Steam pressure self-adaptive control system and method based on self-learning model
The self-learning model-based adaptive steam pressure control system solves the problems of poor adaptability and complex debugging of nonlinear actuators in steam pressure control systems, achieving high-precision, low-cost, and rapid-deployment steam pressure control.
Patent Information
- Application Number
- CN202511734384.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-24
- Publication Date
- 2026-01-13
AI Technical Summary
Existing steam pressure control systems suffer from poor adaptability of nonlinear actuators, complex debugging, and slow response, making it difficult to meet the requirements for high-precision and low-cost control.
A steam pressure adaptive control system based on a self-learning model is adopted. Through non-proportional regulating valves, pressure sensor modules, main control modules, and human-machine interaction units, combined with least squares fitting and fuzzy integral compensation algorithms, fully automated commissioning and high-precision pressure control are achieved.
It achieved a steam pressure steady-state error of less than ±0.001MPa, reduced the commissioning time to 1.8 minutes, reduced hardware costs by 54.4%, and shortened the large disturbance settling time by 48.9%.
Smart Images
Figure CN121325992A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of industrial automation control technology, and more specifically to a steam pressure adaptive control system and method based on a self-learning model. It is particularly suitable for applications requiring stringent steam pressure stability, such as liquor brewing, fine chemicals, and biopharmaceuticals. Background Technology
[0002] Steam pressure control is a critical aspect of industrial production, and traditional control schemes suffer from the following technical shortcomings: 1. Poor adaptability of nonlinear actuators: Non-proportional control valves (which cost only 40% of proportional valves) often have control errors exceeding ±0.003MPa when using standard PID algorithms due to the nonlinearity of their flow curves, which cannot meet the requirements for high precision. 2. The debugging process relies on human experience: The structural differences of different equipment (such as the bottom pot of the still and the reaction vessel) mean that the PID parameters need to be repeatedly debugged. The debugging cycle of a single device generally takes 2 to 6 hours depending on the debugging experience of the debugging personnel, and it is easy to cause overshoot or oscillation due to insufficient experience of the operators. 3. Dynamic response lag: In order to ensure control accuracy, when the target control pressure is adjusted within a large range (such as the pressure suddenly increasing from 0.035MPa to 0.12MPa during the grain steaming stage), the PID tuning time exceeds about 23% of the theoretical valve action time, resulting in a delay in the process cycle.
[0003] Current technologies primarily optimize PID performance through parameter self-tuning or improve control accuracy and tuning efficiency using fuzzy PID or cascade PID methods. However, they fail to address the control overshoot and response lag caused by the nonlinearity of non-proportional control valves. To further improve control accuracy, proportional control valves are often relied upon, which are costly. Therefore, a steam pressure control technology that balances high accuracy, low cost, and ease of adjustment is urgently needed. Summary of the Invention
[0004] The purpose of this invention is to address the technical problems of low control accuracy, complex debugging, and significant response lag in existing nonlinear actuators. This invention provides a steam pressure adaptive control system and method based on a self-learning model, achieving high-precision pressure control and fully automated debugging of non-proportional regulating valves.
[0005] To achieve the above objectives, the present invention specifically adopts the following technical solution: One aspect of the present invention provides a steam pressure adaptive control system based on a self-learning model for a still system, the still system including a still, a still cover disposed on top of the still, a condenser connected to the still cover by a pipe, a boiler disposed at the bottom of the still for heating the still, a steam pipe connected to the internal pipe of the boiler, and a branch pipe disposed on the steam pipe located outside the still in parallel with the steam pipe; The adaptive control system includes a pressure sensor module, a non-proportional regulating valve, a main control module, and a human-machine interaction unit. The pressure sensor module consists of a pressure transmitter and a mechanical pressure gauge installed on the steam pipeline. The non-proportional regulating valve includes a first threaded shut-off valve, an electric shut-off valve, and a third threaded shut-off valve arranged sequentially along the airflow direction on the branch pipe; the non-proportional regulating valve also includes a second threaded shut-off valve located on the steam pipe connected in parallel with the branch pipe. The main control module is connected to the first threaded stop valve, the second threaded stop valve, the electric stop valve, the third threaded stop valve, the pressure transmitter, and the mechanical pressure gauge, respectively.
