A temperature-controlled pressure-controlled combined flow control method and controller

Abstract

The application provides a temperature and pressure combined flow control method and a controller, which are applied to the technical field of fluid control, and through establishing an extended thermodynamic model containing real-time pressure signals and channel outer wall temperature, pressure and environmental temperature are directly embedded into the inverse solution process of mass flow as dynamic compensation parameters to eliminate the interference of gas inlet pressure fluctuation and environmental temperature change on measurement accuracy. Meanwhile, through constant temperature difference closed loop control, the unique mapping relationship between heating power and flow is ensured, and the control valve is adjusted in real time through the adaptive control algorithm to realize high-precision and high-stability flow output. Therefore, the application has the advantages of effectively eliminating the interference of pressure and environmental temperature fluctuation on measurement accuracy without relying on frequent offline calibration, and significantly improving the flow measurement accuracy and stability under complex working conditions.

Description

Technical Field

[0001] This application relates to the field of fluid control technology, and in particular to a flow control method and controller combining temperature control and pressure control. Background Technology

[0002] Traditional thermal mass flow controllers (MFCs) are based on King's Law, measuring gas mass flow rate through the temperature difference created by a heating element and a temperature sensor placed in the flow channel. This technology has the following main drawback: in practical applications, the instrument coefficient K is highly susceptible to fluctuations in gas inlet pressure and ambient temperature, leading to a decrease in measurement accuracy.

[0003] Because existing technologies cannot effectively eliminate system deviations caused by pressure and temperature changes over a wide measurement range, frequent offline calibrations are required, severely limiting their use in fields with extremely high flow control accuracy requirements, such as precision chemicals and semiconductor processes. To maintain measurement stability, traditional solutions are often forced to increase structural complexity, yet still struggle to achieve stable output across the entire measurement range.

[0004] Therefore, how to eliminate the interference of pressure and ambient temperature fluctuations on measurement accuracy and improve the measurement accuracy and stability of flow controllers under complex operating conditions is a key problem that urgently needs to be solved in the field of fluid control.

[0005] Therefore, existing technologies urgently need to be improved to address the aforementioned problems. Summary of the Invention

[0006] In view of the shortcomings of the prior art, this application provides a flow control method and controller that combines temperature control and pressure control. It is applied to the field of fluid control technology. By using an extended thermodynamic model to introduce pressure and outer wall temperature compensation terms, it realizes real-time and accurate correction of fluid flow rate, effectively overcomes the influence of pressure and temperature fluctuations on the accuracy of flow measurement, and significantly improves the control accuracy of fluid flow rate.

[0007] In a first aspect, a flow control method combining temperature and pressure control is provided, the method comprising the steps of: S1: Control the heating element to heat the fluid flowing through the channel, and collect the temperature difference signal before and after the fluid is heated, the real-time pressure signal of the fluid in the channel, and the temperature of the outer wall of the channel. S2: Based on the temperature difference signal, the heating power of the heating element is controlled through closed-loop feedback to maintain the temperature difference signal at a preset constant temperature difference; S3: Under constant temperature difference conditions, monitor the heating power of the heating element in real time, and substitute the heating power, the real-time pressure signal and the temperature of the outer wall of the channel into the preset extended thermodynamic model to obtain the real-time mass flow rate of the fluid. S4: Compare the real-time quality flow rate with the preset set flow rate, and calculate the flow rate error and the error change rate; S5: Input the flow error and the error change rate into the adaptive control algorithm to generate a control signal and drive the control valve to adjust the fluid flow rate.

[0008] Furthermore, after step S1, the following is also included: S11: Synchronously sample the temperature difference signal and the real-time pressure signal; S12: The Kalman filter algorithm is used to perform multimodal fusion processing on the synchronously sampled temperature difference signal and the real-time pressure signal to obtain the optimal estimated temperature difference signal and the optimal estimated real-time pressure signal after noise suppression. S13: The optimal estimated temperature difference signal is used as the temperature difference signal to facilitate subsequent control of the heating power of the heating element, and the optimal estimated real-time pressure signal is used as the real-time pressure signal to facilitate subsequent calculation in the preset thermodynamic model.

[0009] Furthermore, in step S3, the heating power is calculated by detecting the real-time voltage and real-time current of the heating element.

[0010] Furthermore, in step S3, the extended thermodynamic model is: ,in, The heating power, For real-time quality flow to the power of n, For real-time pressure signals, For standard pressure, The outer wall temperature, The standard temperature is represented by A, B, n, k, m, and C, which are instrument coefficients. A represents the zero-flow power bias coefficient, B represents the mass flow rate to heating power ratio coefficient, C represents the system power offset constant, n represents the flow nonlinearity index, k represents the pressure compensation index, and m represents the temperature compensation index.

