Self-adaptive valve control method and system suitable for complex working conditions

Abstract

The invention relates to the technical field of valve control, in particular to a self-adaptive valve control method and system suitable for complex working conditions, and the method comprises the steps: installing a sensor group on an installation pipeline of a valve body, and collecting the opening characteristic data of the valve body in real time through the sensor group; and inputting the collected opening characteristic data into a pre-constructed first machine learning model, outputting valve opening data, and generating a valve body opening instruction based on the valve opening data to control the opening of the valve. According to the self-adaptive valve control method suitable for the complex working conditions, the pipeline vibration data and the medium fluctuation data are introduced, the cross influence between the pipeline vibration data and the medium fluctuation data is combined, the valve opening compensation instruction is generated, errors caused by vibration or medium fluctuation can be effectively compensated, it is ensured that the response of the valve is more accurate, and the valve opening compensation accuracy is improved. And secondly, in combination with the self-learning capability of the first machine learning model, the system can continuously optimize the compensation algorithm so as to adapt to wider working condition changes.

Description

Technical Field

[0001] The present invention relates to the field of valve control technology, and in particular to an adaptive valve control method and system suitable for complex working conditions. Background Art

[0002] A valve is a device that controls the flow of fluids. It is widely used in pipeline systems of liquids, gases, steam and other fluids. The valve can adjust the flow of fluids through the pipeline. By adjusting the opening of the valve, the flow can be accurately controlled to meet the process needs. The valve is mainly used for flow regulation, pressure control, safety protection and other functions. There are many methods for adjusting the valve opening in the prior art, including manual, pneumatic, electric, hydraulic and other methods. With the development of automatic control technology, more precise control methods such as intelligent regulation, PID regulation and adaptive regulation have become a trend of widespread application in modern industry.

[0003] The Chinese patent application with publication number CN118244622A discloses a method for automatically adjusting flow and pressure by adjusting valve opening, including the following steps: S1, collecting historical data; historical data includes the opening of valves at all levels and the flow and pressure of each system, S2, pre-processing historical data, and obtaining historical data sets by interval division according to historical data; S3, using historical data sets to train neural network models to obtain valve opening prediction models; S4, inputting target flow and target pressure into valve opening prediction models to obtain valve predicted opening intervals; S5, controlling the opening of valves at all levels according to valve predicted opening intervals. This invention simplifies the process of repeatedly adjusting valves to ensure flow and pressure by training network models and using patterns to predict valve opening, thereby improving work efficiency.

[0004] As mentioned in the above application, the existing valve body intelligent control method is used to control the opening of the valve. Under normal working conditions, the valve opening instruction is generated based on the flow rate and pressure of the system. However, in actual use, the vibration of the pipeline and the fluctuation of the medium in the pipeline will affect the size of the valve opening. Therefore, in the prior art, simply adjusting the valve opening based on the flow rate and pressure of the system cannot guarantee the accuracy of valve control, which in turn affects the stability of the entire system. Summary of the invention

[0005] In order to solve the above problems, the present invention provides an adaptive valve control method and system suitable for complex working conditions.

[0006] The present invention adopts the following technical solutions, which are applicable to the adaptive valve control method under complex working conditions, including:

[0007] Step S01: installing a sensor group on the installation pipe of the valve body, and collecting the opening characteristic data of the valve body in real time through the sensor group;

[0008] Step S02: inputting the collected opening characteristic data into a pre-built first machine learning model, outputting valve opening data, and generating a valve body opening instruction based on the valve opening data to control the opening of the valve;

[0009] Step S03: collecting pipeline vibration data and medium fluctuation data in real time, wherein the pipeline vibration data includes vibration amplitude and vibration angular frequency, and the medium fluctuation data includes fluctuation amplitude and fluctuation angular frequency;

[0010] Step S04: Generate a valve opening compensation instruction based on the pipeline vibration data, the medium fluctuation data, and the cross-influence of the pipeline vibration and the medium fluctuation, and adaptively adjust the valve opening based on the valve opening compensation instruction.