[0006] In one embodiment, a steam flow meter is installed on the steam pipeline, and the steam flow meter is signal-connected to the main control module.
[0007] In one embodiment, the first threaded stop valve, the second threaded stop valve, the third threaded stop valve, and the electric stop valve are all DN40 non-proportional electric regulating stop valves with a full stroke time of 35 seconds.
[0008] In one embodiment, the electric shut-off valve is a 4-20mA electric regulating shut-off valve with a full stroke action time of 35 seconds; the main control module is model Huichuan AM320-0808TN with a word processing time of 14.3ns; and the human-machine interaction unit is a 10-inch touch screen from the Huichuan IT7000 series.
[0009] In one embodiment, the pressure sensor is a high-precision polycrystalline silicon pressure transmitter with a measurement range of 0 MPa to 0.6 MPa and an accuracy of ±0.25%FS.
[0010] In one embodiment, the steady-state error of the steam pressure in the steam pipeline is ≤ ±0.001 MPa, and the overshoot frequency is ≤ 1 time / month.
[0011] Another aspect of the present invention provides a steam pressure adaptive control method based on a self-learning model, which employs the above-mentioned steam pressure adaptive control system based on a self-learning model and includes the following steps: S1. Divide the opening (0, 100%) of the non-proportional regulating valve into 10 equal segments in 10% increments, or into multiple segments of varying lengths; fit the pressure equations for each segment using the least squares method. ; S2. The system automatically executes the full stroke of the non-proportional regulating valve; it collects stable pressure values for each segment. And store it in the power-loss retained data area; S3, Real-time calculation of pressure deviation ( );when Time; through formula Dynamically adjust the integral coefficient; output adjustment commands.
[0012] In one implementation, in step S2, the pressure acquisition time for each segment is ≥3s.
[0013] In one implementation, in step S1, the opening degree (0, 100%) of the non-proportional regulating valve is divided into 10 segments in 10% increments; the specific segmentation is as follows: Segment 1: 0%~10%; Segment 2: 10%~20%; ...; Segment i: (i-1)×10% ~ i×10%; ...; Segment 10: 90%~100%; Sampling strategy: Select 3 sampling points (n=3) evenly within each opening interval; and sample data at 0%, 50% and 100% of each step size, respectively. Each segment corresponds to an independent linear sub-model: ; Linear coefficients are estimated using the least squares method. and Construct a system of equations: ; Solving the system of equations simultaneously yields the following results: ; ; In the formula, Q i The steam pressure of the i-th segment (unit: MPa); Let k be the valve opening degree within the i-th segment (value range [(i-1)x10%, ix10%]). i b is the linear slope (sensitivity coefficient) of the i-th segment; i Let be the intercept term for the i-th segment. Example: The fitted model is The maximum deviation between the measured pressure and the model calculation value is ≤0.006MPa.
[0014] In one implementation, step S3 introduces a fuzzy integral compensation algorithm; static errors are eliminated by dynamically adjusting the integral coefficients. ; In the formula, a(t) is the time-varying integral coefficient, which is output using limit logic (value range 0~0.6). a0 is the initial integration coefficient (initial value 0.5, which can be modified according to the actual debugging situation, with a value range of 0.05~0.9); The deviation between the target value and the real-time value; This is the decay reduction factor (controlling the decay rate of the integral as the deviation increases, with an initial value of 0.0015).
[0015] The beneficial effects of this invention are as follows: 1. The steady-state pressure error of the present invention is ±0.001MPa, which is 66.7% higher than that of the traditional PID (proportional valve) steady-state pressure error of ±0.003MPa.
[0016] 2. This invention completes parameter calibration and coefficient tuning through one-click self-learning. The debugging time for a single device is 1.8 minutes. Compared with the traditional PID (proportional valve) single device debugging time of 4 hours (average), the improvement of this invention is 99.2%.
[0017] 3. The large disturbance settling time of the present invention is 24 seconds, which is 48.9% higher than that of the traditional PID (proportional valve) with a large disturbance settling time of 47 seconds.
[0018] 4. The hardware investment (6 devices) of this invention is RMB 26,000, which is RMB 57,000 for the traditional PID (proportional valve) hardware investment (6 devices). The improvement of this invention is 54.4%. Attached Figure Description
[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.
