Thermal expansion self-adaptive gate valve

Abstract

The invention discloses a thermal expansion self-adaptive gate valve, which solves the problems of deformation and leakage caused by thermal expansion of a gate valve sealing surface under a wide temperature range working condition through deep fusion of temperature measurement signal processing and sealing position optimization control. According to the invention, a temperature compensation adaptive filtering algorithm and a variable step gradient descent sealing position optimization algorithm are fused, a bidirectional information flow of temperature signal processing and position control is established, and closed-loop adaptive control is realized. The method has the following beneficial effects that the temperature measurement precision and the sealing position control precision are remarkably improved, the system response speed is increased, the service life of equipment is prolonged, and the maintenance cost is reduced.

Description

Technical Field

[0001] The present invention relates to the field of industrial fluid control equipment, and particularly to a thermally expandable self-adaptive gate valve. Background Art

[0002] In modern industrial production, gate valves, as key fluid control equipment, are widely used for the cutting-off and regulation of various media. In high-temperature industrial environments, the sealing performance of gate valves is directly related to production safety and economic benefits.

[0003] When the working temperature fluctuates within the range of -40°C to 400°C, gate valves face severe technical challenges. First, temperature fluctuations cause uneven deformation of the valve body and the sealing components due to differences in the thermal expansion coefficients of the materials, resulting in a change in the sealing surface gap of 0.5 - 2.0 mm, far exceeding the allowable leakage standard (≤0.1 mm). Taking a fluid catalytic cracking unit as an example, the average leakage rate of high-temperature gate valves is 5 - 8 times higher than that at normal temperature, and the product losses, energy waste, and environmental pollution caused by leakage result in economic losses of tens of millions of yuan every year.

[0004] Secondly, the problem of temperature measurement accuracy severely restricts the control effect. The signal-to-noise ratio (SNR) of traditional temperature monitoring systems is only 10 - 15 dB, and it performs even worse especially in an electromagnetic interference environment; the sensor drift rate reaches 0.5 - 1.5°C / month, and the long-term operation accuracy decays significantly; the temperature measurement delay is 1 - 3 seconds, and it cannot promptly reflect rapid temperature changes.

[0005] Thirdly, traditional control methods are difficult to meet the precise control requirements in a wide temperature range. The mechanical compensation device has a response lag, and the adjustment period is as long as 30 - 60 seconds; traditional PID control has poor effects in the temperature non-linear region, and the overshoot reaches 15 - 25%; single-point temperature feedback cannot cope with the temperature field gradient problem, and local overheating / overcooling phenomena are widespread.

[0006] In addition, the problem of long-term system stability is prominent. The average service life of high-temperature gate valves is only 65% of the designed life, and the cost of early replacement is high; the unplanned shutdowns caused by seal failures result in losses of up to 50 - 2 million yuan per day; the average annual cost of maintaining and replacing seals accounts for 15 - 25% of the total investment in the valves.

[0007] The prior art mainly adopts the following several methods to address the thermal expansion problem of gate valves: Compensating for thermal expansion through mechanical structures such as bellows and springs, but the structure is complex, the reliability is poor, the maintenance cost is high, and the compensation accuracy is limited. Selecting materials with similar thermal expansion coefficients, but it is difficult to maintain the match within a wide temperature range, and certain special working conditions have specific requirements for materials, resulting in limited selection. Monitoring the temperature of the valve body through temperature sensors and adjusting the sealing position, but the prior art has problems such as low temperature measurement accuracy, simple control algorithms, and lack of intelligent self-adaptive capabilities.

[0008] Therefore, there is an urgent need for an adaptive control system that can simultaneously solve the problems of temperature measurement accuracy and sealing position control accuracy to meet the high-reliability sealing requirements under wide-temperature conditions. Summary of the Invention

[0009] The object of the present invention is to provide a thermally expanded adaptive gate valve. This system solves the problems of deformation and leakage caused by thermal expansion of the gate valve sealing surface under wide-temperature conditions (-40°C to 400°C) through the deep integration of temperature measurement signal processing and sealing position optimization control.

[0010] To achieve the above object, the technical solution adopted by the present invention is: A thermally expanded adaptive gate valve includes a temperature monitoring subsystem, an adaptive temperature processing unit, a sealing position optimization control unit, an algorithm fusion interaction unit, and an actuator control system.

[0011] The temperature monitoring subsystem includes a multi-point temperature sensor array and a signal processing unit. The multi-point temperature sensor array is distributed at various positions of the gate valve and is used to collect temperature signals in real time. The signal processing unit is used to condition and preprocess the temperature signals. The multi-point temperature sensor array adopts a specific layout strategy. Symmetric temperature sensors are set on both sides of the sealing surface to monitor the temperature gradient. Temperature sensors are set at the fluid channels inside the gate valve body to monitor the medium temperature. Temperature sensors are set on the outer wall of the gate valve to monitor the influence of the ambient temperature. The signal processing unit uses a high-precision analog-to-digital converter and has functions of signal filtering and outlier processing.

[0012] The adaptive temperature processing unit is connected to the temperature monitoring subsystem and is used to execute the temperature compensation adaptive filtering algorithm. This unit dynamically adjusts the filtering parameters based on the temperature change rate and performs adaptive filtering and nonlinear temperature compensation. The adaptive temperature processing unit has the ability to recognize temperature change patterns, can extract multi-dimensional feature vectors from the filtered temperature signals, automatically classify the temperature changes into steady-state patterns, rising / falling patterns, fluctuating patterns, and mutation patterns, and automatically adjusts the filtering algorithm parameters for different patterns.

[0013] The sealing position optimization control unit is connected to the adaptive temperature processing unit and is used to execute the variable step size gradient descent sealing position optimization algorithm. This unit constructs a sealing performance objective function based on the compensated temperature signal and calculates the optimal sealing position. The sealing position optimization control unit adopts a hierarchical optimization strategy, including a fast response layer, a fine optimization layer, and a steady-state maintenance layer, to improve the system's ability to have both fast response and high-precision control performance.

[0014] The algorithm fusion interaction unit is connected between the adaptive temperature processing unit and the seal position optimization control unit, and is used to realize the collaborative optimization of the temperature compensation adaptive filtering algorithm and the variable step size gradient descent seal position optimization algorithm. This unit enables the direct input of the temperature processing result into the position optimization unit to provide an accurate temperature reference; the seal position control effect is fed back to the temperature processing unit to optimize the filtering parameters; a fusion compensation model update mechanism is constructed; and the temperature change trend information is used to optimize the step size parameters. The algorithm fusion interaction unit also includes an adaptive protection mechanism that automatically takes corresponding protection measures in case of anomalies by evaluating the reliability of the system state in real time.

[0015] The actuator control system is connected to the seal position optimization control unit and is used to drive the seal surface to perform corresponding displacements according to the optimized position command, ensuring that the valve seal surface maintains the best sealing state within the full temperature range. The actuator adopts an electric-hydraulic composite mechanism and has high-precision position control capabilities.

[0016] In addition, the system also includes a self-learning optimization unit. Through long-term data collection, parameter self-optimization, and environmental adaptability learning, this unit enables the system to continuously improve its control performance over time and adapt to specific working conditions and environmental conditions.

[0017] The beneficial effects of the present invention are as follows: Through the deep integration of temperature measurement signal processing and seal position optimization control, the present invention solves the problems of deformation and leakage of the valve seal surface caused by thermal expansion under wide temperature range working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the overall structure of the thermal expansion adaptive valve of the present invention.

[0019] Figure 2 It is a schematic diagram of the architecture of the temperature monitoring subsystem of the present invention.

[0020] Figure 3 It is a functional block diagram of the adaptive temperature processing unit of the present invention.

[0021] Figure 4 It is a processing flow chart of the seal position optimization control unit of the present invention.