[0011] Furthermore, step S5 includes: S51: Based on the flow error and the error change rate, determine the proportional coefficient of the adaptive control algorithm online using fuzzy inference rules. Integral coefficient and differential coefficients ; S52: Adjust the aforementioned proportional coefficient Integral coefficient and differential coefficients Substitute into adaptive control algorithm The control signal is obtained; among which, For control signals, The flow error is... The error change rate is given.

[0012] In a second aspect, a temperature and pressure control combination controller is provided, the controller being used to implement the steps of any of the methods described above, the controller comprising at least: a main channel, a control valve disposed on the main channel, a flow sensing unit, and a microprocessor controller; The flow sensing unit includes a heating element, a temperature sensor, and a pressure sensor; The microprocessor controller is electrically connected to the heating element, temperature sensor, pressure sensor, and control valve; The microprocessor controller is used to control the heating element to heat the fluid flowing through the channel, and to collect the temperature difference signal before and after the fluid is heated, the real-time pressure signal of the fluid in the channel, and the temperature of the outer wall of the channel. Based on the temperature difference signal, the heating power of the heating element is controlled through closed-loop feedback to maintain the temperature difference signal at a preset constant temperature difference. Under constant temperature difference conditions, the heating power of the heating element is monitored in real time. The heating power, the real-time pressure signal, and the temperature of the outer wall of the channel are substituted into a preset extended thermodynamic model, and the real-time mass flow rate of the fluid is obtained by inverse solution. The real-time quality flow rate is compared with the preset set flow rate, and the flow rate error and error change rate are calculated. The flow error and the rate of change of the error are input into the adaptive control algorithm to generate a control signal and drive the control valve to adjust the fluid flow rate.

[0013] Furthermore, the temperature sensor is a high-frequency response MEMS thermopile structure based on graphene or boron nitride nanosheets, and the response time of the high-frequency response MEMS thermopile structure is less than 10ms.

[0014] Furthermore, the temperature sensor includes at least an upstream temperature sensor, a downstream temperature sensor, and a reference temperature sensor; The upstream temperature sensor and the downstream temperature sensor are disposed inside the main channel and located on both sides of the heating element, respectively, to collect the upstream temperature value and the downstream temperature value of the fluid, so as to calculate the temperature difference signal of the fluid before and after being heated. The reference temperature sensor is disposed outside the main channel and is used to measure the temperature of the outer wall. The pressure sensor is a MEMS pressure sensor integrated inside the main flow channel, used to collect the real-time pressure signal of the fluid.

[0015] Furthermore, the control valve is a piezoelectric ceramic driven valve.

[0016] Furthermore, the heating element is a micron-sized platinum thin-film resistor, which is sputtered or electroplated onto the inner wall of the main channel.

[0017] Beneficial Effects: The temperature- and pressure-controlled combined flow control method and controller proposed in this application establishes an extended thermodynamic model that simultaneously incorporates real-time pressure signals and channel outer wall temperatures. Pressure and ambient temperature are directly embedded as dynamic compensation parameters into the inverse process of mass flow rate, fundamentally eliminating the interference of gas inlet pressure fluctuations and ambient temperature changes on measurement accuracy. This ensures the instrument coefficient remains stable across the entire range, eliminating the need for frequent offline calibration. Simultaneously, constant temperature difference closed-loop control ensures a unique mapping relationship between heating power and flow rate. Combined with an adaptive control algorithm for real-time adjustment of the control valve, high-precision and high-stability flow output can be achieved even under wide range and variable operating conditions, effectively meeting the stringent flow control requirements of cutting-edge fields such as semiconductor processes and precision chemicals. Therefore, this application has the advantages of effectively eliminating the interference of pressure and ambient temperature fluctuations on measurement accuracy without relying on frequent offline calibration, significantly improving the accuracy and stability of flow measurement under complex operating conditions. Attached Figure Description

[0018] Figure 1 This is a flowchart of a flow control method combining temperature and pressure control proposed in this application.

[0019] Figure 2 This is an overall structural diagram of a temperature and pressure control combination controller proposed in this application.

[0020] Figure 3 This is a schematic diagram of the shunt structure of a temperature and pressure control combination controller proposed in this application.

[0021] Figure 4 Another schematic diagram of the shunt structure of the controller for the temperature control and pressure control combination proposed in this application.

[0022] Figure 5 This is a structural diagram of the laminar flow cavity of a temperature and pressure control combination controller proposed in this application.

[0023] Figure 6 This is a structural diagram of the control valve of a temperature and pressure control combination controller proposed in this application.