[0020] Figure 1 This is a schematic diagram of a steam pressure adaptive control system based on a self-learning model.
[0021] Reference numerals in the attached diagram: 1. Steam flow meter; 2. First threaded stop valve; 3. Second threaded stop valve; 4. Third threaded stop valve; 5. Electric stop valve; 6. Pressure transmitter; 7. Mechanical pressure gauge. Detailed Implementation
[0022] To make the technical problems, technical solutions, and technical effects of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0023] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0024] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0025] In the description of the embodiments of the present invention, it should be noted that the terms "inner", "outer", "upper", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship in which the product of the invention is usually placed when in use. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the present invention.
[0026] Example 1 This embodiment provides a steam pressure adaptive control system based on a self-learning model for a still system. The still system includes a still, a lid on top of the still, a condenser connected to the lid via a pipe, a boiler at the bottom of the still for heating the still, a steam pipe connected to the inside of the boiler, and a branch pipe on the steam pipe located outside the still. The adaptive control system includes a pressure sensor module, a non-proportional regulating valve, a main control module, and a human-machine interaction unit. The pressure sensor module consists of a pressure transmitter 6 and a mechanical pressure gauge 7 installed on the steam pipeline. The non-proportional regulating valve includes a first threaded shut-off valve 2, an electric shut-off valve 5, and a third threaded shut-off valve 4 arranged sequentially along the airflow direction on the branch pipe; the non-proportional regulating valve also includes a second threaded shut-off valve 3 located on the steam pipe connected in parallel with the branch pipe; The main control module is connected to the first threaded stop valve 2, the second threaded stop valve 3, the electric stop valve 5, the third threaded stop valve 4, the pressure transmitter 6, and the mechanical pressure gauge 7, respectively.
[0027] A steam flow meter 1 is installed on the steam pipeline, and the steam flow meter 1 is connected to the main control module via signal.
[0028] The first threaded stop valve 2, the second threaded stop valve 3, the third threaded stop valve 4, and the electric stop valve 5 are all DN40 non-proportional electric regulating stop valves with a full stroke time of 35 seconds.
[0029] The electric shut-off valve 5 is a 4-20mA electric regulating shut-off valve with a full stroke action time of 35 seconds; the main control module is model Huichuan AM320-0808TN with a word processing time of 14.3ns; the human-machine interaction unit is a 10-inch touch screen from the Huichuan IT7000 series.
[0030] The pressure sensor is a high-precision polycrystalline silicon pressure transmitter 6, with a measurement range of 0MPa-0.6MPa and an accuracy of ±0.25%FS.
[0031] The steady-state error of the steam pressure in the steam pipeline is ≤ ±0.001 MPa, and the overshoot frequency is ≤ 1 time / month. Example 2 This embodiment provides a steam pressure adaptive control method based on a self-learning model, including the following steps: S1. Divide the opening (0, 100%) of the non-proportional regulating valve into 10 equal segments in 10% increments, or into multiple segments of varying lengths; fit the pressure equations for each segment using the least squares method. ; S2. The system automatically executes the full stroke of the non-proportional regulating valve; it collects stable pressure values for each segment. And store it in the power-loss retained data area; S3, Real-time calculation of pressure deviation ( );when Time; through formula Dynamically adjust the integral coefficient; output adjustment commands.
[0032] In step S2, the pressure acquisition time for each segment is ≥3s.
[0033] In step S1, the opening degree (0, 100%) of the non-proportional regulating valve is divided into 10 equal segments in 10% increments; the specific segmentation is as follows: Segment 1: 0%~10%; Segment 2: 10%~20%; ...; Segment i: (i-1)×10% ~ i×10%; ...; Segment 10: 90%~100%; Sampling strategy: Select 3 sampling points (n=3) evenly within each opening interval; and sample data at 0%, 50% and 100% of each step size, respectively. Each segment corresponds to an independent linear sub-model: ; Linear coefficients are estimated using the least squares method. and Construct a system of equations: ; Solving the system of equations simultaneously yields the following results: ; ; In the formula, Q i The steam pressure of the i-th segment (unit: MPa); Let k be the valve opening degree within the i-th segment (value range [(i-1)x10%, ix10%]). i b is the linear slope (sensitivity coefficient) of the i-th segment; i Let be the intercept term for the i-th segment. Example: The fitted model is The maximum deviation between the measured pressure and the model calculation value is ≤0.006MPa.