[0022] Figure 5 It is a data flow diagram of the algorithm fusion interaction unit of the present invention.

[0023] Figure 6 It is a comparison chart of the control performance of the present invention under different temperature conditions.

[0024] Figure 7 It is a response characteristic diagram of the present invention under the condition of rapid temperature change.

[0025] Figure 8This is the graph of the long-term stability test results of the system of the present invention. Detailed implementation manners

[0026] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0027] As Figure 1 shown, the thermal expansion adaptive gate valve of the present invention includes a temperature monitoring subsystem 10, an adaptive temperature processing unit 20, a sealing position optimization control unit 30, an algorithm fusion interaction unit 40, and an actuator control system 50.

[0028] The temperature monitoring subsystem 10 is located at the input end of the system and is used to collect the temperature data of each part of the gate valve. The adaptive temperature processing unit 20 processes the temperature signal and executes the temperature compensation adaptive filtering algorithm. The sealing position optimization control unit 30 executes the variable step size gradient descent sealing position optimization algorithm according to the processed temperature data. The algorithm fusion interaction unit 40 connects the two algorithm units of temperature processing and position control to realize two-way information flow and collaborative optimization. The actuator control system 50 drives the sealing surface to execute the corresponding displacement according to the position optimization result.

[0029] Each functional unit is interconnected through a digital communication bus to realize data interaction. The system adopts a modular design. The controller adopts an industrial-grade embedded computing platform, which has real-time processing capabilities and anti-interference characteristics.

[0030] As Figure 2 shown, the temperature monitoring subsystem 10 includes a multi-point temperature sensor array 11 and a signal processing unit 12.

[0031] The multi-point temperature sensor array 11 is composed of 8-12 high-precision temperature sensors, and adopts a specific layout strategy: 4 symmetric temperature sensors (T1-T4) are arranged on both sides of the sealing surface to monitor the temperature gradient of the sealing surface; 2 temperature sensors (T5-T6) are arranged at the fluid channel inside the gate valve body to monitor the medium temperature; 2 temperature sensors (T7-T8) are arranged on the outer wall of the gate valve to monitor the influence of the ambient temperature.

[0032] The sensor adopts a PT100 platinum resistance temperature sensor, with a temperature measurement range of -50°C to 450°C, an accuracy of ±0.5°C, and a response time of less than 2 seconds. For special high-temperature working conditions, a K-type thermocouple can be selected, and the upper temperature measurement limit can reach 1000°C.

[0033] The signal processing unit 12 adopts a 24-bit high-precision analog-to-digital converter, with a sampling frequency of 10Hz and a resolution of 0.1°C. This unit has functions of signal amplification, filtering, cold junction compensation, and outlier processing to ensure the accuracy and reliability of the temperature signal.

[0034] Multiple-point temperature signals are fused through weighting to generate a comprehensive temperature index. The calculation formula is as follows: T combined (k)=∑[i = 1 to N]ω i (k)·T i (k) Where, T combined (k) represents the comprehensive temperature index at the k-th moment (unit: °C), T i (k) represents the temperature reading of the i-th sensor at the k-th moment (unit: °C), ω i (k) represents the weight coefficient of the i-th sensor at the k-th moment (dimensionless), satisfying ∑[i = 1 to N]ω i (k)=1, and N represents the total number of sensors.

[0035] The weight coefficient is dynamically adjusted according to the sensor reliability. The calculation formula is as follows: ω i (k)=exp(-λ ω ·|T i (k)-T pred,i (k)| 2 ) / ∑[j = 1 to N]exp(-λ ω ·|T j (k)-T pred,j (k)| 2 )。

[0036] Where, λ ω is the reliability evaluation coefficient (dimensionless), usually taking a value of 0.5 - 2.0; T pred,i (k) is the predicted temperature value of sensor i (unit: °C), estimated based on historical data and heat conduction model; |T i (k)-T pred,i (k)| represents the deviation between the actual temperature and the predicted temperature (unit: °C). The smaller the deviation, the higher the reliability of the sensor and the greater the weight.

[0037] Through multi-point temperature monitoring and data fusion, the system can identify non-uniform temperature distributions and improve the compensation accuracy for non-uniform thermal expansion caused by thermal gradients.

[0038] As Figure 3 shown, the adaptive temperature processing unit 20 processes and optimizes the temperature signal based on the temperature compensation adaptive filtering algorithm. The processing flow of this unit includes the following steps:

[0039] )]]Calculation of temperature change rate: ΔT(k)=T m (k)-T m (k - 1).

[0040] Among them, T m (k) represents the currently collected temperature signal (unit: °C), which is T combined (k) from the temperature monitoring subsystem; T m (k - 1) represents the temperature signal collected at the previous moment (unit: °C); ΔT(k) represents the current temperature change rate (unit: °C). The temperature change rate reflects the speed of temperature change and is an important basis for adjusting the filtering parameters.

[0041] Dynamic filtering parameter adjustment: α(k)=α base +β·exp(-γ·|ΔT(k)| 2 )

[0042] Among them, α(k) represents the filtering coefficient (dimensionless), which controls the smoothness of filtering; α base represents the basic filtering coefficient (dimensionless), with a value range of 0.30 - 0.95, and the basic value is usually set to 0.85; β represents the filtering parameter gain (dimensionless), with a value range of 0.05 - 0.50, and the basic value is usually set to 0.12; γ represents the filtering parameter attenuation coefficient (dimensionless), with a value range of 0.5 - 2.5, and the basic value is usually set to 1.5; |ΔT(k)| represents the absolute value of the temperature change rate (unit: °C).

[0043] The above formula enables the filtering coefficient α(k) to be automatically adjusted according to the temperature change rate: when the temperature change rate is small, exp(-γ·|ΔT(k)| 2 ) is close to 1, and α(k) is close to α base +β, making the filtering smoother; when the temperature change rate is large, exp(-γ·|ΔT(k)| 2 ) is close to 0, and α(k) is close to α base , making the filtering respond more quickly to changes.

[0044] Adaptive filtering execution: T f (k)=α(k)·T f (k - 1)+(1 - α(k))·T m (k)

[0045] Among them, T f (k) represents the filtered temperature signal (unit: °C), which is the filtering result at the current moment; T f (k - 1) represents the filtered temperature signal at the previous moment (unit: °C); α(k) represents the filtering coefficient (dimensionless); T m (k) represents the currently collected temperature signal (unit: °C).

[0046] The above is a first-order low-pass filter, and α(k) controls the smoothness of filtering. The larger α(k) is, the greater the weight of historical data, the smoother the filtering, but the slower the response; the smaller α(k) is, the greater the weight of the current measurement value, the faster the response, but the worse the anti-noise performance. By dynamically adjusting α(k), the system achieves the best balance between smoothness and response speed.

[0047] Nonlinear temperature compensation: T corr (k)=T f (k)+∑[i=1 to n]c i ·f i (T f (k)); Among them, T corr (k) represents the compensated temperature signal (unit: °C), which is the final output temperature value; T f (k) represents the filtered temperature signal (unit: °C); c i represents the compensation coefficient (dimensionless), which is obtained through adaptive learning; f i (T f (k)) represents the characteristic function, which is used to describe the nonlinear temperature characteristics; n represents the number of characteristic functions, usually taking values from 5 to 8.

[0048] The characteristic function f i usually adopts a piecewise polynomial function or a form based on a spline function. In this embodiment, the following piecewise polynomial characteristic function is adopted:

[0049] f i (T)=max(0,min(T-T i-1 ,T i -T i-1 )) / (T i -T i-1 ); Among them, T i represents the temperature segmentation point (unit: °C). The above characteristic function changes linearly within the corresponding temperature range and is 0 or 1 outside the range, which can effectively capture the nonlinear characteristics of different temperature ranges.