[0024] Labeling Explanation: 1. Controller; 2. Main Flow Channel; 3. Flow Sensing Unit; 31. Heating Element; 32. Temperature Sensor; 33. Pressure Sensor; 4. Microprocessor Controller; 5. Control Valve; 21. Laminar Flow Chamber; 22. Flow Diversion Structure. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and marked in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0026] In semiconductor vapor deposition processes or fuel cell gas supply systems, precise control of gas flow directly determines the yield of the final product or the power generation efficiency of the system. Traditional thermal mass flow controllers are typically based on fixed thermodynamic empirical formulas for measurement. However, in complex industrial applications, frequent fluctuations in gas source pressure and drastic changes in ambient temperature can directly disrupt the thermal equilibrium between the fluid and the heating element 31.

[0027] Furthermore, in extremely low or high flow rates, traditional control methods often suffer from low signal-to-noise ratios or poor linearity, making it difficult to meet the stringent requirements of modern precision manufacturing for wide flow ranges, high precision, and rapid response. Moreover, in semiconductor vapor deposition scenarios, the stability of the supply of precursor gases, carrier gases, or reactant gases directly affects the uniformity of film thickness, compositional consistency, and deposition repeatability; in fuel cell gas supply scenarios, deviations from the target value in cathode or anode gas supply can easily lead to incomplete reactions, localized oversupply, or decreased system efficiency.

[0028] Therefore, flow control is not only a simple execution control problem, but also involves the coordinated design of thermal measurement accuracy, environmental disturbance compensation capability, and valve control dynamic response capability.

[0029] To address this practical need, this solution integrates constant temperature difference measurement, pressure and external wall temperature compensation, and adaptive valve control into the same technical chain, so that measurement and control are no longer isolated from each other, but form a continuous closed loop within the same operating cycle.

[0030] For details, please refer to Figure 1 A flow control method combining temperature and pressure control, the method comprising the following steps: S1: Control the heating element 31 to heat the fluid flowing through the channel, and collect the temperature difference signal before and after the fluid is heated, the real-time pressure signal of the fluid in the channel, and the temperature of the outer wall of the channel. S2: Based on the temperature difference signal, the heating power of the heating element 31 is controlled through closed-loop feedback to maintain the preset constant temperature difference. S3: Under constant temperature difference conditions, monitor the heating power of heating element 31 in real time, substitute the heating power, real-time pressure signal and channel outer wall temperature into the preset extended thermodynamic model, and obtain the real-time mass flow rate of the fluid by inverse solution. S4: Compare the real-time quality flow rate with the preset set flow rate, and calculate the flow error and error change rate; S5: Input the flow error and error change rate into the adaptive control algorithm to generate a control signal and drive the control valve 5 to adjust the fluid flow rate.

[0031] Constant temperature difference refers to the ability of dynamically adjusting the heating power of the heating element 31 during fluid flow to maintain the temperature difference before and after the fluid flows through the heating element 31 at a preset fixed value, which serves as the reference state for flow measurement. Furthermore, this preset fixed value is not necessarily limited to a single value, but is a target temperature difference value ΔTset pre-selected based on the fluid type, channel structure, heat capacity of the heating element 31, target range, and allowable temperature rise range.

[0032] When the system is running, if the actual temperature difference is detected to be lower than ΔTset, it indicates that the heat carried away by the fluid per unit time has increased, and the control system will increase the output of the heating element 31; if the actual temperature difference is detected to be higher than ΔTset, it indicates that the current heat input is relatively large, and the control system will reduce the output of the heating element 31.

[0033] In this way, this application does not directly use the magnitude of the temperature difference signal as the final flow rate value, but indirectly characterizes the flow rate by the heating power required to maintain a constant temperature difference. The reason for this is that when the temperature difference is stably constrained to the same reference state, the heat exchange comparison basis under different operating conditions is more consistent, and the stability of the subsequent flow rate inversion is higher.

[0034] The extended thermodynamic model differs from the traditional ideal model that only considers temperature and power. This model is a mathematical physics model that introduces real-time fluid pressure and external ambient temperature as compensation variables, used to more accurately describe the nonlinear mapping relationship between heating power and fluid mass flow rate under complex operating conditions. Furthermore, in this scheme, the external ambient temperature is preferably represented by the channel outer wall temperature. This is because the channel outer wall temperature can simultaneously reflect changes in the ambient temperature around the equipment and changes in the heat dissipation state of the channel itself, and compared to simply collecting the ambient air temperature far from the channel, it more closely approximates the actual thermal boundary conditions of the heating element 31.

[0035] The nonlinear mapping relationship between heating power and fluid mass flow rate means that under different pressures, different external wall temperatures, and different flow ranges, the same change in heating power does not necessarily correspond to the same magnitude of flow rate change. Therefore, it is necessary to establish a multivariate correlation through pre-calibration, rather than relying solely on a single empirical formula. For example, when the pressure increases, the gas density and convective heat transfer characteristics will change; when the channel's external wall temperature increases, the temperature difference conditions for heat dissipation from the channel to the outside will also change. Both of these factors will affect the actual power required to maintain a constant temperature difference. If not compensated for, environmental changes will be misinterpreted as flow rate changes.