[0034] In step S3, a fuzzy integral compensation algorithm is introduced; static errors are eliminated by dynamically adjusting the integral coefficients. ; In the formula, a(t) is the time-varying integral coefficient, which is output using limit logic (value range 0~0.6). a0 is the initial integration coefficient (initial value 0.5, which can be modified according to the actual debugging situation, with a value range of 0.05~0.9); The deviation between the target value and the real-time value; This is the decay reduction factor (controlling the decay rate of the integral as the deviation increases, with an initial value of 0.0015).
[0035] This solution has an automatic learning mechanism, as follows: Triggering conditions: The learning button is activated in manual mode after the device is powered on; if the stop button is pressed during the learning process, the valve opening output is 0% and the valve is closed; learning cannot be activated in automatic mode. Full stroke scan: The control valve moves in 10% increments from 0% to 100% and back to 0%; within each increment, the learning is divided into three segments (0% to 50% to 100%); the dwell time for each segment is t=3s (to ensure pressure stability); the stable pressure value for each segment is recorded. ); Data calculation and storage: Stable pressure values for each segment ( Recording completes and triggers the calculation logic; calculates the linear slope of each segment. ) and intercept term ( And record; synchronously set the learning completion status and reset the learning button, and all data and status are recorded in the power-off retention data storage area; This solution features an adaptive dynamic adjustment mechanism: Mechanism Overview: The dynamic adjustment mechanism is the core element for achieving precise pressure control. By combining a piecewise linearization model with a fuzzy integral compensation strategy, it enables real-time optimization of valve opening. Its core logic is: to activate different control modes based on the target pressure; to trigger the compensation mechanism through valve position feedback and pressure deviation grading; and ultimately to eliminate static errors and avoid system oscillations.
[0036] Core control logic: Target pressure judgment and initial opening calculation are as follows: Mode 1: When the target pressure is 0 ( ), directly output valve opening command Cut off the steam supply; ensure the system has no flow output; Mode 2: When the target pressure is greater than 0 ( According to the target pressure Query the piecewise linearized model (see the previous 10-segment division); determine its openness interval. (For example; if) The corresponding steam pressure is in the fifth pressure range; therefore, match the fifth range (40%~50%), and then call the... The linear model of the segment calculates the initial valve opening:
[0037] First point; closed-loop adjustment of valve position, as follows: Feedback deviation judgment: Real-time acquisition of valve position feedback values ; Calculate the deviation:
[0038] Adjust trigger conditions: when (Opening deviation threshold; can be adjusted according to valve mechanical precision; recommended range 0.5%~2%); only the initial opening value is executed; no integral compensation is introduced; when At that time, the pressure deviation compensation stage begins; the fuzzy integral compensation algorithm is activated.
[0039] Second point: Fuzzy integral compensation strategy, as follows: Dynamic integral compensation activation condition: After the valve position closed-loop regulation reaches the target ( ); and based on the actual terminal feedback pressure deviation ( When introducing time-varying integral coefficients. Dynamically correct valve opening; adjust terminal pressure to the set target requirement; and achieve steady-state operating conditions; Dynamic integral compensation exit mechanism: To avoid frequent valve operation under small pressure fluctuations, which could lead to mechanical wear or system oscillation; a compensation exit condition is set; when the real-time pressure deviation meets the condition... At the same time, maintain the current integral compensation value and exit dynamic integral control; achieve overshoot-free steady-state control.
Claims
1. A steam pressure adaptive control system based on a self-learning model, characterized in that, For use in a stilling system, the stilling system includes a still, a still cover disposed on top of the still, a condenser connected to the still cover via a pipe, a boiler disposed at the bottom of the still for heating the still, a steam pipe connected to the internal pipe of the boiler, and a branch pipe disposed in parallel with the steam pipe on the steam pipe located outside the still. The adaptive control system includes a pressure sensor module, a non-proportional regulating valve, a main control module, and a human-machine interaction unit. The pressure sensor module consists of a pressure transmitter and a mechanical pressure gauge installed on the steam pipeline. The non-proportional regulating valve includes a first threaded shut-off valve, an electric shut-off valve, and a third threaded shut-off valve arranged sequentially along the airflow direction on the branch pipe; the non-proportional regulating valve also includes a second threaded shut-off valve located on the steam pipe connected in parallel with the branch pipe. The main control module is connected to the first threaded stop valve, the second threaded stop valve, the electric stop valve, the third threaded stop valve, the pressure transmitter, and the mechanical pressure gauge, respectively.