[0050] Compensation coefficient update: c(k+1)=c(k)+μ·F(k)·e(k); Among them, c(k+1) represents the compensation coefficient vector at the next moment; c(k) represents the current compensation coefficient vector; μ represents the learning rate (dimensionless), usually taking values from 0.01 to 0.10, which controls the parameter update speed; F(k) represents the feature vector, F(k)=[f1(T f (k)),f2(T f (k)),...,fn (T f (k))]^T; e(k) represents the temperature error (unit: °C), e(k)=T ref (k)-T corr (k), where T ref (k) is the reference temperature.

[0051] This unit also has a temperature change pattern recognition function, and extracts multi-dimensional feature vectors from the filtered temperature signal: F temp =[T mean ,T std ,dT / dt,d 2 T / dt 2 ,PSD peak ,PSD width ,T pattern ; Among them, T mean represents the average temperature (unit: °C), T std represents the standard deviation of temperature (unit: °C), dT / dt represents the temperature change rate (unit: °C / s), d 2 T / dt 2 represents the temperature change acceleration (unit: °C / s 2 ), PSD peak represents the peak value of power spectral density (unit: °C 2 / Hz), PSD width represents the power spectral width (unit: Hz), T pattern represents the temperature mode feature.

[0052] Based on the feature vector, the system automatically classifies the temperature change into the following patterns: Steady state mode: The temperature fluctuation is less than the preset threshold ε steady (usually 0.5 °C / min), |dT / dt|<ε ratesteady .

[0053] Rise / fall mode: The temperature change rate is large and in the same direction, |dT / dt|>ε rateTrans (usually 1.0 °C / min), and the sign remains unchanged.

[0054] Fluctuation mode: The temperature changes frequently but with limited amplitude, PSD width >ε width .

[0055] Mutation mode: The temperature changes sharply, |dT / dt|>ε ratestep (usually 5.0 °C / min) and the duration is short.

[0056] For different modes, the system automatically adjusts the filtering algorithm parameters: Steady state mode: α base = 0.85 - 0.95, β = 0.05 - 0.10, γ = 1.8 - 2.5, with priority given to smoothness.

[0057] Climbing / cooling mode: α base = 0.65 - 0.75, β = 0.15 - 0.25, γ = 1.0 - 1.5, balancing tracking and smoothness.

[0058] Fluctuation mode: α base = 0.55 - 0.65, β = 0.20 - 0.30, γ = 0.8 - 1.2, enhancing the tracking ability.

[0059] Mutation mode: α base = 0.30 - 0.50, β = 0.35 - 0.50, γ = 0.5 - 0.8, with priority given to the response speed.

[0060] Through mode adaptive filtering, the system can quickly respond to real - time temperature changes while maintaining stability, improving the temperature measurement accuracy.

[0061] As Figure 4 shown, the sealing position optimization control unit 30 realizes the precise control of the sealing surface position based on the variable - step - size gradient - descent sealing position optimization algorithm. The processing flow of this unit includes the following steps:

[0062] Construction of the sealing performance objective function: J(p)=w1·(p - p ref (T corr )) 2 +w2·R leak (p,T corr )+w3·F contact (p,T corr ); Among them, J(p) represents the objective function, comprehensively evaluating the sealing performance; p represents the current sealing surface position (unit: mm), and its value range depends on the gate valve size, usually 0 - 10 mm; p ref (T corr ) represents the ideal sealing position (unit: mm) calculated based on the compensated temperature; R leak (p,T corr ) represents the leakage rate estimation function, which is related to the sealing position and temperature; F contact (p,T corr ) represents the contact force estimation function, which is related to the sealing position and temperature; w1, w2, w3 represent the weight coefficients (dimensionless), corresponding to the weights of position, leakage rate, and contact force respectively, and usually take values w1 = 0.5 - 0.7, w2 = 0.2 - 0.4, w3 = 0.1 - 0.2.

[0063] The leakage rate estimation function usually adopts an exponential form: R leak (p, T corr ) = R0·exp(c1·|p - p optimal (T corr )|); Among them, R0 represents the reference leakage rate (unit: Pa·m 3 / s), c1 represents the leakage rate coefficient (unit: 1 / mm), p optimal (T corr ) represents the optimal sealing position (unit: mm).

[0064] The contact force estimation function can be expressed as: F contact (p, T corr ) = K contact ·(p - p open ) 2 ·H(p - p open ); Among them, K contact represents the contact stiffness coefficient (unit: N / mm 2 ), p open represents the opening position (unit: mm), and H(·) represents the Heaviside step function.

[0065] Ideal sealing position model: p ref (T) = p0 + ∑[i = 1 to m]a i ·T^i + ∑[j = 1 to n]b j ·(dT / dt)^j.

[0066] Among them, p ref (T) represents the ideal sealing position (unit: mm); p0 represents the initial position (unit: mm), usually the design reference position; a i represents the temperature polynomial coefficient (unit: mm / ℃^i), describing the static relationship between temperature and position; b j represents the temperature change rate polynomial coefficient (unit: mm / (℃ / s)^j), describing the influence of temperature dynamic change on position; m represents the order of the temperature polynomial, usually taken as 3; n represents the order of the temperature change rate polynomial, usually taken as 2; T represents the compensated temperature (unit: ℃); dT / dt represents the temperature change rate (unit: ℃ / s).

[0067] Calculation of the objective function gradient: ∇J(p(k)) = 2·w1·(p(k) - p ref (T corr )) + w2·∇Rleak (p(k), T corr ) + w3·∇F contact (p(k), T corr ); Among them, ∇J(p(k)) represents the gradient of the objective function J(p) with respect to p at p(k); ∇R leak and ∇F contact respectively represent the gradients of the leakage rate function and the contact force function with respect to p.

[0068] Adaptive step size adjustment: η(k) = η base ·exp(-λ·k) + η min ; Among them, η(k) represents the current step size (dimensionless), which controls the amplitude of each position adjustment; η base represents the base step size (dimensionless), with a value range of 0.08 - 0.20, and the base value is usually set to 0.15; λ represents the step size decay coefficient (dimensionless), with a value range of 0.04 - 0.12, and the base value is usually set to 0.08; k represents the number of iterations; η min represents the minimum step size (dimensionless), with a value range of 0.01 - 0.03, and the base value is usually set to 0.01.

[0069] The above formula makes the step size gradually decrease as the number of iterations increases: the step size is larger in the initial stage, approaching the target position quickly; as the iteration progresses, the step size gradually decreases, finely adjusting the position; finally, the step size is not less than η min , ensuring that the system can be continuously optimized and adapt to temperature changes.

[0070] Position update calculation: p(k + 1) = p(k) - η(k)·∇J(p(k)); Among them, p(k + 1) represents the sealing surface position at the next moment (unit: mm); p(k) represents the current sealing surface position (unit: mm); η(k) represents the current step size (dimensionless); ∇J(p(k)) represents the objective function gradient.

[0071] This formula realizes gradient descent optimization: updating the position along the negative gradient direction of the objective function, gradually reducing the objective function value and gradually improving the sealing performance. The negative sign indicates moving in the opposite direction of the gradient, and η(k) controls the step size of the movement.

[0072] To improve the control efficiency, this unit adopts a hierarchical optimization strategy: 1. Fast response layer (execution cycle: 100 - 200 ms): Based on the known temperature - position mapping relationship, perform fast position adjustment, and adopt a simplified objective function: J fast(p)=(p - p ref (T corr )) 2 , the maximum position adjustment is limited: |Δp f ast| ≤ Δp maxf ast (usually 0.2 - 0.5 mm) 2. Fine optimization layer (execution period: 500 - 1000 ms): Perform a complete gradient descent optimization, evaluate the sealing performance using a comprehensive objective function, and dynamically adjust the weight coefficient: w i (k) = w ibase ·f i (T corr , ΔT corr / Δt).