[0011] As a further description of the above technical solution: the opening characteristic data of the valve body include the model of the valve body, the medium temperature, the pressure of the pipeline, the medium flow rate, the medium flow rate and the medium density, wherein the medium temperature, the pressure of the pipeline, the medium flow rate, the medium flow rate and the medium density are respectively collected and acquired in real time by the set temperature sensor, pressure sensor, flow sensor, flow rate sensor and electromagnetic density meter.

[0012] As a further description of the above technical solution: the method for constructing the first machine learning model includes:

[0013] Initialize the first machine learning model structure. The first machine learning model structure adopts an MLP type multi-layer forward network structure. The MLP type multi-layer forward network structure includes two input layers, two hidden layers and one output layer. The input layer includes the first input layer and the second input layer. The number of nodes in the first input layer is [1, n], corresponding to the valve body model, n is the number of valve body models, and the number of nodes in the second input layer is 5, corresponding to the medium temperature, pipeline pressure, medium flow, medium flow velocity and medium density respectively; the hidden layer includes the first hidden layer and the second hidden layer. The number of nodes in the first hidden layer of the first hidden layer is 128, and the Relu function is used as the activation function; the number of nodes in the second hidden layer of the second hidden layer is 64, and the Relu function is used as the activation function; the output layer is the first output layer, the number of nodes in the first output layer is 1, and the valve opening data is predicted;

[0014] After initializing the first machine learning model structure, the opening feature data and the valve opening data corresponding to the opening feature data are used to train the first machine learning model. The optimizer is Adam, the loss function is the MSE loss function, the batch size is set to 32, the number of iterations is 200 rounds, and the training is terminated when the loss function converges.

[0015] As a further description of the above technical solution: the first machine learning model training method includes:

[0016] Convert the opening characteristic data into a corresponding set of characteristic vectors;

[0017] The opening characteristic data is used as the input of the first machine learning model. The first machine learning model takes a group of valve opening data corresponding to each group of opening characteristic data as output, takes the valve opening data actually corresponding to each group of opening characteristic data as the prediction target, and takes minimizing the loss function value of the first machine learning model as the training target; training is stopped when the loss function value of the first machine learning model is less than or equal to the preset target loss value.

[0018] As a further description of the above technical solution: the pipeline vibration data collection method includes:

[0019] With the valve body as the center and the preset radius R, draw a circle and mark the pipes in the circular area as the vibration detection area;

[0020] Select the pipeline support points, pipeline elbows, joints or near the vibration source in the vibration detection area to install accelerometers for data collection and obtain time domain signals;

[0021] The time domain signal is converted into a frequency domain signal through fast Fourier transform to obtain the amplitude and angular frequency of pipeline vibration.

[0022] As a further description of the above technical solution: the method for collecting medium fluctuation data includes:

[0023] Install an ultrasonic sensor at the location where the valve water inlet is connected to the pipeline to collect data and obtain time domain signals;

[0024] Use fast Fourier transform to convert the time domain signal into frequency domain signal to obtain the amplitude and angular frequency of medium fluctuation.

[0025] As a further description of the above technical solution: the method for generating a valve opening compensation instruction based on pipeline vibration data, medium fluctuation data, and the cross-influence of pipeline vibration and medium fluctuation includes:

[0026] generating a first opening compensation amount based on pipeline vibration data;

[0027] generating a second opening compensation amount based on the medium fluctuation data;

[0028] Generate a third opening compensation amount based on the cross-influence of pipeline vibration and medium fluctuation;

[0029] A comprehensive opening compensation amount is generated based on the first opening compensation amount, the second opening compensation amount and the third opening compensation amount, a valve opening compensation instruction is generated according to the comprehensive opening compensation amount, and the opening of the valve is adaptively adjusted.