2. The steam pressure adaptive control system based on a self-learning model according to claim 1, characterized in that, A steam flow meter is installed on the steam pipeline, and the steam flow meter is connected to the main control module via signal.
3. The steam pressure adaptive control system based on a self-learning model according to claim 1, characterized in that, The first threaded stop valve, the second threaded stop valve, the third threaded stop valve, and the electric stop valve are all DN40 non-proportional electric regulating stop valves with a full stroke time of 35 seconds.
4. The steam pressure adaptive control system based on a self-learning model according to claim 1, characterized in that, The electric shut-off valve is a 4-20mA electric regulating shut-off valve with a full stroke action time of 35 seconds; the main control module is model Huichuan AM320-0808TN with a word processing time of 14.3ns; the human-machine interaction unit is a 10-inch touch screen from the Huichuan IT7000 series.
5. The steam pressure adaptive control system based on a self-learning model according to claim 1, characterized in that, The pressure sensor is a high-precision polycrystalline silicon pressure transmitter with a measurement range of 0MPa-1.6MPa and an accuracy of ±0.25%FS.
6. The steam pressure adaptive control system based on a self-learning model according to claim 1, characterized in that, The steady-state error of the steam pressure in the steam pipeline is ≤ ±0.001 MPa, and the overshoot frequency is ≤ 1 time / month.
7. A steam pressure adaptive control method based on a self-learning model, employing a steam pressure adaptive control system based on a self-learning model as described in any one of claims 1 to 6, characterized in that, Includes the following steps: S1. Divide the opening (0, 100%) of the non-proportional regulating valve into 10 equal segments in 10% increments, or into multiple segments of varying lengths; fit the pressure equations for each segment using the least squares method. ; S2. The system automatically executes the full stroke of the non-proportional regulating valve; it collects stable pressure values for each segment. And store it in the power-loss retained data area; S3, Real-time calculation of pressure deviation ( );when Time; through formula Dynamically adjust the integral coefficient; output adjustment commands.
8. The steam pressure adaptive control method based on a self-learning model according to claim 7, characterized in that, In step S2, the pressure acquisition time for each segment is ≥3s.
9. The steam pressure adaptive control method based on a self-learning model according to claim 7, characterized in that, In step S1, the opening degree (0, 100%) of the non-proportional regulating valve is divided into 10 equal segments in 10% increments; the specific segmentation is as follows: Segment 1: 0%~10%; Segment 2: 10%~20%; ...; Segment i: (i-1)×10% ~ i×10%; ...; Segment 10: 90%~100%; Sampling strategy: Select 3 sampling points (n=3) evenly within each opening interval; and sample data at 0%, 50% and 100% of each step size, respectively. Each segment corresponds to an independent linear sub-model: ； Linear coefficients are estimated using the least squares method. and Construct a system of equations: ； Solving the system of equations simultaneously yields the following results: ； ； In the formula, Q i The steam pressure of the i-th segment (unit: MPa); Let k be the valve opening degree within the i-th segment (value range [(i-1)x10%, ix10%]). i b is the linear slope (sensitivity coefficient) of the i-th segment; i Let be the intercept term for the i-th segment. Example: The fitted model is The maximum deviation between the measured pressure and the model calculation value is ≤0.006MPa.
10. The steam pressure adaptive control method based on a self-learning model according to claim 7, characterized in that, In step S3, a fuzzy integral compensation algorithm is introduced; static errors are eliminated by dynamically adjusting the integral coefficients. ； In the formula, a(t) is the time-varying integral coefficient, which is output using limit logic (value range 0~0.6). a0 is the initial integration coefficient (initial value 0.5, which can be modified according to the actual debugging situation, with a value range of 0.05~0.9); The deviation between the target value and the real-time value; This is the decay reduction factor (controlling the decay rate of the integral as the deviation increases, with an initial value of 0.0015).