[0073] 3. Steady - state maintenance layer (execution period: 5 - 10 s): Monitor the position deviation, maintain the best sealing state, compensate for long - term drift and creep effects, and update the temperature - position mapping model parameters.

[0074] This hierarchical optimization strategy enables the system to have both fast response ability and high - precision control performance. It can quickly adjust the position when the temperature changes rapidly, and at the same time achieve fine optimization under steady - state conditions, improving the overall efficiency of the system.

[0075] As Figure 5 shown, the algorithm fusion interaction unit 40 is the core of the present invention, realizing the deep fusion of the temperature - compensation adaptive filtering algorithm and the variable - step - size gradient - descent sealing position optimization algorithm. This unit establishes a two - way information flow between temperature signal processing and position control, enabling the two algorithms to promote each other and jointly optimize.

[0076] Fusion filter parameter optimization: α adaptive (k) = α base + β·exp(-γ·|ΔT(k)| 2 ) + δ·|p(k) - p optimal (T f (k))|; Among them, α adaptive (k) represents the filtered coefficient after fusion optimization (dimensionless); α base , β, γ are the same as the previous definitions; δ represents the position feedback gain coefficient (dimensionless), with a value range of 0.03 - 0.08, and the base value is usually set to 0.05; |p(k) - p optimal (T f (k))| represents the deviation between the current sealing surface position and the ideal position calculated based on the filtered temperature (unit: mm).

[0077] The above formula introduces the position control effect feedback into the adjustment of the filtering parameter: when the position deviation is large, the term δ·|p(k)-p optimal (T f (k))| increases, making α adaptive (k) increase, the filtering is smoother, reducing the position oscillation caused by the temperature measurement noise; when the position is close to the optimum, this term decreases, allowing the filtering to respond faster to the temperature change.

[0078] Fusion compensation model update: c fusion (k + 1)=c(k)+μ T ·F(k)·e T (k)+μ p ·G(k)·e p (k); Among them, c fusion (k + 1) represents the compensated coefficient vector after fusion update; c(k) represents the current compensated coefficient vector; μ T represents the temperature learning rate (dimensionless), and its value range is 0.03 - 0.10; F(k) represents the temperature feature vector; e T (k) represents the temperature estimation error (unit: °C); μ p represents the position learning rate (dimensionless), and its value range is 0.01 - 0.05; G(k) represents the position feature vector; e p (k) represents the position error (unit: mm).

[0079] The above formula jointly uses the temperature error and the position error for the compensated coefficient update, accelerating the system convergence and improving the model accuracy. G(k) can be expressed as the sensitivity matrix of the temperature's influence on the position.

[0080] Fusion step size dynamic adjustment: η fusion (k)=[η base ·exp(-λ·k)+η min ·φ(T corr ,ΔT corr / Δt); Among them, η fusion (k) represents the step size after fusion optimization (dimensionless); η base , λ, η min are the same as the previous definitions; φ(T corr ,ΔT corr / Δt) represents the temperature state adjustment function, based on the compensated temperature and its change rate.

[0081] The temperature state adjustment function can be expressed as: φ(T corr ,ΔT corr / Δt)=1+κ·|ΔT corr / Δt| / (1+ε·|ΔT corr / Δt|); Among them, κ represents the temperature change rate gain (dimensionless), and its value range is 0.5 - 2.0; ε represents the suppression coefficient (unit: s / ℃), and its value range is 0.05 - 0.20; |ΔT corr / Δt| represents the absolute value of the temperature change rate (unit: ℃ / s).

[0082] The above function enables the step size to be dynamically adjusted according to the temperature change characteristics: when the temperature change rate is large, the φ value increases, the step size increases, and the system response is accelerated; when the temperature is stable, the φ value approaches 1, the step size decreases, and the system stability is improved. At the same time, the denominator term prevents the system instability caused by excessive increase of the step size when the temperature change rate is too large.

[0083] The algorithm fusion interaction unit also includes an adaptive protection mechanism, which automatically takes corresponding protection measures in case of anomalies by real-time evaluating the system state reliability: Fusion reliability evaluation: R temp =exp(-||T raw -T mode l|| 2 / σ T 2 ); R pos =exp(-||p actual -p optimal || 2 / σ p 2 ); R fusion =w T ·R temp +w p ·R pos ; Among them, R temp represents the temperature signal reliability (dimensionless), and its value range is 0 - 1; T raw represents the original temperature signal (unit: ℃); T mode l represents the model predicted temperature (unit: ℃); σ T represents the temperature error standard deviation (unit: ℃); R pos represents the position control reliability (dimensionless), and its value range is 0 - 1; p actual represents the actual seal position (unit: mm); p optimal represents the optimal seal position (unit: mm); σ p represents the position error standard deviation (unit: mm); R fusionrepresents the fusion reliability index (dimensionless), with a value range of 0 - 1; w T and w p represent the weight coefficients (dimensionless), satisfying w T + w p = 1.

[0084] Protection strategy based on the reliability index: Normal mode (R fusion > 0.8): The system operates normally with full - parameter adaptability.

[0085] Cautious mode (0.5 < R fusion ≤ 0.8): Increase the filtering smoothness, reduce the position adjustment amplitude, and limit the parameter change rate. <http: / / www.example.com / <http: / / www.example.com /

[0086] Protection mode (0.3 < R fusion ≤ 0.5): Adopt conservative parameter settings, prioritize ensuring system stability, and activate redundant sensors. <http: / / www.example.com / <http: / / www.example.com /

[0087] Safety mode (R fusion ≤ 0.3): Lock the current position, prohibit large - scale adjustments, issue a system warning, and wait for manual intervention. <http: / / www.example.com / <http: / / www.example.com /

[0088] This adaptive protection mechanism automatically takes corresponding protection measures in case of anomalies by real - time evaluating the system - state reliability, preventing equipment damage or performance deterioration caused by incorrect control, and improving the overall robustness and reliability of the system. <http: / / www.example.com / <http: / / www.example.com /

[0089] The actuator control system 50 is responsible for optimizing the position command p(k + 1) output by the sealing - position optimization control unit, driving the sealing surface to perform corresponding displacements, and ensuring that the gate - valve sealing surface maintains the best sealing state within the full temperature range. <http: / / www.example.com / <http: / / www.example.com /

[0090] The actuator adopts an electro - hydraulic composite mechanism, combining the precise control ability of an electric actuator and the large - thrust characteristics of a hydraulic actuator. The position - control resolution reaches 0.01 mm, the maximum stroke is designed according to the gate - valve size, usually 5 - 20 mm. The response time is less than 0.5 seconds, and the adjustment accuracy reaches ±2%. <http: / / www.example.com / <http: / / www.example.com /

[0091] The actuator control system includes the following core components: Position controller: Receives the position command and calculates the control output. Drive circuit: Provides precise drive signals for the electric actuator. Hydraulic drive unit: Provides stable hydraulic power. Position feedback sensor: Provides real - time position feedback to form a closed - loop control. Safety protection circuit: Provides fast protection in case of system anomalies. <http: / / www.example.com / <http: / / www.example.com /

[0092] The control algorithm uses an improved PID controller with feedforward compensation and anti-integral saturation design to ensure fast response and precise positioning in position control. The control system also takes into account the non-linear characteristics of the actuator, such as friction, valve weight, and medium pressure, etc., and reduces the influence of these factors through a compensation algorithm.

[0093] The present invention also includes a self-learning optimization unit, which enables the system to continuously improve control performance as the running time increases through continuous data collection and parameter optimization. This unit includes the following modules:

[0094] 1. Long-term data collection module: Records historical data of key parameters such as temperature, position, and sealing performance, and establishes a working condition - performance correlation database.

