the figure, the detection system includes the following detection methods: detecting the solenoid valve to be tested and comparing the data with the data in the database. Specifically, the detection of the solenoid valve to be tested includes the detection of the dynamic characteristic data of the solenoid valve to be tested. The detection of the dynamic characteristic data of the solenoid valve to be tested includes the following steps: placing the solenoid valve to be tested at the detection site 2, energizing the solenoid valve, collecting the relevant data information of the solenoid valve by the magnetic field sensor, and transmitting the information to the data processing module. The trained learning module outputs the electromagnetic force data, displacement data and the corresponding solenoid valve of the target product, and obtains the dynamic characteristic data of the solenoid valve to be tested and whether the target product is qualified. It should be noted that when detecting the sample to be tested, the data processing module does not need to be connected to the force sensor or the displacement sensor.

[0038] As Figure 2 shown in the figure, if the data output by the data processing module is an unqualified sample, the host computer system analyzes whether the electromagnetic change of the solenoid valve to be tested is qualified. If the electromagnetic change is qualified, the solenoid valve to be tested is a qualified product; if the electromagnetic change is unqualified, it is necessary to detect whether the current of the solenoid valve to be tested is equal to the rated current. If the detected current of the solenoid valve to be tested is not equal to the rated current, the solenoid valve winding fails; if the detected current of the solenoid valve to be tested is equal to the rated current, it is necessary to judge whether the magnetic field is equal to the rated magnetic field. If the magnetic field is not equal to the rated magnetic field, the valve core fails; if the magnetic field is equal to the rated magnetic field, it is necessary to judge whether the response time is equal to the rated time. If the response time is not equal to the rated time, the valve core is stuck.

[0039] As Figure 3 shown in the figure, the present application also provides a model training method for a solenoid valve detection system based on a magnetic field sensor. The model training method includes the following steps: using the magnetic field sensor data, current sensor data and / or voltage sensor data, and time data as input data, and using the electromagnetic force data, displacement data, qualified samples and unqualified samples as result data for data training and verification, and establishing a database.

[0040] Establishing the database includes the following steps: placing the target solenoid valve at the detection site 2, making the solenoid valve work through the host computer system, collecting the relevant data of the target solenoid valve by the magnetic field sensor, displacement sensor, current sensor and / or voltage sensor in chronological order, and transmitting the data to the learning module. The data processing module sorts the magnetic field sensor data, current sensor data and / or voltage sensor data, time data, electromagnetic force data, displacement data and the corresponding solenoid valve in chronological order to form a database, and stores it in the host computer system; repeating the above steps for different models of target solenoid valves.

[0041] It should be noted that when there are two or more detection sites 2, the acquisition method is as follows: First, place the solenoid valves separately on the detection site 2 to collect data, and then place all the detection sites 2 in a permutation and combination manner to place the solenoid valves and collect the corresponding data. The purpose is to reduce the detection error caused by the interference between adjacent solenoid valves. The current sensor data and / or voltage sensor data time data are used as input data, and the electromagnetic force data, displacement data, and / or response time are collected by using existing technology detection equipment and then input into the learning module.

[0042] Among them, the establishment of the database includes the establishment of a qualified solenoid valve database and an unqualified solenoid valve database.

[0043] Establishment of the qualified solenoid valve database: Place the qualified solenoid valves on the detection site 2, control the solenoid valves to work through the upper computer system and synchronously collect the data of the magnetic field sensor, current sensor, and voltage sensor in chronological order. During the operation of the solenoid valves, connect them to the external electromagnetic force test equipment and displacement detection equipment and detect synchronously; select qualified products of different models and conduct multiple measurements; the data processing module sorts the magnetic field sensor data, current sensor data, voltage sensor data, time data, electromagnetic force data, displacement data, and the corresponding solenoid valves in chronological order to form a database and store it in the upper computer system.

[0044] Establishment of the unqualified solenoid valve database: Place the unqualified solenoid valves on the detection site 2, control the solenoid valves to work through the upper computer system and synchronously collect the data of the magnetic field sensor, current sensor, and voltage sensor in chronological order. During the operation of the solenoid valves, connect them to the external electromagnetic force test equipment and displacement detection equipment and detect synchronously; select unqualified products of different models and conduct multiple measurements; the data processing module sorts the magnetic field sensor data, current sensor data, voltage sensor data, time data, electromagnetic force data, displacement data, and the corresponding solenoid valves in chronological order to form a database and store it in the upper computer system.

[0045] Then, train the learning module. Specifically: Use the magnetic field sensor data, current sensor data, voltage sensor data, and time data as input data, and use the electromagnetic force data, displacement data, qualified samples, and unqualified samples as result data to train and verify the learning module.

[0046] The above has introduced the present application in detail. Specific examples are used herein to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only for helping to understand the present application and its core idea. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present application, several improvements and modifications can also be made to the present application, and these improvements and modifications also fall within the protection scope of the claims of the present application.

Claims

1. An electromagnetic valve detection system based on a magnetic field sensor, characterized in that Including: A detection table (1), on which there are at least one detection site (2) for placing solenoid valves and at least two first installation sites (3) for placing magnetic field sensors; An upper computer system, which includes a learning module and a data processing module; Magnetic field sensors, which are arranged on the first installation sites (3), at least two magnetic field sensors, and the magnetic field sensors are respectively connected to the learning module and the data processing module. The magnetic field sensors are used to transmit the electromagnetic changes during the operation of the solenoid valve to the learning module. The learning module forms a database with the obtained data and trains a data model through machine learning methods. The data processing module compares the data of the solenoid valve to be tested with the data model in the learning module to determine whether the solenoid valve is qualified.