[0036] In one feasible approach, the raw temperature data of the fluid before and after heating can be directly acquired by a thermistor deployed inside the channel, and the difference can be calculated to obtain a temperature difference signal. At the same time, the raw pressure value inside the channel can be directly read by a pressure transmitter, and these raw signals can be directly transmitted to the microprocessor controller 4 for further processing.

[0037] To more clearly and completely describe the acquisition process, the channel can be divided into an upstream temperature measurement zone, a heating zone, and a downstream temperature measurement zone along the fluid flow direction. The thermistor in the upstream temperature measurement zone is used to sense the temperature of the fluid before it enters the heating zone, while the thermistor in the downstream temperature measurement zone is used to sense the temperature of the fluid after it leaves the heating zone. The installation positions of both the upstream and downstream thermistors are preferably maintained in a stable geometric relationship with respect to the heating zone to reduce the impact of installation deviations on temperature difference measurement.

[0038] Pressure transmitters can be placed on the side wall of the channel near the heating zone or at a pressure tapping location connected to the main channel so that the measured pressure can represent the actual fluid pressure near the heating zone.

[0039] The temperature of the outer wall of the channel can be obtained by a reference temperature sensor attached to the outer surface of the channel, so as to eliminate the influence of ambient temperature fluctuations.

[0040] After receiving the raw signals from each channel, the microprocessor controller 4 can first perform analog-to-digital conversion, zero-point correction, range normalization, and timestamp alignment before proceeding to the subsequent control and calculation process. If a thermistor is used as the temperature sensing element, a static calibration can also be performed during the initialization phase after the system is powered on. This involves recording the baseline outputs of the upstream and downstream temperature sensing zones under no-current or known stable-current conditions to eliminate fixed errors caused by individual device differences and installation bias.

[0041] After proceeding to step S2, this application constructs a closed-loop feedback loop. When the fluid flow rate increases and carries away more heat, causing the temperature difference signal to decrease, the closed-loop feedback mechanism will rapidly increase the heating power; conversely, it will decrease the heating power, thereby forcing the system to maintain a preset constant temperature difference state.

[0042] Furthermore, this closed-loop feedback loop can be understood as a temperature servo system with the temperature difference signal as the controlled object and the heating power as the adjustable object. The microprocessor controller 4 reads the current temperature difference signal in each control cycle and compares it with the preset constant temperature difference ΔTset to obtain the temperature difference deviation. If the temperature difference deviation persists, it indicates that a new balance has not yet been reached between the current heating power and the heat carried away by the fluid. The system then continues to adjust the output of the heating element 31 until the actual temperature difference returns to near the preset constant temperature difference.

[0043] To avoid excessive temperature fluctuations caused by overly strong control, the closed-loop feedback mechanism can include amplitude limiting, gradual increase and decrease, or output rate of change constraints to ensure that the heating power adjustment is both fast enough and not excessively drastic. The significance of this approach is that the constant temperature difference state itself serves as the measurement benchmark for subsequent flow rate inversion. If this benchmark is unstable, even with a reasonably designed extended thermodynamic model, the reliability of the flow rate calculation will be reduced due to continuous oscillations in the input power.

[0044] Therefore, the main function of step S2 is to construct stable measurement conditions.

[0045] Under constant temperature difference conditions, there is a functional relationship between heating power and fluid mass flow rate (i.e., a pre-defined extended thermodynamic model): ,in, For heating power, For real-time quality flow to the power of n, For real-time pressure signals, For standard pressure, The outer wall temperature, The standard temperature is represented by A, B, n, k, m, and C, which are instrument coefficients. A represents the zero-flow power bias coefficient, B represents the mass flow rate to heating power ratio coefficient, C represents the system power offset constant, n represents the flow nonlinearity index, k represents the pressure compensation index, and m represents the temperature compensation index.

[0046] In step S3, the aforementioned preset extended thermodynamic model is inversely solved, referring to the real-time acquisition of the heating power of the heating element 31. Real-time pressure signal P and outer wall temperature By substituting the values ​​into the model and solving in reverse, the real-time mass flow rate can be obtained. .

[0047] To maintain the constant temperature difference, the actual heating power consumed is determined, and then, based on a pre-established multivariate correspondence, the mass flow rate value matching this power state is obtained or calculated. To ensure this step is practically feasible, the pre-defined extended thermodynamic model needs to be calibrated before the system is put into use. Specifically, a calibration platform can be built, connecting the flow control device under test in series in the calibration loop containing the standard flow source or standard flow meter, and operating the device under multiple different pressure conditions, multiple different channel outer wall temperatures, and multiple different flow points. At each calibration point, the heating power, real-time pressure signal, channel outer wall temperature, and standard mass flow rate value corresponding to the stable constant temperature difference are recorded. Subsequently, these calibration data are stored in a unified format, forming a dataset covering the target operating range. Then, the correspondence between heating power and mass flow rate is established based on this dataset.