[0095] 2. Parameter self-optimization module: Optimizes the parameters of the temperature filtering algorithm based on historical data: α baseopt =arg min E[|T actual -T corr | 2 ; Optimizes the parameters of the position control algorithm: η baseopt ,λopt=arg min E[|p actual -p optimal | 2 ; Updates the temperature - position mapping model: [a,b]opt=arg min E[|p actual (T)-p mode l(T)| 2 .

[0096] 3. Environmental adaptability learning: Identifies the influence of different environmental conditions (such as external temperature, humidity, vibration, etc.) on system performance; establishes an environmental factor compensation model; automatically adjusts control parameters for specific environmental conditions.

[0097] Through continuous data collection and parameter optimization, the system can continuously improve control performance as the running time increases, adapt to specific working conditions and environmental conditions, reduce maintenance requirements, and extend the service life of the equipment.

[0098] To better understand the technical principle and application effect of the present invention, the following detailed calculation examples are provided in this application, showing the performance of the thermal expansion self-adaptive gate valve under three typical working conditions: 1. Initial parameter setting This calculation example covers three typical working conditions, corresponding to different temperature ranges and working characteristics respectively: Condition A: High-temperature steam gate valve of petrochemical plant (steady-state high-temperature condition) Gate valve specifications: DN200, pressure rating: Class900; sealing surface material: nickel-based alloy Inconel718; coefficient of linear expansion α = 13.8×10^-6 / ℃; modulus of elasticity E = 205GPa = 2.05×10^11Pa; Sealing surface dimensions: sealing surface diameter D = 210mm; initial length of sealing surface L = 0.12m; sealing contact area A = 0.0025m 2 .

[0099] Operating environment: initial temperature T initial = 25℃ (ambient temperature); operating temperature T work = 365℃; medium pressure P = 16.5MPa Temperature change characteristics: steady state, temperature fluctuation ±5℃.

[0100] Condition B: Main steam gate valve of nuclear power plant (rapid temperature change condition) Gate valve specifications: DN350, pressure rating: Class1500; sealing surface material: cobalt-based alloy Stellite6; coefficient of linear expansion α = 15.6×10^-6 / ℃; modulus of elasticity E = 215GPa = 2.15×10^11Pa; Sealing surface dimensions: sealing surface diameter D = 380mm; initial length of sealing surface L = 0.15m; sealing contact area A = 0.004m 2 ; Operating environment: initial temperature T initial = 25℃ (ambient temperature); operating temperature T work Change range: 180℃ to 320℃; medium pressure P = 17.5MPa; Temperature change characteristics: heating / cooling rate up to 3.5℃ / min.

[0101] Condition C: LNG cryogenic gate valve (extremely low-temperature condition) Gate valve specifications: DN150, pressure rating: Class600; sealing surface material: austenitic stainless steel 316L; coefficient of linear expansion α = 16.0×10^-6 / ℃; modulus of elasticity E = 195GPa = 1.95×10^11Pa Sealing surface dimensions: sealing surface diameter D = 165mm; initial length of sealing surface L = 0.075m; sealing contact area A = 0.0018m 2 Operating environment: initial temperature T initial = 20℃ (ambient temperature); operating temperature T work = -162℃ (liquefied natural gas temperature); medium pressure P = 7.5MPa; Temperature change characteristics: rapid cooling, cold start temperature change rate up to 5℃ / min.

[0102] Algorithm control parameter setting Parameters of the temperature compensation adaptive filtering algorithm: Operating condition A (steady-state high temperature): α base = 0.85, β = 0.08, γ = 2.0; Operating condition B (rapid change): α base = 0.68, β = 0.20, γ = 1.2; Operating condition C (extremely low temperature): α base = 0.88, β = 0.06, γ = 2.2.

[0103] Parameters of the variable step-size gradient descent seal position optimization algorithm: Operating condition A: η base = 0.12, λ = 0.08, η min = 0.01; Operating condition B: η base = 0.18, λ = 0.06, η min = 0.015; Operating condition C: η base = 0.10, λ = 0.09, η min = 0.008.

[0104] Parameters of the fusion interaction algorithm: Operating condition A: δ = 0.05, κ = 1.0, ε = 0.12; Operating condition B: δ = 0.04, κ = 1.5, ε = 0.08; Operating condition C: δ = 0.06, κ = 0.8, ε = 0.15.

[0105] 2. Operating condition A: Calculation process of the high-temperature steam gate valve in the petrochemical plant Step 1: Multi-point temperature acquisition and fusion processing Assume that at time k, the actual readings of 8 temperature sensors are as follows (unit: °C): T1 = 362.7 (upper part of the seal surface); T2 = 366.3 (lower part of the seal surface); T3 = 364.8 (left side of the seal surface); T4 = 365.4 (right side of the seal surface); T5 = 367.2 (inlet of the fluid channel); T6 = 368.1 (outlet of the fluid channel); T7 = 358.5 (upper part of the valve outer wall); T8 = 357.2 (lower part of the valve outer wall).

[0106] First, calculate the predicted temperature of each sensor (assuming a simple linear prediction based on historical data): T pred,1 = 363.2 °C; T pred,2 = 365.8 °C; T pred,3 = 364.5 °C; T pred,4 = 365.0 °C; T pred,5= 367.5 °C; T pred,6 = 367.8 °C; T pred,7 = 358.0 °C; T pred,8 = 357.5 °C.

[0107] Calculate the squared deviation of each sensor: |T1 - T pred,1 | 2 = |362.7 - 363.2| 2 = 0.25; |T2 - T pred,2 | 2 = |366.3 - 365.8| 2 = 0.25; |T3 - T pred,3 | 2 = |364.8 - 364.5| 2 = 0.09; |T4 - T pred,4 | 2 = |365.4 - 365.0| 2 = 0.16; |T5 - T pred,5 | 2 = |367.2 - 367.5| 2 = 0.09; |T6 - T pred,6 | 2 = |368.1 - 367.8| 2 = 0.09; |T7 - T pred,7 | 2 = |358.5 - 358.0| 2 = 0.25; |T8 - T pred,8 | 2 = |357.2 - 357.5| 2 = 0.09.

[0108] Substitute the reliability evaluation coefficient λ ω = 1.5, and calculate the weight denominator term for each sensor: ∑[j = 1 to 8] exp(-λ ω ·|T j - T pred,j | 2 ) = exp(-1.5 × 0.25) + exp(-1.5 × 0.25) + exp(-1.5 × 0.09) + exp(-1.5 × 0.16) + exp(-1.5 × 0.09) + exp(-1.5 × 0.09) + exp(-1.5 × 0.25) + exp(-1.5 × 0.09) =0.6873 + 0.6873 + 0.8738 + 0.7866 + 0.8738 + 0.8738 + 0.6873 + 0.8738 = 6.3437。

[0109] Calculate the weights of each sensor: ω1 = exp(-1.5 × 0.25) / 6.3437 = 0.6873 / 6.3437 = 0.1083; ω2 = exp(-1.5 × 0.25) / 6.3437 = 0.6873 / 6.3437 = 0.1083; ω3 = exp(-1.5 × 0.09) / 6.3437 = 0.8738 / 6.3437 = 0.1377; ω4 = exp(-1.5 × 0.16) / 6.3437 = 0.7866 / 6.3437 = 0.1240; ω5 = exp(-1.5 × 0.09) / 6.3437 = 0.8738 / 6.3437 = 0.1377; ω6 = exp(-1.5 × 0.09) / 6.3437 = 0.8738 / 6.3437 = 0.1377; ω7 = exp(-1.5 × 0.25) / 6.3437 = 0.6873 / 6.3437 = 0.1083; ω8 = exp(-1.5 × 0.09) / 6.3437 = 0.8738 / 6.3437 = 0.1377。

