Robust driving control method for hydrogen energy flow meter based on composite learning
By establishing a dynamic model of the hydrogen flowmeter and designing a composite learning strategy for neural networks, combining perturbation observer and non-singular terminal sliding mode control, the problem of insufficient robust control of the PI control algorithm under harsh working conditions is solved, and high Accurate hydrogen flowmeter control.
Patent Information
- Application Number
- CN202411149181.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-12-03
AI Technical Summary
In the prior art, PI control algorithms cannot effectively control hydrogen flowmeters under harsh operating conditions, especially in the case of temperature changes and complex vibration environments, and it is difficult to meet the requirements of robust control.
Using a composite learning method, a vibration tube dynamic model of a hydrogen-energy Coriolis mass flowmeter is established, a neural network composite learning strategy is designed to approximate unknown nonlinear dynamics in a variable temperature environment, and mechanical vibration interference is estimated through perturbation observers. , combined with non-singular terminal slip mode control, feedforward unknown dynamics and interference estimation to achieve robust control.
High-precision control under harsh working conditions is achieved, the influence of temperature and vibration on the flowmeter is suppressed, and the robust driving control needs of the flowmeter are met.
Smart Images

Figure CN119065243A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of hydrogen energy flow meter drive control, and in particular relates to a robust drive control method for a hydrogen energy flow meter based on composite learning. Background Art
[0002] The hydrogen flowmeter is the core component of the 70MPa hydrogenation machine calibration device, and its measurement accuracy is the basis for ensuring the measurement reliability of the calibration device. The hydrogen flowmeter has a complex working environment and must cope with the test of factors such as temperature and pipeline vibration.
[0003] At present, some studies have been conducted to deal with temperature and vibration environments through precise modeling and vibration isolation installation. The influence of temperature effect on flowmeter accuracy is reflected in the nonlinear changes in the elastic modulus and Poisson's ratio of the measuring tube material caused by temperature changes. The article "Analysis of Temperature Influence Characteristics of U-Type Coriolis Mass Flowmeter" (Yu Lianyan, Pei Xiangxiang, Li Xiang, Zhang Xiaobin, "Instrument Technology and Sensor", Issue 2, 2024) established an accurate mathematical model of the flowmeter and introduced a correction coefficient into the model to weaken the influence of temperature. At the same time, the vibration of the upstream and downstream pipelines, or the vibration of other flowmeters during operation, causes the vibration tube of the flowmeter to receive vibration, thereby affecting the metering accuracy of the flowmeter. The article "Analysis of Factors Affecting the Accuracy of Mass Flowmeters in LNG Measurement" (Jiang Jun, Gong Li, Li Wenyi, Lu Longxing, He Wei, "Chemical Management", Issue 32, 2023) points out that staying away from the vibration source is an important means to eliminate mechanical vibration.
[0004] However, these studies on flow meter measurement accuracy do not consider the impact of environmental factors on the flow meter drive control system. The drive control system is the core component of the flow meter, and its control performance directly determines the measurement accuracy. At present, the PI control algorithm is usually used in engineering to improve the measurement accuracy of the flow meter, but it can only achieve good accuracy when measuring a relatively stable flow rate. Because it cannot adjust parameters online, it is difficult to predict complex measurement environments such as temperature changes and vibrations. Its comprehensive control effect is not ideal and can no longer meet the robust control requirements under harsh working conditions. Summary of the invention
[0005] The purpose of the present invention is to solve the problem in the prior art that the hydrogen energy flow meter drive control system controlled by using PI control algorithm cannot meet the robust control requirements under harsh working conditions, and to provide a robust drive control method for a hydrogen energy flow meter based on composite learning.
[0006] To achieve the above purpose, the technical solution provided by the present invention is:
[0007] A robust drive control method for a hydrogen energy flow meter based on composite learning comprises the following steps:
[0008] Step 1, establishing a dynamic model of a vibrating tube of a hydrogen energy Coriolis mass flowmeter;
[0009] Step 2, design a neural network composite learning strategy to approximate the unknown nonlinear dynamics of the flow meter drive control system caused by the variable temperature environment, and design a disturbance observer to estimate the external disturbance of the system caused by mechanical vibration;
[0010] Step 3, based on non-singular terminal sliding mode control, determine the electrostatic driving force generated by the driving coil of the driver in the flowmeter in the dynamic model established in step 1, and feed forward the unknown nonlinear dynamics and external interference obtained in step 2 to the determined electrostatic driving force, so as to realize robust control of the hydrogen energy flowmeter under harsh working conditions.