[0048] During long-term industrial operation, the heating element 31 is affected by thermal stress and material aging, and its resistance value often undergoes irreversible drift. At the same time, there may also be slight voltage fluctuations in the on-site power supply. In this case, if the actual heating power is equivalently replaced by the theoretical command value issued by the microprocessor controller 4 (that is, the heating power used in step S2 to maintain the temperature difference signal at a preset constant temperature difference using closed-loop feedback control), a significant deviation will occur between the theoretical value and the actual physical consumption value.

[0049] This distortion of power parameters, once substituted into the extended thermodynamic model, will directly lead to systematic errors in the flow rate inverse kinematics results. Furthermore, this error is insidious: while closed-loop isothermal control may superficially maintain temperature stability, if the power value input to the model deviates from the true value, the system will continuously output erroneous mass flow rate results even under the premise of stable temperature difference. In other words, temperature difference control addresses the problem of maintaining thermal equilibrium, while accurate power measurement addresses the problem of flow rate inverse kinematics reference; both are indispensable.

[0050] To eliminate errors caused by such underlying hardware changes, this specific implementation scheme clarifies the precise method for obtaining heating power. More specifically, a precision sampling resistor with an extremely low temperature coefficient can be connected in series in the power supply circuit of the heating element 31, and a high-precision voltage detection circuit can be connected in parallel. During each control cycle of system operation, the actual voltage drop applied across the heating element 31 and the actual current flowing through the heating element 31 are acquired in real time. Subsequently, the real-time voltage and real-time current are multiplied by a hardware multiplier or an underlying software algorithm to obtain the actual physical heating power currently consumed by the heating element 31. That is, in step S3, the heating power is calculated by detecting the real-time voltage and real-time current of the heating element 31.

[0051] The detected heating power Real-time pressure signal P and outer wall temperature Substitute into the extended thermodynamic model During the reverse solution process The effect of pressure on gas density is introduced, and k represents the pressure compensation index. The magnitude of k determines the compensation weight for pressure fluctuations. This introduces a correction for the heat dissipation boundary conditions based on ambient temperature, where m represents the temperature compensation index, reflecting the nonlinear impact of changes in the outer wall temperature on the overall thermal balance.

[0052] After the fluid flows through the heating zone, the boundary layer thickness, local velocity distribution, and gas thermal properties all change. Especially in the high flow rate range, increasing the flow rate by the same amount may not necessarily result in a completely proportional increase in heat removal capacity. The pressure compensation term is used to incorporate the density change of the gas after being compressed into the model. For example, the same volumetric flow rate often corresponds to a larger mass flow rate at higher pressures, and the convective heat transfer capacity will also change accordingly. The temperature compensation term is used to correct for changes in heat dissipation conditions between the outer wall of the main flow channel 2 and the surrounding environment. For example, although the internal fluid flow rate is the same, the heat exchange boundary on the outer wall is significantly different when the equipment is installed in a temperature-controlled cavity versus an ambient temperature cabinet.

[0053] Therefore, this model does not make empirical corrections to a single sensing quantity, but rather incorporates zero-point heat dissipation, heat transfer changes caused by flow rate, density changes caused by pressure, and boundary condition changes caused by external wall temperature into the same inverse solution framework, thus reflecting the real working conditions more completely.

[0054] Through this specific implementation scheme, the system completely overcomes the limitations of traditional empirical formulas at the underlying algorithm level. This extended thermodynamic model can not only adaptively adjust the compensation weights of pressure and temperature, but also maintain extremely high fitting accuracy across the entire range.

[0055] Its technical advantage lies in the fact that, in semiconductor processes, even if the gas source pressure changes abruptly or the temperature around the reaction chamber fluctuates drastically, the model can instantly correct the mapping relationship between power and flow rate, avoiding misinterpreting environmental disturbances as flow rate changes, thereby ensuring the absolute stability of the precursor gas supply and the uniformity of thin film deposition.

[0056] Further analysis reveals that traditional thermal mass flow controllers, based on King's Law, typically assume relatively constant heat exchange conditions and only perform first-order corrections on a single variable. Therefore, rapid changes in external pressure will manifest as abrupt changes in power demand, which can be easily interpreted as flow rate changes in simple models.