[0110] Calculate the comprehensive temperature through weight fusion: T combined (k) = 0.1083 × 362.7 + 0.1083 × 366.3 + 0.1377 × 364.8 + 0.1240 × 365.4 + 0.1377 × 367.2 + 0.1377 × 368.1 + 0.1083 × 358.5 + 0.1377 × 357.2 = 363.82 °C。

[0111] Step 2: Calculate the temperature change rate Assume that the fused temperature T at the previous moment combined (k - 1) = 363.25 °C, then the temperature change rate: ΔT(k) = T m (k) - T m (k - 1) = 363.82 - 363.25 = 0.57 °C Step 3: Adjust the dynamic filtering parameters Use the filtering parameters of working condition A (α base = 0.85, β = 0.08, γ = 2.0): α(k)=α base +β·exp(-γ·|ΔT(k)| 2 )=0.85+0.08·exp(-2.0·|0.57| 2 ) =0.85+0.08·exp(-2.0×0.3249)=0.85+0.08×0.5227=0.85+0.0418=0.8918; Step 4: Adaptive filtering execution Assume that the filtering result at the previous moment is T f (k - 1)=363.40 °C, then: T f (k)=α(k)·T f (k - 1)+(1 - α(k))·T m (k)=0.8918×363.40+(1 - 0.8918)×363.82=324.0505+39.3547=363.41 °C; Step 5: Nonlinear temperature compensation For working condition A, we use 4 characteristic functions and compensation coefficients. Assume that the current system compensation coefficients and characteristic function values are:

[0112] c1 = 0.22, f1(T f (k)) = 0.35 (characteristic function in the 350 - 370 °C interval) c2=-0.12, f2(T f (k)) = 0.28 (characteristic function in the 360 - 380 °C interval) c3 = 0.08, f3(T f (k)) = 0.15 (sensor nonlinear characteristic function) c4=-0.05, f4(T f (k)) = 0.42 (material property compensation characteristic function) Then the compensated temperature: T corr (k)=T f (k)+∑[i = 1 to n]c i ·f i (T f (k))T corr (k) =363.41+(0.22×0.35+(-0.12)×0.28+0.08×0.15+(-0.05)×0.42) = 363.41 + (0.077 - 0.0336 + 0.012 - 0.021) = 363.41 + 0.0344 = 363.44 °C.

[0113] Step 6: Calculation of the ideal seal position model Using the temperature-dependent seal position model (parameters for operating condition A): p ref (T) = p0 + ∑[i = 1 to m] a i · T^i + ∑[j = 1 to n] b j · (dT / dt)^j; Given the following parameters: p0 = 7.5 mm (initial position); a1 = 0.0018 mm / °C (first-order temperature coefficient); a2 = 2.8×10^-6 mm / °C 2 (second-order temperature coefficient); a3 = -1.7×10^-9 mm / °C 3 (third-order temperature coefficient); b1 = 0.038 mm / (°C / min) (first-order rate coefficient); b2 = 0.0025 mm / (°C / min) 2 (second-order rate coefficient); T = 363.44 °C (compensated temperature); dT / dt = 0.57 °C / min (temperature change rate).

[0114] Calculate the ideal seal position: p ref (T) = 7.5 + 0.0018 × 363.44 + 2.8×10^-6 × 363.44 2 - 1.7×10^-9 × 363.44 3 + 0.038 × 0.57 + 0.0025 × 0.57 2 = 7.5 + 0.6542 + 0.3699 - 0.0817 + 0.0217 + 0.0008 = 8.4649 mm.

[0115] Step 7: Construction of the seal performance objective function Assume the current seal face position p(k) = 8.38 mm, then: Position error term: (p(k) - p ref (T corr )) 2 = (8.38 - 8.4649) 2 = (-0.0849) 2 = 0.0072; Leakage rate estimation function (assuming R0 = 8.0×10^-5 Pa·m 3 / s, c1 = 18.5 mm^-1): Rleak (p, T corr ) = R0·exp(c1·|p - p optimal (T corr )|) = 8.0×10^-5·exp(18.5×|8.38 - 8.4649|) = 8.0×10^-5·exp(18.5×0.0849) = 8.0×10^-5·exp(1.5707) = 8.0×10^-5×4.8100 = 3.848×10^-4 Pa·m 3 / s.

[0116] Contact force estimation function (assuming K contact = 4800 N / mm 2 , p open = 6.2 mm): F contact (p, T corr ) = K contact ·(p - p open ) 2 ·H(p - p open ) = 4800×(8.38 - 6.2) 2 ×1 = 4800×4.7524 = 22811.5 N.

[0117] Objective function (weight coefficients for operating condition A: w1 = 0.65, w2 = 0.25, w3 = 0.1, where the w3 term is normalized): J(p) = w1·(p - p ref (T corr )) 2 + w2·R leak (p, T corr ) + w3·F contact (p, T corr ) / F max J(p) = 0.65×0.0072 + 0.25×3.848×10^-4 + 0.1×22811.5 / 25000 = 0.00468 + 0.0000962 + 0.09125 J(p) = 0.09603.

[0118] Step 8: Calculation of the objective function gradient Use the numerical differentiation method to calculate the gradient: ∇J(p(k)) ≈ [J(p(k) + Δp) - J(p(k) - Δp)] / (2×Δp).

[0119] Take Δp = 0.01 mm, and calculate J(8.39) and J(8.37): For p = 8.39 mm: Position error term: (8.39 - 8.4649) 2 = (-0.0749) 2 = 0.00561 Leakage rate: R leak = 8.0×10^-5·exp(18.5×0.0749) = 8.0×10^-5×3.9698 = 3.176×10^-4; Contact force: F contact = 4800×(8.39 - 6.2) 2 = 4800×4.7961 = 23021.3N

[0120] Objective function value: J(8.39) = 0.65×0.00561 + 0.25×3.176×10^-4 + 0.1×23021.3 / 25000 = 0.00365 + 0.0000794 + 0.09209 = 0.09582

[0121] For p = 8.37mm: Position error term: (8.37 - 8.4649) 2 = (-0.0949) 2 = 0.00901

[0122] Leakage rate: R leak = 8.0×10^-5·exp(18.5×0.0949) = 8.0×10^-5×5.8342 = 4.667×10^-4

[0123] Contact force: F contact = 4800×(8.37 - 6.2) 2 = 4800×4.7089 = 22603.0N

[0124] Objective function value: J(8.37) = 0.65×0.00901 + 0.25×4.667×10^-4 + 0.1×22603.0 / 25000 = 0.00586 + 0.0001167 + 0.09041 = 0.09639

[0125] Gradient calculation: ∇J(p(k)) = [J(8.39) - J(8.37)] / (2 × 0.01) = [0.09582 - 0.09639] / 0.02 = -0.0057 / 0.02 = -0.285. The negative gradient indicates that the seal position value should be increased to reduce the objective function.

[0126] Step 9: Adaptive step size adjustment Assume the current iteration number k = 7, and use the step size parameters for operating condition A (η base = 0.12, λ = 0.08, η min = 0.01): η(k) = η base · exp(-λ · k) + η min η(7) = 0.12 · exp(-0.08 × 7) + 0.01 η(7) = 0.12 · exp(-0.56) + 0.01 η(7) = 0.12 × 0.5712 + 0.01 η(7) = 0.06854 + 0.01 η(7) = 0.07854.

[0127] Step 10: Fusion step size dynamic adjustment Use the fusion step size dynamic adjustment formula (parameters for operating condition A: κ = 1.0, ε = 0.12): η fusion (k) = [η base · exp(-λ · k) + η min · φ(T corr , ΔT corr / Δt); where φ(T corr , ΔT corr / Δt) = 1 + κ · |ΔT corr / Δt| / (1 + ε · |ΔT corr / Δt|); Substitute the parameters κ = 1.0, ε = 0.12, |ΔT corr / Δt| = 0.57 °C / min: φ(T corr , ΔT corr / Δt) = 1 + 1.0 × 0.57 / (1 + 0.12 × 0.57) = 1 + 0.57 / 1.0684 φ = 1 + 0.5336 = 1.5336.

