High-speed switch valve flow soft measurement method based on neural network

Abstract

The invention relates to a high-speed switch valve flow soft measurement method based on a neural network, and the method comprises the following steps: S1, constructing a neural network training data set through measuring the valve port flow, the duty ratio of an input voltage, the frequency of the input voltage, the oil temperature, a valve body structure parameter and an armature structure parameter of a high-speed switch valve under different working conditions; s2, performing neural network training on the data set collected in the S1; s3, using the trained neural network model to realize structure calibration of the high-speed switch valve with unknown valve body structure parameters and armature structure parameters; s4, transplanting the trained neural network model into a digital controller used for soft measurement of the flow of the high-speed switch valve; and S5, the digital controller calculates the corresponding flow. Compared with the prior art, accurate high-frequency response flow measurement in an extreme environment and self-adaptive flow soft measurement of valves of the same type can be realized, and the measurement efficiency and accuracy are improved.

Description

Technical Field

[0001] The invention relates to the technical field of flow control and measurement, and in particular to a high-speed switch valve flow soft measurement method based on a neural network. Background Art

[0002] As a key industrial automation component, the application background of high-speed switch valves mainly stems from the urgent need of modern industry for high-precision, fast response and reliable control technology. With the intelligent upgrading of the manufacturing industry and the continuous improvement of the automation level, high-speed switch valves have been widely used in hydraulic transmission systems, pneumatic control systems, precision machinery, fluid control, aerospace and other fields due to their ability to quickly open and close and accurately control flow and pressure, and have become one of the key technologies indispensable for achieving efficient, precise and energy-saving control. The flow rate of high-speed switch valves is one of the most critical parameters in its performance evaluation, which is directly related to its efficiency and adaptability in various industrial application scenarios. The size of the flow rate not only determines the volume of media (such as gas, liquid) that the valve can handle or adjust per unit time, but also directly affects the system's response speed, control accuracy and overall operating efficiency.

[0003] However, the traditional flow measurement method is to directly use traditional flow meters for measurement. Traditional flow meters have poor dynamic performance, high cost, and large size. At the same time, in some extreme application scenarios such as fuel measurement of aircraft engines. When the aircraft is flying at high speed, the aircraft engine is in extremely harsh environments such as strong vibration loads, strong temperature differences, and high pressure. At the same time, under the requirements of lightweight aviation equipment, the volume and weight of the flow meter will be limited. In the above application environment, traditional flow meters are difficult to achieve real-time and accurate measurement of flow due to their own mechanical structure matching tolerances and inertia, and cannot meet application scenarios with high-frequency response requirements.

[0004] In order to solve the problems of poor dynamic performance, high cost, large size and low accuracy in extreme environments of traditional flowmeters, Chinese patent CN202410128580.6 discloses a soft measurement method for electro-hydraulic control valve flow. The patent forms a data set by measuring the valve port pressure, valve core displacement, oil night temperature and main valve port flow, and uses a neural network to establish a mathematical model of valve port pressure, valve core displacement, oil night temperature and main valve port flow, so that the main valve port flow can be indirectly obtained by measuring the easily measured parameters of valve port pressure, valve core displacement and oil night temperature. However, the patent does not have the adaptive flow measurement capability of valves of the same type. When the structural parameters of the valve change slightly, the entire model needs to re-establish the data set training. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a high-speed switch valve flow soft measurement method based on neural network, which can realize accurate high-frequency response measurement of flow in extreme environments and adaptive flow soft measurement of valves of the same type, thereby improving the measurement efficiency and accuracy.

[0006] Innovation in the integration of neural network and high-speed switching valve structural parameters: This invention innovatively uses the structural parameters of the high-speed switching valve as one of the core elements of the input variables of the neural network. This move breaks the limitation of traditional flow measurement methods that ignore the structural characteristics of the valve body, realizes the correlation between the valve structure characteristics and the flow characteristics, and lays a solid foundation for adaptive flow measurement.

[0007] Pre-calibration of valves with unknown dimensions: Before the high-speed on-off valve is officially put into use, this patent innovatively introduces a pre-calibration process. By accurately giving the frequency and duty cycle of the input voltage of the high-speed on-off valve and measuring its corresponding valve port flow in real time, the trained neural network is used to calibrate its dimensions.

[0008] Adaptive flow measurement capability for valve body structure: Through the comprehensive application of the above innovations, this patent successfully realizes the adaptive flow measurement capability of the same type of high-speed switch valve, realizing the fundamental transformation of flow measurement technology from the traditional static, single mode to the dynamic, intelligent mode.