[0057] In this scheme, since pressure and external wall temperature are synchronously included in the calculation as inverse solution inputs, the microprocessor controller 4 can identify changes in heat transfer capacity caused by changes in gas state, rather than a uniform change in flow rate itself. Similarly, when the temperature around the chamber rises, the heat dissipation from the main flow channel 2 to the outside weakens, and the heating power required to maintain the same constant temperature difference may decrease. Without external wall temperature compensation, the system would mistakenly assume a decrease in flow rate; however, in this model, the external wall temperature term synchronously corrects for this change, ensuring that the real-time mass flow rate obtained from the inverse solution remains stable.

[0058] In other words, the effect of this technology does not come from the optimization of a single parameter, but from the result of the unified modeling of the coupling relationship between power, pressure, outer wall temperature, and mass flow rate. Therefore, it has stronger robustness under wide pressure range, wide temperature range, and dynamic disturbance conditions.

[0059] After achieving high-precision real-time flow inverse kinematics, the system enters the flow regulation execution phase. As mentioned earlier, an adaptive control algorithm is used to regulate fluid flow. In fuel cell gas supply scenarios, frequent load changes require the gas supply system to have extremely fast dynamic response capabilities.

[0060] However, gas flow has inherent physical hysteresis and nonlinearity. Traditional fixed-parameter PID controllers are prone to problems such as excessive overshoot, excessively long settling time, or even system oscillation when faced with large-scale flow step changes or external pressure disturbances, leading to an imbalance in the supply of fuel cell reactant gas.

[0061] To solve this problem, a strategy is needed that can dynamically adjust control parameters in real time based on the current flow error status.

[0062] Furthermore, step S5 includes: S51: Based on the flow error and the rate of change of error, the proportional coefficient of the adaptive control algorithm is determined online using fuzzy inference rules. Integral coefficient and differential coefficients ; S52: Adjust the proportional coefficient after tuning Integral coefficient and differential coefficients Substitute into adaptive control algorithm The control signal is obtained; among which, For control signals, For flow error, This represents the rate of change of error.

[0063] For ease of implementation, the fuzzy inference rules in step S51 can be pre-established offline and embedded in the microprocessor controller 4. Online tuning involves retrieving the corresponding parameter correction direction and magnitude from the preset rules within each control cycle based on the newly acquired flow error and error change rate. (Proportional coefficient) It mainly affects response speed and driving force, and the integral coefficient. The main factor affecting the ability to eliminate steady-state deviation is the differential coefficient. Its main effects are on suppressing rapid changes and enhancing damping.

[0064] The control signal in step S52 can be expressed as valve drive current, target valve core displacement value, PWM duty cycle, or control voltage of the valve actuator. The specific form can be determined according to the type of control valve 5. For example, for a proportional solenoid valve, the control signal can directly correspond to the coil drive current; for an electric regulating valve with position feedback, the control signal can correspond to the target valve stem opening.

[0065] This control strategy deeply integrates fuzzy logic reasoning with classic PID control. The core of the fuzzy reasoning rule lies in simulating the adjustment experience of human experts, dynamically allocating the weights of proportional, integral, and derivative actions based on the current deviation state of the system. Specifically, the system predefines fuzzy subsets (e.g., negative large, negative small, zero, positive small, positive large) of the flow error e(t) and the error change rate ec, and targets... , , The correction amount was used to establish a multidimensional fuzzy rule table.

[0066] Among them, "negative large, negative small, zero, positive small, positive large" are not required to be limited to five levels. They can also be subdivided into more levels according to the control accuracy requirements. However, in order to balance the amount of calculation and the control effect, it is preferable to use a finite set of levels with clear meanings.

[0067] In this context, fuzzy subsets can be understood as dividing continuously changing errors and their rates of change into several intervals according to control requirements, allowing for smooth transitions between these boundary intervals. For example, when the flow error is significantly positive and the rate of change is close to zero, it indicates that the actual flow rate is consistently lower than the target value and the catch-up speed is insufficient. In this case, the PID controller should emphasize rapid acceleration. Conversely, when the flow error is close to zero and the rate of change is significantly negative, it indicates that the system may be moving downwards from above the target value. In this situation, increased damping is needed to prevent oscillations.

[0068] The pre-defined fuzzy rule table can be obtained during the prototype debugging phase by experienced engineers through item-by-item testing, combining valve response characteristics, main channel volume 2, gas compressibility, and target control bandwidth. For example, within a large error range, different... , , By combining rise time, overshoot, and settling time, a set of parameter correction directions that balance speed and stability is selected. These experimental results are then organized into a correspondence between error states and parameter correction suggestions, solidified into a rule table. This rule table is derived from the physical response of the specific device and is easily accessible to the microprocessor controller 4 in real time.