[0011] Furthermore, in step 1, the kinetic model is established as:
[0012]
[0013] In the formula, and x are the vibration acceleration, vibration velocity and vibration displacement of the flowmeter vibration tube, respectively, m is the mass of the flowmeter vibration system, c and k are the damping coefficient and stiffness coefficient of the vibration tube, respectively, d(t) is the vibration received by the flowmeter vibration tube, and f is the electrostatic driving force generated by the driving coil.
[0014] Furthermore, step 2 includes the following sub-steps:
[0015] Step 2.1, define the unknown nonlinear function under variable temperature environment:
[0016] Step 2.2, use a neural network to approximate the unknown nonlinear function ρ:
[0017]
[0018] In the formula, is the estimated value of ρ, is the n-dimensional adjustable neural network weight vector, is the n-dimensional basis vector;
[0019] Step 2.3, define the model prediction error:
[0020]
[0021] In the formula, for Estimates:
[0022]
[0023] In the formula, β is a constant, and β>0, is the estimated value of the composite external disturbance D, ε is the neural network approximation error;
[0024] Step 2.4, design the neural network weight vector update law:
[0025]
[0026] Where α1, α2 and α3, as well as p and r are constants, and α1>0, α2>0, α3>0, p>0, 1<r<2, s is the non-singular terminal sliding surface;
[0027] Step 2.5: Design a disturbance observer to obtain an estimate of the external disturbance
[0028]
[0029] Where L is a constant and L>0, is the intermediate variable, and e is the tracking error of the drive control system.
[0030] Furthermore, step 3 includes the following sub-steps:
[0031] Step 3.1, define the tracking error e of the drive control system:
[0032] e=xx r
[0033] In the formula, x r is the reference trajectory of the vibration displacement of the vibrating tube, A r is the amplitude of the reference trajectory, is the frequency of the reference trajectory;
[0034] Step 3.2, define the non-singular terminal sliding surface:
[0035] Step 3.3, design the drive controller:
[0036]
[0037] Where κ1 and κ2 are constants, κ1>0, κ2>0, and sign(·) is the sign function.
[0038] The advantages of the present invention are:
[0039] The robust drive control method of the hydrogen energy flowmeter based on composite learning of the present invention establishes a dynamic model for the vibration tube of the hydrogen energy flowmeter, constructs a parallel estimation model of the dynamic model for the unknown nonlinear dynamics of the system caused by temperature, designs a neural network composite learning strategy in combination with the model prediction error and the tracking control error, estimates the unknown nonlinear dynamics online, designs a disturbance observer to estimate the external interference caused by mechanical vibration, and feeds forward the unknown dynamics estimate and the interference estimate to the control system based on non-singular terminal sliding mode control. Therefore, the present invention designs an intelligent learning strategy based on composite learning to cope with the unknown nonlinear dynamics of the system caused by the variable temperature environment, designs a disturbance observer to cope with the external interference caused by mechanical vibration, suppresses the influence of temperature and vibration by feedforward compensating the unknown dynamics estimate and the interference estimate in the control system, and then realizes the robust drive control of the flowmeter under harsh working conditions, and meets the high-precision control requirements of the flowmeter. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] The above and / or other features and advantages of the present invention will become more easily understood through the following description with reference to the accompanying drawings, in which:
[0041] Figure 1 It is a schematic diagram of the robust drive control system of the hydrogen energy flow meter based on composite learning of the present invention;
[0042] Figure 2 It is a flow chart of the robust drive control method of the hydrogen energy flow meter based on composite learning of the present invention. DETAILED DESCRIPTION
[0043] The present invention will be described in detail below with reference to the accompanying drawings by means of exemplary embodiments of the present invention. It should be noted that the following detailed description of the present invention is only for the purpose of illustration, and is not intended to limit the present invention.
[0044] The present invention provides a robust drive control method for a hydrogen energy flowmeter based on composite learning, which is used to control a drive control system of a hydrogen energy Coriolis mass flowmeter and realize high-precision control of the control system under harsh working conditions.
[0045] Reference Figure 1The hydrogen energy flow meter drive control system includes three links: vibration tube, digital-to-analog conversion module, and digital drive control module. Among them, the hydrogen energy flow meter vibration tube is the control object; the digital-to-analog conversion link amplifies the driving force f and the detection displacement signal x and performs digital-to-analog and analog-to-digital conversion; the digital drive link implements the intelligent robust drive control algorithm of the hydrogen energy flow meter based on composite learning. The specific idea is: give the dynamic model of the hydrogen energy flow meter vibration tube; consider the unknown nonlinear dynamics of the system caused by temperature, build a parallel estimation model of the dynamic model, combine the model prediction error and tracking control error to design a neural network composite learning strategy, and estimate the unknown nonlinear dynamics online; consider the external interference caused by mechanical vibration, and design a disturbance observer to estimate it; by feedforward compensation of unknown dynamics estimation and interference estimation in the control system, suppress the influence of temperature and vibration, and achieve high-precision control of the flow meter under harsh working conditions.