[0128] Fusion step size: η fusion (7) = 0.07854 × 1.5336 = 0.12045.

[0129] Step 11: Position update calculation Use the position update formula: p(k + 1)=p(k)-η fusion (k)·∇J(p(k)); p(8)=8.38 - 0.12045×(-0.285)=8.38 + 0.03433p(8)=8.41433mm。

[0130] Step 12: Fusion Filter Parameter Optimization Use the fusion filter parameter optimization formula (parameters for operating condition A: δ = 0.05): α adaptive (k)=α base +β·exp(-γ·|ΔT(k)| 2 )+δ·|p(k)-p optimal (T f (k))| Assume p optimal (T f (k)) = 8.4649mm, δ = 0.05: α adaptive (k)=0.85 + 0.08·exp(-2.0·|0.57| 2 )+0.05·|8.38 - 8.4649| =0.85 + 0.08×0.5227 + 0.05×0.0849 = 0.85 + 0.0418 + 0.00425 = 0.89605。

[0131] The optimized filtering coefficient increases by approximately 0.45%, further improving the filtering smoothness and reducing the position adjustment oscillation.

[0132] Step 13: System Performance Evaluation Thermal adaptation performance index: Assume the ideal compensation amount ΔL ideal =0.955mm, the actual compensation amount ΔL comp =0.914mm: η Thermal =ΔL comp / ΔL ideal =0.914 / 0.955 = 0.957。

[0133] Sealing reliability index: The maximum allowable leakage rate L max =8×10^-3Pa·m 3 / s, the calculated leakage rate L = 3.848×10^-4Pa·m 3 / s; R seal =1-(L / L max) = 1 - (3.848×10^-4 / 8×10^-3) = 1 - 0.0481 = 0.952 Thermal stress calculation: σ = E×α×ΔT = 2.05×10^11×13.8×10^-6×(363.44 - 25) = 2.05×10^11×13.8×10^-6×338.44 = 9.566×10^8 Pa = 956.6 MPa.

[0134] Considering the constraint coefficient in the structural design is 0.18, the actual thermal stress is: σ actual = 956.6×0.18 = 172.2 MPa.

[0135] Expected service life: The reference life L0 = 40000 hours, the allowable stress of the material σ max = 480 MPa, the actual thermal stress σ actual = 172.2 MPa, the material life index m = 2.8:<{ Lifetime = L0×(σ max / σ actual )^m = 40000×(480 / 172.2)^2.8 = 40000×(2.79)^2.8 = 40000×17.74 = 709,600 hours (about 81 years).

[0136] 3. Condition B: Calculation process of the main steam isolation valve of the nuclear power plant Temperature change rate and filter parameter adjustment Under the condition of rapid temperature change, assume that the system measured temperature at a certain moment is T m (k) = 253.6 °C, the previous moment is T m (k - 1) = 250.1 °C, then: ΔT(k) = T m (k) - T m (k - 1) = 253.6 - 250.1 = 3.5 °C.

[0137] Using the filtering parameters of Condition B (α base = 0.68, β = 0.20, γ = 1.2): α(k) = α base + β·exp(-γ·|ΔT(k)| 2 ) α(k) = 0.68 + 0.20·exp(-1.2·|3.5| 2 ) = 0.68 + 0.20·exp(-1.2×12.25) = 0.68 + 0.20×0.0000196 = 0.68 + 0.00000392 = 0.68000392.

[0138] At this time, the filtering coefficient is close to the base value αbase , indicating that the system has recognized a rapid temperature change, significantly reducing the weight of historical data and enhancing the response speed to new measurement values.

[0139] Calculation of ideal sealing position Using the temperature position mapping parameters of operating condition B, calculate the ideal seal position (substitute T = 253.6 °C, dT / dt = 3.5 °C / min): p ref (T)=9.2 + 0.0025×253.6 + 1.8×10^-6×253.6 2 -1.2×10^-9×253.6 3 +0.065×3.5 + 0.0045×3.5 2 =9.2 + 0.634 + 0.116 - 0.019 + 0.228 + 0.055 = 10.214 mm.

[0140] Dynamic adjustment of fusion step size Using the parameters of operating condition B (κ = 1.5, ε = 0.08, η base =0.18, λ = 0.06, η min =0.015): φ(T corr ,ΔT corr / Δt)=1 + 1.5×3.5 / (1 + 0.08×3.5)=1 + 5.25 / 1.28 = 1 + 4.102 = 5.102.

[0141] Assume the number of iterations k = 8: η(8)=0.18·exp(-0.06×8)+0.015 = 0.18×0.618 + 0.015 = 0.126.

[0142] Fusion step size: η fusion (8)=0.126×5.102 = 0.643.

[0143] The significantly increased step size value above (about 5 times that of operating condition A) enables the system to quickly respond to the seal position requirements under rapid temperature change conditions and achieve timely tracking of rapid temperature changes.

[0144] Performance evaluation results The calculation results of the performance evaluation for operating condition B show that: Thermal adaptation performance index: η Thermal =0.923 Seal reliability index: R seal =0.947 Expected service life: Lifetime = 292,500 hours (about 33 years) 4. Operating condition C: Calculation process of LNG cryogenic gate valve Cryogenic characteristic compensation For extremely low temperature operating conditions, temperature characteristic compensation is particularly important. Assume that at -162 °C, the system uses 8 characteristic functions to capture the non-linear characteristics in the low temperature region, and focuses on the non-linear behavior in the temperature range from -165 °C to -155 °C and near the freezing point.

[0145] Thermal stress calculation Under cryogenic operating conditions, calculation of thermal stress (considering temperature change ΔT = -182 °C): σ = E × α × ΔT = 1.95×10^11 × 16.0×10^-6 × (-182) = 1.95×10^11 × 16.0×10^-6 × (-182) = -5.683×10^8 Pa = -568.3 MPa (negative value indicates compressive stress) Considering the structural constraint coefficient of 0.22: σ actual = -568.3 × 0.22 = -125.0 MPa.

[0146] Low-temperature sealing characteristics Under cryogenic conditions, the sealing material hardens and the contact force characteristics change. The model adapts by modifying the coefficient of the contact force estimation function:

[0147] K contact (-162 °C) = K contact (20 °C) × 1.35 = 4200 × 1.35 = 5670 N / mm 2 .

[0148] The above adjustments ensure accurate estimation of the sealing contact force in a cryogenic environment, avoiding over-compression or insufficient sealing.

[0149] Performance evaluation results The calculation results of the performance evaluation for operating condition C show that: Thermal adaptation performance index: η Thermal = 0.935; Sealing reliability index: R seal = 0.968; Expected service life: Lifetime = 368,000 hours (about 42 years).

[0150] 5. Comparative analysis of the algorithm performance under three operating conditions Through the calculation examples of three typical operating conditions, it can be clearly shown the performance of the thermal expansion self-adaptive gate valve of the present invention under different temperature conditions:

[0151] 6. Technical effect analysis Through the above calculation examples, the following technical effect analysis results can be obtained: (1) The temperature measurement accuracy is significantly improved Under the steady-state high-temperature condition (condition A), the temperature measurement error is only ±0.28°C, which is 84% higher than that of the traditional system (±1.8°C); Under the condition of rapid temperature change (condition B), the temperature measurement error is ±0.72°C, which is 84% higher than that of the traditional system (±4.5°C); Under the extremely low-temperature condition (condition C), the temperature measurement error is ±0.35°C, which is 84% higher than that of the traditional system (±2.2°C).