[0030] As a further description of the above technical solution: the method for obtaining the first opening compensation amount includes:

[0031] Δθvib=Kvib×Avib×sin(ωvib×t);

[0032] In the formula, Δθvib is the first opening compensation, Kvib is the sensitivity coefficient of pipeline vibration to the valve, which indicates the degree of valve response to pipeline vibration, Avib is the amplitude of pipeline vibration, ωvib is the angular frequency of vibration, and t is the data acquisition time;

[0033] The method for obtaining the second opening compensation amount includes:

[0034] Δθflow=Kflow×Aflow×sin(ωflow×t);

[0035] In the formula, Δθflow is the second opening compensation, Kflow is the sensitivity coefficient of the medium fluctuation to the valve, which indicates the degree of valve response to the medium fluctuation, Aflow is the amplitude of the medium fluctuation, ωflow is the angular frequency of the medium fluctuation, and t is the data acquisition time;

[0036] The method for obtaining the third opening compensation amount includes:

[0037] Δθcross=Kv-f×Avib×Aflow×cos(φv-f)×sin(ωv-f);

[0038] Wherein, Δθcross is the third opening compensation, Kv-f is the coupling coefficient of the cross-action of pipeline vibration and medium fluctuation, which indicates the influence of the combined action of pipeline vibration and medium fluctuation on the valve opening; φv-f is the phase difference between pipeline vibration and medium fluctuation, which indicates the time delay or synchronization of the interaction between pipeline vibration and medium fluctuation, and ωv-f is the frequency of the cross-effect of pipeline vibration and medium fluctuation.

[0039] As a further description of the above technical solution: the calculation formula of the comprehensive opening compensation amount is:

[0040] Δθtotal=Δθvib+Δθflow+Δθcross;

[0041] Wherein, Δθtotal is the comprehensive opening compensation, Δθvib is the first opening compensation, Δθflow is the second opening compensation, and Δθcross is the third opening compensation.

[0042] An adaptive valve control system applicable to complex working conditions implements the adaptive valve control method applicable to complex working conditions, and the system comprises:

[0043] A first data acquisition module is provided, wherein a sensor group is installed on the installation pipe of the valve body, and the opening characteristic data of the valve body is collected in real time through the sensor group;

[0044] A model training module inputs the collected opening characteristic data into a pre-built first machine learning model, outputs valve opening data, and generates a valve body opening instruction based on the valve opening data to control the valve opening;

[0045] The second data acquisition module collects pipeline vibration data and medium fluctuation data in real time, wherein the pipeline vibration data includes vibration amplitude and vibration angular frequency, and the medium fluctuation data includes fluctuation amplitude and fluctuation angular frequency;

[0046] The data analysis module generates valve opening compensation instructions based on pipeline vibration data, medium fluctuation data, and the cross-effect of pipeline vibration and medium fluctuation;

[0047] The execution module adaptively adjusts the valve opening based on the valve opening compensation instruction.

[0048] Beneficial effects:

[0049] In the above technical scheme, the adaptive valve control method suitable for complex working conditions provided by the present invention collects the opening characteristic data of the valve body in real time. The opening characteristic data of the valve body includes the model of the valve body, the medium temperature, the pressure of the pipeline, the medium flow rate, the medium flow rate and the medium density. The first machine learning model can more accurately predict the opening of the valve, avoiding the accuracy problem caused by the single regulation method relying on flow and pressure. Secondly, the opening characteristic data of the valve body is collected in real time and input into the pre-built first machine learning model, and the valve opening data is output. Based on the valve opening data, a valve body opening instruction is generated to control the opening of the valve, thereby realizing real-time adaptive adjustment of the valve opening. The valve can automatically adjust the opening to adapt to different working conditions to ensure that the system always operates stably in a changing environment.