[0009] The present invention provides a high-speed switch valve flow soft measurement method based on a neural network, comprising the following steps:

[0010] S1: By measuring the valve port flow rate, input voltage duty cycle, input voltage frequency, oil temperature, valve body structure parameters and armature structure parameters of the high-speed switch valve under different working conditions, a data set for neural network training is constructed;

[0011] S2: Perform neural network training on the data set collected in S1 to obtain a trained neural network model;

[0012] S3: Before actual use, the duty cycle of the input voltage, the frequency of the input voltage, the oil temperature and the valve port flow rate of the high-speed switch valve are measured, and the trained neural network model is used to realize the structural calibration of the high-speed switch valve with unknown valve body structural parameters and armature structural parameters;

[0013] S4: Then transplant the trained neural network model into the digital controller used for high-speed switch valve flow soft measurement;

[0014] S5: In actual use, for a calibrated high-speed switching valve, the digital controller only needs to obtain easy-to-measure signals such as the duty cycle of the input voltage, the frequency of the input voltage, and the oil temperature to calculate the corresponding flow rate, thereby achieving accurate high-frequency response measurement of flow in extreme environments and adaptive flow soft measurement of valves of the same type.

[0015] Furthermore, the valve body structural parameters used for neural network training include valve inlet diameter and valve outlet diameter.

[0016] Furthermore, the armature structural parameters used for neural network training include armature diameter and total armature stroke.

[0017] Furthermore, the oil temperature parameter used for neural network training is measured by a temperature sensor.

[0018] Furthermore, the temperature sensor is installed at the oil inlet of the high-speed switching valve.

[0019] Furthermore, in S5, the method for obtaining the target flow rate by transplanting the digital controller with the trained neural network model is:

[0020] The digital controller obtains the input voltage duty cycle, input voltage frequency and oil temperature of the high-speed switching valve in real time; the digital controller inputs the input voltage duty cycle, input voltage frequency and oil temperature obtained in real time into the trained neural network model to obtain the target flow rate corresponding to the input voltage duty cycle, input voltage frequency and oil temperature.

[0021] Further, in S2, the neural network is trained according to the data set composed of relevant parameters, and the method includes:

[0022] Obtain the valve port flow, input voltage duty cycle, input voltage frequency, oil temperature, valve body structure parameters and armature structure parameters of the high-speed switch valve under different working conditions;

[0023] Preprocessing the acquired relevant parameters to obtain preprocessed relevant parameters;

[0024] The preprocessed related parameters are used as sample data sets and input into an untrained neural network model for training to obtain a target neural network model.

[0025] Further, the acquired related parameters are preprocessed to obtain the preprocessed related parameters, and the method includes:

[0026] The relevant parameters are filtered to remove noise interference, sampled, and processed by digital signals to obtain the pre-processed relevant parameters.

[0027] Further, in S3, the calibration method for a high-speed switch valve with unknown armature structural parameters and valve body structural parameters includes:

[0028] Apply a voltage with a certain duty cycle and frequency to the high-speed switching valve;

[0029] Use a flow meter to measure the valve port flow of a high-speed switching valve;

[0030] The trained target neural network is used to calibrate the structural parameters of the high-speed switching valve.

[0031] Furthermore, the neural network model is a BP neural network, a RBF neural network or a LSTM neural network.

[0032] Compared with the prior art, the present invention has the following advantages:

[0033] Based on the neural network model that can realize the measurement task, it can realize accurate high-frequency response measurement of flow in extreme environments and adaptive flow soft measurement of valves of the same type, thus improving the efficiency and accuracy of measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flowchart of the high-speed switch valve flow soft measurement method based on neural network;

[0035] Figure 2 A flow chart of the high-speed switching valve calibration method;

[0036] Figure 3 A flowchart of the neural network training method;

[0037] Figure 4 This is a schematic diagram of the structure of the high-speed switch valve used in the high-speed switch valve flow soft measurement method based on neural network.

[0038] Figure markings: 1-blocking frame, 2-screw plug, 3-housing, 4-coil, 5-iron core, 6-armature, 7-valve sleeve, 8-sealing iron, 9-upper spring, 10-push rod, 11-valve core, 12-lower spring, 13-valve seat, A-total stroke of high-speed switching valve, B-armature diameter, C-outlet diameter, D-inlet diameter. DETAILED DESCRIPTION

[0039] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Component models, material names, connection structures, control methods, algorithms and other features not clearly described in this technical solution are all considered to be common technical features disclosed in the prior art.