[0069] In practice, the microprocessor calculates the current e(t) and ec in each control cycle and performs fuzzification. When the set flow rate experiences a large step change (i.e., e(t) is large), the fuzzy rule will output a larger value. To accelerate valve response while outputting a smaller amount of power. To prevent severe overshoot caused by integral saturation, the fuzzy rules will be appropriately increased when the actual flow rate rapidly approaches the setpoint (i.e., e(t) decreases but ec is large). To increase system damping and suppress overshoot tendency; when the actual flow rate is close to the set value and changes slowly (i.e., both e(t) and ec are small), the fuzzy rule will be increased. To eliminate steady-state error and reduce and To avoid frequent valve vibration around the target value, the microprocessor calculates specific parameter correction values ​​using defuzzification algorithms such as the centroid method, updates the coefficients in the PID formula in real time, and finally calculates the control signal u(t) to drive the control valve 5.

[0070] To further improve feasibility, in a specific example, the microprocessor can execute the following sequence: First, read the set flow rate and real-time quality flow rate for the current cycle to obtain the flow error; then, compare the flow error of the current cycle with the flow error of the previous cycle to obtain the error change rate; next, map these two quantities to the membership degree of the corresponding fuzzy subsets according to a pre-set quantization range; subsequently, simultaneously trigger multiple corresponding fuzzy rules to obtain the corresponding... , , Several modification suggestions were proposed; then, these modification suggestions were synthesized into the proportional coefficient, integral coefficient, and derivative coefficient actually used in this cycle using the center of gravity method; finally, the drive signal for control valve 5 was output by combining the upper limit, lower limit, and integral limiting strategy of the control signal.

[0071] In practical engineering, two additional protective measures are preferred: First, when the valve has reached its mechanical limit opening or limit closing, the integral term is limited from accumulating further to prevent integral saturation caused by long-term deviation; second, the error rate of change signal is lightly filtered to suppress noise-induced differential term jitter. This avoids frequent micro-movements of control valve 5 due to measurement noise when it approaches its stable point.

[0072] Furthermore, in some preferred embodiments, step S1 is followed by: S11: Synchronously sample the temperature difference signal and the real-time pressure signal; S12: The Kalman filter algorithm is used to perform multi-modal fusion processing on the synchronously sampled temperature difference signal and real-time pressure signal to obtain the optimal estimated temperature difference signal and the optimal estimated real-time pressure signal after noise suppression. S13: Use the optimal estimated temperature difference signal as the temperature difference signal to facilitate subsequent control of the heating power of the heating element 31, and use the optimal estimated real-time pressure signal as the real-time pressure signal to facilitate subsequent calculation in the preset thermodynamic model.

[0073] In real-world industrial environments, fluid pipelines are often accompanied by vibrations from mechanical pumps, interference from electromagnetic equipment, and turbulent pulsations within the fluid itself. These external disturbances cause a significant amount of high-frequency random noise to be mixed into the directly acquired raw temperature difference and pressure signals. If closed-loop control is directly based on these unprocessed raw signals, it can easily lead to frequent and drastic fluctuations in heating power, subsequently causing oscillations in the entire flow control system. Furthermore, although the temperature difference and pressure signals originate from different sources, they are both affected by the same flow process, thus exhibiting temporal and physical correlation. If the sampling times of the two signals are inconsistent, the temperature difference signal at one moment may correspond to the pressure state of the previous instant, while the pressure signal at another moment may correspond to the flow state of the next instant. This temporal misalignment weakens the effectiveness of joint judgment based on multi-source information. Especially during periods of rapid flow change or when valves have just actuated, the temporal misalignment is further amplified, causing controller 1 to make delayed or distorted judgments about the system state.

[0074] To address this issue, this specific implementation introduces a synchronous sampling and Kalman filtering multimodal fusion mechanism. Referring to the following example, in step S11, the system uses a high-precision timer to trigger the analog-to-digital converter, ensuring that the temperature difference signal and pressure signal are acquired within the same microsecond-level time window, thus guaranteeing strict alignment of the multi-source data in the time dimension.

[0075] In step S12, the system pre-constructs state transition equations and observation equations that incorporate fluid dynamics characteristics. During operation, the Kalman filter algorithm dynamically calculates the Kalman gain based on the current observation value and the previous prediction value, thereby extracting the optimal estimated temperature difference signal and the optimal estimated real-time pressure signal that best approximate the true physical state from the noisy measurement data.

[0076] To explain the process more clearly and completely, it can be broken down into the following consecutive steps: First, in step S11, the main control chip simultaneously starts the temperature acquisition channel and the pressure acquisition channel through a unified time base, so that the two data have the same or corresponding timestamps; if a multiplexed analog-to-digital converter is used in the hardware, quasi-synchronous sampling can be achieved by fixing the sampling order and the extremely short switching interval, and the data of the same sampling batch are bound into a group in the software.

[0077] Secondly, in step S12, the system needs to pre-set the initial state, noise level estimation, and state update rules required for filtering. This preset information can be obtained through experiments: under typical operating conditions such as stable flow, step flow, and pressure disturbance, a segment of raw temperature difference signal and pressure signal is collected, and their fluctuation characteristics are observed. Based on this, a reasonable range for the strength of measurement noise and the rate of state change is determined, and the corresponding parameters are written into the program.