[0046] Reference Figure 2 As an exemplary embodiment of the present invention, a robust driving control method for a hydrogen energy flow meter based on composite learning includes:
[0047] Step S1, establishing a dynamic model of a vibrating tube of a hydrogen energy Coriolis mass flowmeter;
[0048] Step S2, designing a neural network composite learning strategy to approximate the unknown nonlinear dynamics of the flow meter drive control system caused by the variable temperature environment, and designing a disturbance observer to estimate the external disturbance of the system caused by mechanical vibration;
[0049] Step S3, based on non-singular terminal sliding mode control, determines the electrostatic driving force generated by the driving coil of the driver in the flowmeter in the dynamic model established in step S1, and feeds forward the unknown nonlinear dynamics and external interference obtained in step S2 to the determined electrostatic driving force, so as to realize robust control of the hydrogen energy flowmeter under harsh working conditions.
[0050] In an exemplary embodiment of the present invention, step S1 includes: establishing a dynamic model of the flow meter as
[0051]
[0052] In the formula, and x are the vibration acceleration, vibration velocity and vibration displacement of the flowmeter vibration tube respectively; m is the mass of the flowmeter vibration system; c and k are the damping coefficient and stiffness coefficient of the vibration tube respectively, and their values are related to the structural parameters and dynamic characteristics of the flowmeter; d(t) is the vibration received by the flowmeter vibration tube, which can be selected as d(t)=10+2sin(5.8t); f is the electrostatic driving force generated by the driving coil.
[0053] Dividing both sides of equation (1) by m yields:
[0054]
[0055] The ambient temperature affects the material parameters (elastic modulus, Poisson's ratio and linear expansion coefficient) and geometric dimensions of the vibration tube, making the resonant frequency ω of the vibration tube n As a result, the stiffness coefficient and damping coefficient c = -2mξω n , Change, where ξ is the damping ratio of the vibration tube. Under normal temperature, m=5.5kg, ξ=0.0005, ω n =2π×100rad / s.
[0056] Step S2 is used to perform unknown nonlinear approximation and interference estimation under severe working conditions, and specifically includes:
[0057] definition From the above analysis, it can be seen that ρ is an unknown nonlinear function under variable temperature environment.
[0058] Using neural network to approximate the unknown nonlinear function ρ, we have:
[0059]
[0060] in, is the estimated value of ρ; is an n-dimensional adjustable neural network weight vector, preferably, is a 100-dimensional adjustable neural network weight vector; is the n-dimensional basis vector.
[0061] When the neural network weights are optimal, there exists:
[0062] ρ=ω T θ+ε (4)
[0063] Among them, ε is the neural network approximation error; is the 100-dimensional optimal weight vector; is a 100-dimensional basis vector, and the i-th (i=1,2,…100) element of the vector is a Gaussian function:
[0064]
[0065] In the formula, is the input vector of the neural network; is the Gaussian function θ i The center of , whose value can be arbitrarily selected between [-50 50]×[-5 5]; σ i is the Gaussian function θ i The standard deviation of i =1.
[0066] The model prediction error is defined as:
[0067]
[0068] in, for The estimate of is obtained by integrating the following parallel estimation model:
[0069]
[0070] Where, β = 0.8, For composite external disturbance The estimated value of .
[0071] The update law of the neural network weight vector is designed as:
[0072]
[0073] Among them, α1>0, α2>0 and α3>0 are constants, and the possible values are α1=400, α2=10, α3=1.2; p>0 and 1<r<2 are constants, and the possible values are p=80, r=1.2; s is the non-singular terminal sliding surface, which will be described in detail below.
[0074] It is obtained by the following disturbance observer:
[0075]
[0076] Among them, L>0 is a constant, which can be selected as L=20; is the intermediate variable; e is the tracking error of the drive control system, and its definition will be given below.
[0077] Step S3 designs a robust drive controller, which may include:
[0078] Define the reference trajectory x of the flowmeter vibrating tube vibration displacement r for
[0079]
[0080] Among them, A r is the amplitude of the reference trajectory; is the frequency of the reference trajectory. Optionally, A r =5mm,
[0081] The tracking error e of the drive control system is defined as:
[0082] e=xx r (11)
[0083] The non-singular terminal sliding surface is defined as:
[0084]
[0085] Solving for the derivative of s is:
[0086]
[0087] Design the drive controller as:
[0088]
[0089] Wherein, κ1>0 and κ2>0 are constants to be designed, preferably, κ1=120, κ2=60; sign(·) is a sign function.