2. The solenoid valve detection system based on a magnetic field sensor according to claim 1, characterized in that It also includes a displacement sensor, which is used to measure the moving amount during the movement of the solenoid valve. The displacement sensor is respectively connected to the upper computer system and the solenoid valve.

3. The solenoid valve detection system based on a magnetic field sensor according to claim 1, characterized in that, It also includes a current sensor and / or a voltage sensor, and the current sensor and / or the voltage sensor are respectively electrically connected to the upper computer system and the solenoid valve.

4. The solenoid valve detection system based on a magnetic field sensor according to claim 1, characterized in that, The electromagnetic change of the solenoid valve is the dynamic characteristic data of the solenoid valve, including displacement, response time, and electromagnetic force.

5. The solenoid valve detection system based on a magnetic field sensor according to claim 1, characterized in that When there are more than three magnetic field sensors at the first installation site (3), at least one magnetic field sensor is not at the same height as the remaining magnetic field sensors.

6. The solenoid valve detection system based on a magnetic field sensor according to claim 1, wherein The detection site (2) is a groove.

7. The solenoid valve detection system based on a magnetic field sensor according to claim 1, characterized in that, The magnetic field sensor is an amorphous wire GMI magnetic sensor or a GMR sensor or a TMR sensor or a Hall sensor.

8. The solenoid valve detection system based on a magnetic field sensor according to claim 1, wherein, The learning module is a convolutional neural network model or a BP neural network model.

9. The solenoid valve detection system based on a magnetic field sensor according to claim 1 or 2 or 3, characterized in that, The detection system includes the following detection methods: the detection of the solenoid valve to be tested and the comparison with the data in the database.

10. The solenoid valve detection system based on a magnetic field sensor according to claim 9, wherein, The detection of the solenoid valve to be tested includes the detection of the dynamic characteristic data of the solenoid valve to be tested. The detection of the dynamic characteristic data of the solenoid valve to be tested includes the following steps: Place the solenoid valve to be tested on the detection site (2), energize the solenoid valve, the magnetic field sensor collects the relevant data information of the solenoid valve, and transmits this information to the data processing module. Through the trained learning module, the electromagnetic force data, displacement data, and the corresponding solenoid valve of the target product are output, and the dynamic characteristic data of the solenoid valve to be tested and whether the target product is qualified are obtained.

11. The solenoid valve detection system based on a magnetic field sensor according to claim 9, characterized in that, The detection of the solenoid valve to be tested also includes the fault analysis of the unqualified solenoid valve. The fault analysis of the unqualified solenoid valve includes the following steps: If the result output by the data processing module is an unqualified product, the upper computer system analyzes whether the electromagnetic change of the solenoid valve to be tested is qualified. If the electromagnetic change is qualified, the solenoid valve to be tested is a qualified product; if the electromagnetic change is unqualified, it is necessary to detect whether the current of the solenoid valve to be tested is equal to the rated current. If the detected current of the solenoid valve to be tested is not equal to the rated current, there is a fault in the solenoid valve winding; if the detected current of the solenoid valve to be tested is equal to the rated current, then it is judged whether the magnetic field is equal to the rated magnetic field. If the magnetic field is not equal to the rated magnetic field, there is a fault in the valve core; if the magnetic field is equal to the rated magnetic field, then it is judged whether the response time is equal to the rated time. If the response time is not equal to the rated time, there is a jam in the valve core.

12. The model training method for the solenoid valve detection system based on a magnetic field sensor according to any one of the above claims, characterized in that, Model training includes the following steps: using magnetic field sensor data, current sensor data, and / or voltage sensor data, and time data as input data, and using electromagnetic force data, displacement data, qualified samples, and unqualified samples as result data for data training and verification, and establishing a database.

13. The model training method of the solenoid valve detection system based on a magnetic field sensor according to claim 12, characterized in that, Establishing the database includes the following steps: placing the target solenoid valve at the detection site (2), operating the solenoid valve through the host computer system, collecting relevant data of the target solenoid valve by the magnetic field sensor, displacement sensor, current sensor, and / or voltage sensor in chronological order, and transmitting the data to the learning module. The data processing module sorts the magnetic field sensor data, current sensor data, and / or voltage sensor data, time data, electromagnetic force data, displacement data, and the corresponding solenoid valve in chronological order to form a database and stores it in the host computer system; repeating the above steps for target solenoid valves of different models.

14. The model training method of the solenoid valve detection system based on a magnetic field sensor according to claim 13, characterized in that, When there are two or more detection sites (2), the acquisition method is as follows: first, place the solenoid valves separately on the detection sites (2) for data acquisition, and then place all the detection sites (2) in a permutation and combination manner to place the solenoid valves and collect the corresponding data.

15. The model training method of the solenoid valve detection system based on a magnetic field sensor according to claim 12, characterized in that, The current sensor data and / or voltage sensor data and time data are used as input data, and the electromagnetic force data, displacement data, and / or response time are collected by using existing technology detection equipment and then input into the learning module.

Images