[0050] Furthermore, when the valve body is in use, pipeline vibration data and medium fluctuation data are also collected in real time. Valve opening compensation instructions are generated based on the pipeline vibration data, medium fluctuation data, and the cross-influence of pipeline vibration and medium fluctuation. By introducing pipeline vibration data and medium fluctuation data and combining the cross-influence between the two, valve opening compensation instructions are generated, which can effectively compensate for errors caused by vibration or fluctuation and ensure that the valve response is more accurate to meet complex working conditions. Secondly, combined with the self-learning ability of the first machine learning model, the system can continuously optimize the compensation algorithm to adapt to a wider range of working conditions. It not only considers the independent effects of pipeline vibration and medium fluctuation, but also includes the mutual influence between them. It can more comprehensively consider the influencing factors of valve opening, further improve the accuracy of valve opening, and improve the stability of the entire system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention will be further explained below in conjunction with the accompanying drawings and embodiments:

[0052] Figure 1 A flow chart of an adaptive valve control method applicable to complex working conditions provided by an embodiment of the present invention;

[0053] Figure 2 A flow chart of a method for generating a valve opening compensation instruction based on pipeline vibration data, medium fluctuation data, and the cross-influence of pipeline vibration and medium fluctuation provided in an embodiment of the present invention;

[0054] Figure 3 A flow chart of a pipeline vibration data collection method provided by an embodiment of the present invention;

[0055] Figure 4 A module connection diagram of an adaptive valve control system suitable for complex working conditions provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the technical means, creative features, objectives and effects of the present invention easy to understand, the present invention is further described below with reference to specific diagrams. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0057] Example 1

[0058] See also Figure 1 The embodiment of the present invention provides a technical solution: an adaptive valve control method suitable for complex working conditions, specifically comprising the following steps:

[0059] Step S01: installing a sensor group on the installation pipe of the valve body, and collecting the opening characteristic data of the valve body in real time through the sensor group;

[0060] The opening characteristic data of the valve body include the model of the valve body, the medium temperature, the pressure of the pipeline, the medium flow rate, the medium flow rate and the medium density, wherein the medium temperature, the pressure of the pipeline, the medium flow rate, the medium flow rate and the medium density are respectively collected and acquired in real time by the set temperature sensor, pressure sensor, flow sensor, flow rate sensor and electromagnetic density meter.

[0061] Step S02: inputting the collected opening characteristic data into a pre-built first machine learning model, outputting valve opening data, and generating a valve body opening instruction based on the valve opening data to control the opening of the valve;

[0062] The method for constructing the first machine learning model includes:

[0063] Initialize the first machine learning model structure. The first machine learning model structure adopts an MLP type multi-layer forward network structure. The MLP type multi-layer forward network structure includes two input layers, two hidden layers and one output layer. The input layer includes the first input layer and the second input layer. The number of nodes in the first input layer is [1, n], corresponding to the valve body model, n is the number of valve body models, and the number of nodes in the second input layer is 5, corresponding to the medium temperature, pipeline pressure, medium flow, medium flow velocity and medium density respectively; the hidden layer includes the first hidden layer and the second hidden layer. The number of nodes in the first hidden layer of the first hidden layer is 128, and the Relu function is used as the activation function; the number of nodes in the second hidden layer of the second hidden layer is 64, and the Relu function is used as the activation function; the output layer is the first output layer, the number of nodes in the first output layer is 1, and the valve opening data is predicted;

[0064] After initializing the first machine learning model structure, the opening feature data and the valve opening data corresponding to the opening feature data are used to train the first machine learning model. The optimizer is Adam, the loss function is the MSE loss function, the batch size is set to 32, the number of iterations is 200 rounds, and the training is terminated when the loss function converges.

[0065] It should be noted that the Adam optimizer is one of the five major optimizers commonly used in machine learning, and its full name is Adaptive Moment Estimation.

[0066] The first machine learning model training method comprises:

[0067] The opening characteristic data is converted into a corresponding set of characteristic vectors.

[0068] The opening characteristic data is used as the input of the first machine learning model. The first machine learning model takes a group of valve opening data corresponding to each group of opening characteristic data as output, takes the valve opening data actually corresponding to each group of opening characteristic data as the prediction target, and takes minimizing the loss function value of the first machine learning model as the training target; training is stopped when the loss function value of the first machine learning model is less than or equal to the preset target loss value.