[0090] Therefore, the present invention adopts a neural network composite learning strategy (3) (8) to approximate the unknown nonlinear dynamics under a variable temperature environment, adopts a disturbance observer (9) to estimate the external disturbance, and feeds the unknown dynamics estimate and the disturbance estimate forward to the controller (14) to realize the closed-loop drive control of the hydrogen energy flowmeter. Thus, according to the determined flowmeter drive controller, the vibration tube dynamics model is returned to track the tracking error of the flowmeter drive control system, so that the robust drive control of the flowmeter drive control system under harsh working conditions can be realized.
[0091] As described above, the present invention designs an intelligent learning strategy based on composite learning to cope with the unknown nonlinear dynamics of the system caused by the variable temperature environment, designs a disturbance observer to cope with the external interference caused by mechanical vibration, and suppresses the influence of temperature and vibration by feedforward compensation of unknown dynamic estimation and interference estimation in the control system, thereby realizing robust drive control of the flowmeter under harsh working conditions and meeting the high-precision control requirements of the flowmeter.
[0092] Finally, it should be noted that the features mentioned and / or shown in the above description of the exemplary embodiments of the present invention may be combined in the same or similar manner into one or more other embodiments, combined with the features in other embodiments or substituted for the corresponding features in other implementations. The technical solutions obtained by these combinations or substitutions shall also be deemed to be included in the protection scope of the present invention.
Claims
1. A robust drive control method for a hydrogen energy flow meter based on composite learning, characterized in that: The following steps are involved: Step 1, establishing a dynamic model of a vibrating tube of a hydrogen energy Coriolis mass flowmeter; Step 2, design a neural network composite learning strategy to approximate the unknown nonlinear dynamics of the flow meter drive control system caused by the variable temperature environment, and design a disturbance observer to estimate the external disturbance of the system caused by mechanical vibration; Step 3, based on non-singular terminal sliding mode control, determine the electrostatic driving force generated by the driving coil of the driver in the flowmeter in the dynamic model established in step 1, and feed forward the unknown nonlinear dynamics and external interference obtained in step 2 to the determined electrostatic driving force, so as to achieve robust control of the hydrogen energy flowmeter under harsh working conditions.
2. According to claim 1, the intelligent robust drive control method of hydrogen energy flowmeter based on composite learning is characterized in that: In step 1, the kinetic model is established as: In the formula, and x are the vibration acceleration, vibration velocity and vibration displacement of the flowmeter vibration tube, respectively, m is the mass of the flowmeter vibration system, c and k are the damping coefficient and stiffness coefficient of the vibration tube, respectively, d(t) is the vibration received by the flowmeter vibration tube, and f is the electrostatic driving force generated by the driving coil.
3. The intelligent robust drive control method for hydrogen energy flowmeter based on composite learning according to claim 2 is characterized in that: The step 2 includes the following sub-steps: Step 2.1, define the unknown nonlinear function under variable temperature environment: Step 2.2, use a neural network to approximate the unknown nonlinear function ρ: In the formula, is the estimated value of ρ, is the n-dimensional adjustable neural network weight vector, is the n-dimensional basis vector; Step 2.3, define the model prediction error: In the formula, for of estimate: In the formula, β is a constant, and β>0, is the estimated value of the composite external disturbance D, ε is the neural network approximation error; Step 2.4, design the neural network weight vector update law: Where α1, α2 and α3, as well as p and r are constants, and α1>0, α2>0, α3>0, p>0, 1<r<2, s is the non-singular terminal sliding surface; Step 2.5: Design a disturbance observer to obtain an estimate of the external disturbance Where L is a constant and L>0, is the intermediate variable, and e is the tracking error of the drive control system.
4. The intelligent robust drive control method for a hydrogen energy flow meter based on composite learning according to claim 3 is characterized in that: The step 3 includes the following sub-steps: Step 3.1, define the tracking error e of the drive control system: e=x-x r In the formula, x r is the reference trajectory of the vibration displacement of the vibrating tube, A r is the amplitude of the reference trajectory, is the frequency of the reference trajectory; Step 3.2, define the non-singular terminal sliding surface: Step 3.3, design the drive controller: Where κ1 and κ2 are constants, κ1>0, κ2>0, and sign(·) is the sign function.