[0040] Example 1

[0041] This embodiment provides a high-speed switch valve flow soft measurement method based on neural network, such as Figure 1 As shown, the following steps are included:

[0042] S1: By measuring the valve port flow rate, input voltage duty cycle, input voltage frequency, oil temperature, valve body structure parameters and armature structure parameters of the high-speed switch valve under different working conditions, a data set for neural network training is constructed;

[0043] S2: Perform neural network training on the data set collected in S1 to obtain a trained neural network model;

[0044] S3: Before actual use, the duty cycle of the input voltage, the frequency of the input voltage, the oil temperature and the valve port flow rate of the high-speed switch valve are measured, and the trained neural network model is used to realize the structural calibration of the high-speed switch valve with unknown valve body structural parameters and armature structural parameters;

[0045] S4: Then transplant the trained neural network model into the digital controller used for high-speed switch valve flow soft measurement;

[0046] S5: In actual use, for a calibrated high-speed switching valve, the digital controller only needs to obtain easy-to-measure signals such as the duty cycle of the input voltage, the frequency of the input voltage, and the oil temperature to calculate the corresponding flow rate, thereby achieving accurate high-frequency response measurement of flow in extreme environments and adaptive flow soft measurement of valves of the same type.

[0047] In a specific embodiment, the valve body structural parameters used for neural network training include valve inlet diameter D and valve outlet diameter C.

[0048] In a specific implementation, the armature structural parameters used for neural network training include the armature diameter B and the total stroke of the armature 6 .

[0049] In a specific embodiment, the oil temperature parameter used for neural network training is measured by a temperature sensor.

[0050] In a specific implementation, the temperature sensor is installed at the oil inlet of the high-speed switching valve.

[0051] In a specific implementation, in S5, the method for obtaining the target flow rate by transplanting the digital controller with the trained neural network model is:

[0052] The digital controller obtains the input voltage duty cycle, input voltage frequency and oil temperature of the high-speed switching valve in real time; the digital controller inputs the input voltage duty cycle, input voltage frequency and oil temperature obtained in real time into the trained neural network model to obtain the target flow rate corresponding to the input voltage duty cycle, input voltage frequency and oil temperature.

[0053] like Figure 3 As shown, in a specific implementation, in S2, a neural network is trained according to a data set composed of relevant parameters, and the method includes:

[0054] Obtain the valve port flow, input voltage duty cycle, input voltage frequency, oil temperature, valve body structure parameters and armature structure parameters of the high-speed switch valve under different working conditions;

[0055] Preprocessing the acquired relevant parameters to obtain preprocessed relevant parameters;

[0056] The preprocessed related parameters are used as sample data sets and input into an untrained neural network model for training to obtain a target neural network model.

[0057] In a specific implementation manner, the acquired related parameters are preprocessed to obtain the preprocessed related parameters, and the method includes:

[0058] The relevant parameters are filtered to remove noise interference, sampled, and processed by digital signals to obtain the pre-processed relevant parameters.

[0059] like Figure 2 As shown, in a specific implementation, in S3, the calibration method for a high-speed switch valve with unknown armature structural parameters and valve body structural parameters includes:

[0060] Apply a voltage with a certain duty cycle and frequency to the high-speed switching valve;

[0061] Use a flow meter to measure the valve port flow of a high-speed switching valve;

[0062] The trained target neural network is used to calibrate the structural parameters of the high-speed switching valve.

[0063] In a specific implementation, the neural network model is a BP neural network.

[0064] like Figure 4 As shown, the high-speed switch valve used in the high-speed switch valve flow soft measurement method based on neural network, the high-speed switch valve includes: a plug frame 1, a screw plug 2, a shell 3, a coil 4, an iron core 5, an armature 6, a valve sleeve 7, a sealing iron 8, an upper spring 9, a push rod 10, a valve core 11, a lower spring 12, and a valve seat 13; A is the total stroke of the high-speed switch valve;

[0065] The housing 3 is a bearing component of the whole, providing protection and installation space for the internal components. The coil 4 is installed inside the housing 3, which surrounds the iron core 5. When the coil 4 is energized, it will generate electromagnetic force to drive the subsequent components to move. The iron core 5 is located at the center surrounded by the coil 4, providing a core magnetic conductive medium for the generation of electromagnetic force.

[0066] The armature 6 is arranged corresponding to the iron core 5. When the coil 4 is energized to generate an electromagnetic force, the armature 6 will be displaced. The armature 6 is connected to the push rod 10, so as to transmit its own displacement to the push rod 10.

[0067] The other end of the push rod 10 is connected to the valve core 11. Driven by the displacement of the armature 6, the push rod 10 pushes or pulls the valve core 11 to move. The armature 6 is assembled in the valve sleeve 7. The valve sleeve 7 guides and limits the movement of the armature 6, ensuring that the armature 6 performs reciprocating switching actions in the valve sleeve 7 along a predetermined route to achieve the opening and closing control of the valve.