[0069] The loss function value of the first machine learning model is the mean square error;

[0070] Mean square error is one of the commonly used loss functions. By transforming the loss function formula The model is trained with minimization as the goal, so that the first machine learning model fits the data better, thereby improving the performance and accuracy of the model.

[0071] In the loss function, MSE is the loss function value of the first machine learning model, x is the feature vector group number; e is the number of feature vector groups; y is the loss function value of the first machine learning model, x is the number of feature vector groups ...x is the valve opening data output by the first machine learning model corresponding to the xth group of feature vectors, It is the valve opening data actually corresponding to the xth group of characteristic vectors.

[0072] It is particularly noteworthy that, since this is strategic data training, the optimal decision control group of the initial sample of valve opening data corresponding to the opening characteristic data is obtained by manually making the optimal decision and recording it after manual experiments. At this stage, the initial stage of decision-making training data requires manual decision-making, and machine learning is obtained by logical judgment and classification summary based on manual decision-making.

[0073] In this embodiment, by collecting the opening characteristic data of the valve body in real time, the opening characteristic data of the valve body includes the model of the valve body, the medium temperature, the pressure of the pipeline, the medium flow rate, the medium flow rate and the medium density. The first machine learning model can more accurately predict the opening of the valve, avoiding the accuracy problem caused by the single regulation method relying on flow and pressure. Secondly, by collecting the opening characteristic data of the valve body in real time, inputting it into the pre-built first machine learning model, outputting the valve opening data, and generating a valve body opening instruction based on the valve opening data to control the opening of the valve, thereby realizing real-time adaptive adjustment of the valve opening. The valve can automatically adjust the opening to adapt to different working conditions, ensuring that the system always operates stably in a changing environment.

[0074] Example 2

[0075] See also Figures 1 to 3 , step S03: collecting pipeline vibration data and medium fluctuation data in real time, wherein the pipeline vibration data includes vibration amplitude and vibration angular frequency, and the medium fluctuation data includes fluctuation amplitude and fluctuation angular frequency;

[0076] The pipeline vibration data collection method comprises:

[0077] With the valve body as the center and the preset radius R, draw a circle and mark the pipes in the circular area as the vibration detection area;

[0078] Select the pipeline support points, pipeline elbows, joints or near the vibration source in the vibration detection area to install accelerometers for data collection and obtain time domain signals;

[0079] It should be noted that the vibration sources include: pump body, compressor, etc.

[0080] The time domain signal is converted into a frequency domain signal through fast Fourier transform to obtain the amplitude and angular frequency of pipeline vibration.

[0081] The method for collecting medium fluctuation data comprises:

[0082] Install an ultrasonic sensor at the location where the valve water inlet is connected to the pipeline to collect data and obtain time domain signals;

[0083] Use fast Fourier transform to convert the time domain signal into frequency domain signal to obtain the amplitude and angular frequency of medium fluctuation.

[0084] Step S04: Generate a valve opening compensation instruction based on the pipeline vibration data, the medium fluctuation data, and the cross-influence of the pipeline vibration and the medium fluctuation, and adaptively adjust the valve opening based on the valve opening compensation instruction.

[0085] The method for generating a valve opening compensation instruction based on pipeline vibration data, medium fluctuation data, and the cross-influence of pipeline vibration and medium fluctuation includes:

[0086] A first opening compensation amount is generated based on pipeline vibration data, and a method for obtaining the first opening compensation amount includes:

[0087] Δθvib=Kvib×Avib×sin(ωvib×t);

[0088] In the formula, Δθvib is the first opening compensation, Kvib is the sensitivity coefficient of pipeline vibration to the valve, which indicates the degree of valve response to pipeline vibration, Avib is the amplitude of pipeline vibration, ωvib is the angular frequency of vibration, and t is the data acquisition time;

[0089] It should be noted that the calculation formula for angular frequency is: ω=2πf, where f is the frequency of the vibration signal (in Hertz), and sin(ωvib×t) is used to describe periodic vibrations that change over time. Its amplitude and frequency determine the size and period of the vibration signal.