[0152] (2) The control accuracy of the sealing position is greatly improved Under the steady-state high-temperature condition, the position control accuracy is improved to ±0.046 mm, which is 85% higher than that of the traditional system (±0.3 mm); Under the condition of rapid temperature change, the position control accuracy reaches ±0.087 mm, which is 89% higher than that of the traditional system (±0.8 mm); Under the extremely low-temperature condition, the position control accuracy reaches ±0.051 mm, which is 87% higher than that of the traditional system (±0.4 mm).

[0153] (3) The system response speed is significantly accelerated The average response time is shortened from the traditional 30 - 55 seconds to 6.8 - 9.5 seconds, an increase of 76 - 88%. Especially under the condition of rapid temperature change, the response time is only 6.8 seconds, meeting the requirements of harsh conditions.

[0154] (4) The long-term reliability is greatly improved The expected service lives under the three conditions reach 709,600 hours, 292,500 hours, and 368,000 hours respectively; the average service life is extended by 2.4 times, significantly reducing the equipment replacement and maintenance frequency; all the sealing reliability indicators are higher than 0.94, greatly reducing the leakage risk.

[0155] (5) A breakthrough is achieved in the wide temperature range adaptability The system can work stably in the ultra-wide temperature range from -162°C to 365°C. The temperature adaptation range is expanded by more than 60% compared with the traditional system, and the adaptability to rapid temperature changes is significantly enhanced, and it can handle a temperature change rate of up to 5°C / min.

[0156] 7. Conclusion The thermal expansion adaptive gate valve of the present invention realizes high-precision sealing control under wide temperature range working conditions by integrating a temperature compensation adaptive filtering algorithm and a variable step size gradient descent sealing position optimization algorithm. Computational examples prove that the system exhibits excellent performance under three typical working conditions: steady-state high temperature, rapid temperature change, and extremely low temperature: high-precision temperature measurement ability with an error controlled within the range of ±0.28 - 0.72 °C; excellent sealing position control accuracy with an error controlled within the range of ±0.046 - 0.087 mm; fast system response ability with the response time shortened to 6.8 - 9.5 seconds; long service life characteristics with the service life extended to more than 2.4 times that of the traditional system. The significant improvement of the above indicators enables the present invention to effectively solve the problem of sealing failure of gate valves in high-temperature industrial applications, reduce the leakage risk, extend the service life of equipment, reduce maintenance costs, and provide a highly reliable fluid control technical solution for the high-temperature industrial field.

Claims

1. A thermally expandable self - adapting gate valve, characterized in that Comprising: A temperature monitoring subsystem, which includes a multi-point temperature sensor array and a signal processing unit. The multi-point temperature sensor array is distributed at various positions of the gate valve and is used to collect temperature signals in real time. The signal processing unit is used to condition and preprocess the temperature signals; An adaptive temperature processing unit, which is connected to the temperature monitoring subsystem and is used to execute a temperature compensation adaptive filtering algorithm, dynamically adjust the filtering parameters based on the temperature change rate, and perform adaptive filtering and non-linear temperature compensation; A sealing position optimization control unit, which is connected to the adaptive temperature processing unit and is used to execute a variable step size gradient descent sealing position optimization algorithm, construct a sealing performance objective function based on the compensated temperature signal, and calculate the optimal sealing position; An algorithm fusion and interaction unit, which is connected between the adaptive temperature processing unit and the sealing position optimization control unit and is used to realize the collaborative optimization of the temperature compensation adaptive filtering algorithm and the variable step size gradient descent sealing position optimization algorithm; An actuator control system, which is connected to the sealing position optimization control unit and is used to drive the sealing surface to perform corresponding displacements according to the optimized position command, ensuring that the gate valve sealing surface maintains the best sealing state within the full temperature range.

2. The thermally-expansion self-adaptive gate valve according to claim 1, wherein: The temperature change rate calculation formula executed by the adaptive temperature processing unit is: ΔT(k)=T m (k)-T m (k - 1); where, T m (k) represents the currently collected temperature signal, T m (k - 1) represents the temperature signal collected at the previous moment, and ΔT(k) represents the current temperature change rate.

3. The thermally expandable self - adapting gate valve according to claim 1, wherein: The dynamic filtering parameter adjustment formula executed by the adaptive temperature processing unit is: α(k)=α base +β·exp(-γ·|ΔT(k)|^2); where, α(k) represents the filtering coefficient, α base represents the basic filtering coefficient, and its value range is 0.30 - 0.95, β represents the filtering parameter gain, and its value range is 0.05 - 0.50, γ represents the filtering parameter attenuation coefficient, and its value range is 0.5 - 2.5, and ΔT(k) represents the current temperature change rate.

4. The thermally-expansion self-adaptive gate valve according to claim 1, wherein: The adaptive filtering formula executed by the adaptive temperature processing unit is: T f (k) = α(k)·T f (k - 1) + (1 - α(k))·T m (k); where T f (k) represents the filtered temperature signal, T f (k - 1) represents the filtered temperature signal at the previous moment, α(k) represents the filtering coefficient, and T m (k) represents the currently acquired temperature signal.

5. The thermal expansion self - adapting gate valve according to claim 1, wherein: The non - linear temperature compensation formula executed by the adaptive temperature processing unit is: T corr (k)=T f (k)+∑[i = 1 to n]c i ·f i (T f (k)); where, T corr (k) represents the compensated temperature signal, T f (k) represents the filtered temperature signal, c i represents the compensation coefficient, f i (T f (k)) represents the characteristic function, which is used to describe the non - linear temperature characteristic, and n represents the number of characteristic functions.

6. The thermally expandable self-adaptive gate valve according to claim 1, characterized in that: The formula for constructing the sealing performance objective function executed by the sealing position optimization control unit is: J(p)=w1·(p - p ref (T corr ))^2 + w2·R leak (p, T corr ) + w3·F contact (p, T corr );where J(p) represents the objective function, p represents the current sealing surface position, p ref (T corr ) represents the ideal sealing position calculated based on the compensated temperature, R leak (p, T corr ) represents the leakage rate estimation function, F contact (p, T corr ) represents the contact force estimation function, and w1, w2, and w3 represent the weight coefficients, corresponding to the weights of position, leakage rate, and contact force respectively.

7. The thermal expansion self-adaptive gate valve according to claim 1, characterized in that: The calculation formula for the position update amount executed by the sealing position optimization control unit is: p(k + 1) = p(k) - η(k)·∇J(p(k)); where, p(k + 1) represents the sealing surface position at the next moment, p(k) represents the current sealing surface position, η(k) represents the current step size, and ∇J(p(k)) represents the objective function gradient.

8. The thermally expandable self-adaptive gate valve according to claim 1, wherein: The adaptive step size adjustment formula executed by the sealed position optimization control unit is: η(k)=η base ·exp(-λ·k)+η min ; where η(k) represents the current step size, η base represents the base step size, with a value range of 0.08 - 0.20, λ represents the step size attenuation coefficient, with a value range of 0.04 - 0.12, k represents the number of iterations, and η min represents the minimum step size, with a value range of 0.01 - 0.

03.

9. The thermally expandable self-adaptive gate valve according to claim 1, wherein: The fusion filtering parameter optimization formula executed by the algorithm fusion interaction unit is: α adaptive (k)=α base +β·exp(-γ·|ΔT(k)|^2)+δ·|p(k)-p optimal (T f (k))|; where, α adaptive (k) represents the filtering coefficient after fusion optimization, α base represents the basic filtering coefficient, β represents the filtering parameter gain, γ represents the filtering parameter attenuation coefficient, ΔT(k) represents the current temperature change rate, δ represents the position feedback gain coefficient, and its value range is 0.03 - 0.08, p(k) represents the current seal surface position, p optimal (T f (k)) represents the ideal position calculated based on the filtered temperature.

Images