[0068] The upper spring 9 and the lower spring 12 provide a restoring elastic force for the valve core 11. When the coil 4 is powered off and the electromagnetic force disappears, the spring force causes the valve core 11 to return to the initial position.

[0069] The sealing iron 8 plays the role of auxiliary sealing and magnetic conductivity, and enhances the sealing and electromagnetic performance of the entire valve. The valve seat 13 is located at the outlet position corresponding to the valve core 11, and cooperates with the valve core 11 to accurately control the on-off and flow rate of the liquid; the plug frame 1 and the screw plug 2 are used to plug the assembly holes and process holes reserved on the shell 3 to prevent oil leakage and ensure the internal sealing of the valve.

[0070] Components not described in detail in this embodiment are all existing components that can be purchased through public channels.

[0071] The above description of the embodiments is to facilitate the understanding and use of the invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments, and improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the present invention should be within the scope of protection of the present invention.

Claims

1. A high-speed switch valve flow soft measurement method based on neural network, characterized in that: The following steps are involved: S1: By measuring the valve port flow rate, input voltage duty cycle, input voltage frequency, oil temperature, valve body structure parameters and armature structure parameters of the high-speed switch valve under different working conditions, a data set for neural network training is constructed; S2: Perform neural network training on the data set collected in S1 to obtain a trained neural network model; S3: Before actual use, the duty cycle of the input voltage, the frequency of the input voltage, the oil temperature and the valve port flow rate of the high-speed switch valve are measured, and the trained neural network model is used to realize the structural calibration of the high-speed switch valve with unknown valve body structural parameters and armature structural parameters; S4: transplant the trained neural network model into the digital controller used for high-speed switch valve flow soft measurement; S5: In actual use, for a calibrated high-speed switching valve, the digital controller only needs to obtain easy-to-measure signals such as the duty cycle of the input voltage, the frequency of the input voltage, and the oil temperature to calculate the corresponding flow rate, thereby achieving accurate high-frequency response measurement of flow in extreme environments and adaptive flow soft measurement of valves of the same type.

2. The high-speed switch valve flow soft measurement method based on neural network according to claim 1 is characterized in that: The valve body structural parameters used for neural network training include valve inlet diameter and valve outlet diameter.

3. The high-speed switch valve flow soft measurement method based on neural network according to claim 1 is characterized in that: The armature structural parameters used for neural network training include the armature diameter and the total stroke of the armature (6).

4. The high-speed switch valve flow soft measurement method based on neural network according to claim 1 is characterized in that: The oil temperature parameter used for neural network training is measured by a temperature sensor.

5. The high-speed switch valve flow soft measurement method based on neural network according to claim 4 is characterized in that: The temperature sensor is installed at the oil inlet of the high-speed switching valve.

6. The high-speed switch valve flow soft measurement method based on neural network according to claim 1 is characterized in that: In S5, the method for obtaining the target flow rate by transplanting the digital controller with the trained neural network model is: The digital controller obtains the input voltage duty cycle, input voltage frequency and oil temperature of the high-speed switching valve in real time; the digital controller inputs the input voltage duty cycle, input voltage frequency and oil temperature obtained in real time into the trained neural network model to obtain the target flow rate corresponding to the input voltage duty cycle, input voltage frequency and oil temperature.

7. The neural network-based high-speed switch valve flow soft measurement method according to claim 1 is characterized in that: In S2, a neural network is trained according to a data set consisting of relevant parameters, and the method includes: Obtain the valve port flow, input voltage duty cycle, input voltage frequency, oil temperature, valve body structure parameters and armature structure parameters of the high-speed switch valve under different working conditions; Preprocessing the acquired relevant parameters to obtain preprocessed relevant parameters; The preprocessed related parameters are used as sample data sets and input into an untrained neural network model for training to obtain a target neural network model.

8. The neural network-based high-speed switch valve flow soft measurement method according to claim 7 is characterized in that: Preprocessing the acquired related parameters to obtain the preprocessed related parameters, the method includes: The relevant parameters are filtered to remove noise interference, sampled, and processed by digital signals to obtain the pre-processed relevant parameters.

9. The neural network-based high-speed switch valve flow soft measurement method according to claim 1, characterized in that: In S3, the calibration method for the high-speed switch valve with unknown armature structural parameters and valve body structural parameters includes: Apply a voltage with a certain duty cycle and frequency to the high-speed switching valve; Use a flow meter to measure the valve port flow of a high-speed switching valve; The trained target neural network is used to calibrate the structural parameters of the high-speed switching valve.

10. The high-speed switch valve flow soft measurement method based on neural network according to claim 1 is characterized in that: The neural network model is a BP neural network, a RBF neural network or a LSTM neural network.

Images