[0090] A second opening compensation amount is generated based on the medium fluctuation data, and a method for obtaining the second opening compensation amount includes:

[0091] Δθflow=Kflow×Aflow×sin(ωflow×t);

[0092] In the formula, Δθflow is the second opening compensation, Kflow is the sensitivity coefficient of the medium fluctuation to the valve, which indicates the degree of valve response to the medium fluctuation, Aflow is the amplitude of the medium fluctuation, ωflow is the angular frequency of the medium fluctuation, and t is the data acquisition time;

[0093] It should be noted that the pipeline vibration data and the medium fluctuation data are collected synchronously, that is, the data collection time of the two is the same.

[0094] A third opening compensation amount is generated based on the cross-influence of pipeline vibration and medium fluctuation, and a method for obtaining the third opening compensation amount includes:

[0095] Δθcross=Kv-f×Avib×Aflow×cos(φv-f)×sin(ωv-f);

[0096] Wherein, Δθcross is the third opening compensation, Kv-f is the coupling coefficient of the cross-action of pipeline vibration and medium fluctuation, which indicates the influence of the combined action of pipeline vibration and medium fluctuation on the valve opening; φv-f is the phase difference between pipeline vibration and medium fluctuation, which indicates the time delay or synchronization of the interaction between pipeline vibration and medium fluctuation, and ωv-f is the frequency of the cross-effect of pipeline vibration and medium fluctuation;

[0097] It should be noted that Kvib, Kflow and Kv-f are determined by those skilled in the art based on a large number of experiments.

[0098] A comprehensive opening compensation amount is generated based on the first opening compensation amount, the second opening compensation amount and the third opening compensation amount. The calculation formula of the comprehensive opening compensation amount is:

[0099] Δθtotal=Δθvib+Δθflow+Δθcross;

[0100] Wherein, Δθtotal is the comprehensive opening compensation, Δθvib is the first opening compensation, Δθflow is the second opening compensation, and Δθcross is the third opening compensation.

[0101] A valve opening compensation instruction is generated according to the comprehensive opening compensation amount, and the opening of the valve body is adaptively adjusted.

[0102] In this embodiment, on the basis of the above embodiment, when the valve body is in use, pipeline vibration data and medium fluctuation data are also collected in real time, and valve opening compensation instructions are generated based on the pipeline vibration data, medium fluctuation data, and the cross-influence of pipeline vibration and medium fluctuation. By introducing pipeline vibration data and medium fluctuation data and combining the cross-influence between the two, valve opening compensation instructions are generated, which can effectively compensate for errors caused by vibration or fluctuation and ensure that the valve response is more accurate to meet complex working conditions. Secondly, combined with the self-learning ability of the first machine learning model, the system can continuously optimize the compensation algorithm to adapt to a wider range of working conditions. It not only considers the independent effects of pipeline vibration and medium fluctuation, but also includes the mutual influence between them. It can more comprehensively consider the influencing factors of valve opening, further improve the accuracy of valve opening, and improve the stability of the entire system operation.

[0103] Example 3

[0104] See also Figure 4 , an adaptive valve control system applicable to complex working conditions, implementing the adaptive valve control method applicable to complex working conditions, the system comprises:

[0105] A first data acquisition module is provided, wherein a sensor group is installed on the installation pipe of the valve body, and the opening characteristic data of the valve body is collected in real time through the sensor group;

[0106] A model training module inputs the collected opening characteristic data into a pre-built first machine learning model, outputs valve opening data, and generates a valve body opening instruction based on the valve opening data to control the valve opening;

[0107] The second data acquisition module collects pipeline vibration data and medium fluctuation data in real time, wherein the pipeline vibration data includes vibration amplitude and vibration angular frequency, and the medium fluctuation data includes fluctuation amplitude and fluctuation angular frequency;

[0108] The data analysis module generates valve opening compensation instructions based on pipeline vibration data, medium fluctuation data, and the cross-effect of pipeline vibration and medium fluctuation;

[0109] The execution module adaptively adjusts the valve opening based on the valve opening compensation instruction.

[0110] The basic principles, main features and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and the description in the specification are only to illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An adaptive valve control method suitable for complex working conditions, characterized in that: include: Step S01: installing a sensor group on the installation pipe of the valve body, and collecting the opening characteristic data of the valve body in real time through the sensor group; Step S02: inputting the collected opening characteristic data into a pre-built first machine learning model, outputting valve opening data, and generating a valve body opening instruction based on the valve opening data to control the opening of the valve; Step S03: collecting pipeline vibration data and medium fluctuation data in real time, wherein the pipeline vibration data includes vibration amplitude and vibration angular frequency, and the medium fluctuation data includes fluctuation amplitude and fluctuation angular frequency; Step S04: Generate a valve opening compensation instruction based on the pipeline vibration data, the medium fluctuation data, and the cross-influence of the pipeline vibration and the medium fluctuation, and adaptively adjust the valve opening based on the valve opening compensation instruction.

2. The adaptive valve control method applicable to complex working conditions according to claim 1, characterized in that: The opening characteristic data of the valve body include the model of the valve body, the medium temperature, the pressure of the pipeline, the medium flow rate, the medium flow rate and the medium density, wherein the medium temperature, the pressure of the pipeline, the medium flow rate, the medium flow rate and the medium density are respectively acquired in real time by the set temperature sensor, pressure sensor, flow sensor, flow rate sensor and electromagnetic density meter.

3. The adaptive valve control method applicable to complex working conditions according to claim 1, characterized in that: The method for constructing the first machine learning model includes: Initialize the first machine learning model structure. The first machine learning model structure adopts an MLP type multi-layer forward network structure. The MLP type multi-layer forward network structure includes two input layers, two hidden layers and one output layer. The input layer includes the first input layer and the second input layer. The number of nodes in the first input layer is [1, n], corresponding to the valve body model, n is the number of valve body models, and the number of nodes in the second input layer is 5, corresponding to the medium temperature, pipeline pressure, medium flow, medium flow velocity and medium density respectively; the hidden layer includes the first hidden layer and the second hidden layer. The number of nodes in the first hidden layer of the first hidden layer is 128, and the Relu function is used as the activation function; the number of nodes in the second hidden layer of the second hidden layer is 64, and the Relu function is used as the activation function; the output layer is the first output layer, the number of nodes in the first output layer is 1, and the valve opening data is predicted; After initializing the first machine learning model structure, the opening feature data and the valve opening data corresponding to the opening feature data are used to train the first machine learning model. The optimizer is Adam, the loss function is the MSE loss function, the batch size is set to 32, the number of iterations is 200 rounds, and the training is terminated when the loss function converges.

4. The adaptive valve control method applicable to complex working conditions according to claim 3, characterized in that: The first machine learning model training method comprises: Convert the opening characteristic data into a corresponding set of characteristic vectors; The opening characteristic data is used as the input of the first machine learning model. The first machine learning model takes a group of valve opening data corresponding to each group of opening characteristic data as output, takes the valve opening data actually corresponding to each group of opening characteristic data as the prediction target, and takes minimizing the loss function value of the first machine learning model as the training target; training is stopped when the loss function value of the first machine learning model is less than or equal to the preset target loss value.

5. The adaptive valve control method applicable to complex working conditions according to claim 1, characterized in that: The pipeline vibration data collection method comprises: With the valve body as the center and the preset radius R, draw a circle and mark the pipes in the circular area as the vibration detection area; Select the pipeline support points, pipeline elbows, joints or near the vibration source in the vibration detection area to install accelerometers for data collection and obtain time domain signals; The time domain signal is converted into a frequency domain signal through fast Fourier transform to obtain the amplitude and angular frequency of pipeline vibration.

6. The adaptive valve control method applicable to complex working conditions according to claim 5, characterized in that: The method for collecting medium fluctuation data comprises: Install an ultrasonic sensor at the location where the valve water inlet is connected to the pipeline to collect data and obtain time domain signals; Use fast Fourier transform to convert the time domain signal into frequency domain signal to obtain the amplitude and angular frequency of medium fluctuation.

7. The adaptive valve control method applicable to complex working conditions according to claim 6, characterized in that: The method for generating a valve opening compensation instruction based on pipeline vibration data, medium fluctuation data, and the cross-influence of pipeline vibration and medium fluctuation includes: generating a first opening compensation amount based on pipeline vibration data; generating a second opening compensation amount based on the medium fluctuation data; Generate a third opening compensation amount based on the cross-influence of pipeline vibration and medium fluctuation; A comprehensive opening compensation amount is generated based on the first opening compensation amount, the second opening compensation amount and the third opening compensation amount, a valve opening compensation instruction is generated according to the comprehensive opening compensation amount, and the opening of the valve is adaptively adjusted.

8. The adaptive valve control method applicable to complex working conditions according to claim 7, characterized in that: The method for obtaining the first opening compensation amount includes: Δθvib=Kvib×Avib×sin(ωvib×t); In the formula, Δθvib is the first opening compensation, Kvib is the sensitivity coefficient of pipeline vibration to the valve, which indicates the degree of valve response to pipeline vibration, Avib is the amplitude of pipeline vibration, ωvib is the angular frequency of vibration, and t is the data acquisition time; The method for obtaining the second opening compensation amount includes: Δθflow=Kflow×Aflow×sin(ωflow×t); In the formula, Δθflow is the second opening compensation, Kflow is the sensitivity coefficient of the medium fluctuation to the valve, which indicates the degree of valve response to the medium fluctuation, Aflow is the amplitude of the medium fluctuation, ωflow is the angular frequency of the medium fluctuation, and t is the data acquisition time; The method for obtaining the third opening compensation amount includes: Δθcross=Kv-f×Avib×Aflow×cos(φv-f)×sin(ωv-f); Wherein, Δθcross is the third opening compensation, Kv-f is the coupling coefficient of the cross-action of pipeline vibration and medium fluctuation, which indicates the influence of the combined action of pipeline vibration and medium fluctuation on the valve opening; φv-f is the phase difference between pipeline vibration and medium fluctuation, which indicates the time delay or synchronization of the interaction between pipeline vibration and medium fluctuation, and ωv-f is the frequency of the cross-effect of pipeline vibration and medium fluctuation.

9. The adaptive valve control method applicable to complex working conditions according to claim 7, characterized in that: The calculation formula of the comprehensive opening compensation is: Δθtotal=Δθvib+Δθflow+Δθcross; Wherein, Δθtotal is the comprehensive opening compensation, Δθvib is the first opening compensation, Δθflow is the second opening compensation, and Δθcross is the third opening compensation.

10. Adaptive valve control system suitable for complex working conditions, characterized by: The adaptive valve control method applicable to complex working conditions according to any one of claims 1 to 9 is implemented, and the system comprises: A first data acquisition module is provided, wherein a sensor group is installed on the installation pipe of the valve body, and the opening characteristic data of the valve body is collected in real time through the sensor group; A model training module inputs the collected opening characteristic data into a pre-built first machine learning model, outputs valve opening data, and generates a valve body opening instruction based on the valve opening data to control the valve opening; The second data acquisition module collects pipeline vibration data and medium fluctuation data in real time, wherein the pipeline vibration data includes vibration amplitude and vibration angular frequency, and the medium fluctuation data includes fluctuation amplitude and fluctuation angular frequency; The data analysis module generates valve opening compensation instructions based on pipeline vibration data, medium fluctuation data, and the cross-effect of pipeline vibration and medium fluctuation; The execution module adaptively adjusts the valve opening based on the valve opening compensation instruction.

Images