[0078] During operation, the Kalman filter algorithm first predicts the current state based on the estimation result from the previous time step, and then fuses the observations obtained from the current synchronous sampling with the prediction result. If the current observation fluctuates abnormally and is inconsistent with the continuous change trend of the system, the filtering result will retain more prediction information; if the current observation is consistent with the change trend of the system, the filtering result will absorb more current observation information.

[0079] In this way, random spike noise, short-term jitter, and local abnormal sampled values ​​will not be directly transmitted to subsequent control loops. Step S13 then replaces the original inputs with the filtered optimal estimated temperature difference signal and the optimal estimated real-time pressure signal, respectively, for continued use in steps S2 and S3, thereby establishing constant temperature difference control and flow inversion on a more stable data basis.

[0080] The aforementioned flow control method combining temperature and pressure control needs to be implemented on specific hardware devices. To ensure efficient coordination between high-speed acquisition of multi-dimensional state information, real-time calculation of the extended thermodynamic model, and precise actuation of control valve 5 within the same operating cycle, a highly integrated physical device needs to be constructed. This device not only requires a reasonable fluid channel design but also must achieve deep integration of sensing, computation, and execution in its hardware architecture.

[0081] Therefore, please refer to Figure 2 This application also provides a temperature control and pressure control combination controller, the controller 1 is used to implement the steps of any of the above methods, and the controller 1 includes at least: a main channel 2, a control valve 5 disposed on the main channel 2, a flow sensing unit 3 and a microprocessor controller 4; The flow sensing unit 3 includes a heating element 31, a temperature sensor 32, and a pressure sensor 33; The microprocessor controller 4 is electrically connected to the heating element 31, the temperature sensor 32, the pressure sensor 33, and the control valve 5; The microprocessor controller 4 is used to control the heating element 31 to heat the fluid flowing through the channel, and to collect the temperature difference signal before and after the fluid is heated, the real-time pressure signal of the fluid in the channel, and the temperature of the outer wall of the channel. Based on the temperature difference signal, the heating power of the heating element 31 is controlled through closed-loop feedback to maintain the preset constant temperature difference. Under constant temperature difference conditions, the heating power of the heating element 31 is monitored in real time. The heating power, real-time pressure signal and channel outer wall temperature are substituted into the preset extended thermodynamic model and the real-time mass flow rate of the fluid is obtained by inverse solution. The real-time quality flow rate is compared with the preset set flow rate, and the flow error and error change rate are calculated. The flow error and the rate of change of error are input into the adaptive control algorithm to generate a control signal and drive the control valve 5 to adjust the fluid flow.

[0082] For details, please refer to [link / reference]. Figures 3 to 5 The main flow channel 2 can be rectangular, elliptical, or tunnel-shaped to facilitate uniform gas flow. A heating and diversion assembly is installed within the main flow channel 2. This assembly consists of stacked multi-layer diversion structures 22. Each layer of diversion structure 22 distributes the incoming airflow evenly in stages, dividing it into two, three, or four equal multiples. These diversion structures 22 are arranged at intervals along the fluid flow direction within the main flow channel 2, thus dividing part of the main flow channel 2 into several laminar flow chambers 21. Each laminar flow chamber 21 is composed of microchannels between adjacent diversion layers. After being diverted stage by stage, the gas enters the corresponding laminar flow chamber 21 and, under the constraint of the microchannels, forms a laminar flow state with stable velocity and parallel streamlines.

[0083] Downstream temperature and pressure sensors 33 are installed within the laminar flow cavity 21 to measure the temperature of the heated fluid and the real-time pressure signal of the fluid, respectively. Through the coordinated arrangement of the multi-layered flow distribution structure 22, which uniformly distributes the gas in stages and simultaneously forms the laminar flow cavity 21, a stable laminar flow state is maintained for the gas across a wide flow range. Furthermore, the heated gas is uniformly distributed within the microchannels of each laminar flow cavity 21, significantly improving the consistency of temperature difference detection and the repeatability of flow measurement.

[0084] The heating and flow-diverting components and flow-diverting structure 22 can be cylindrical or square. Each laminar flow cavity 21 is uniformly distributed within the component.

[0085] In practical implementation, the microprocessor controller 4 integrates a high-resolution analog-to-digital converter (ADC) and a high-frequency pulse-width modulation (PWM) generator. The ADC channels are connected to the signal conditioning circuits of the temperature sensor 32 and the pressure sensor 33, respectively, to achieve synchronous high-speed acquisition of multi-dimensional signals; the PWM generator is connected to the control valve 5 through a power amplifier circuit, outputting a drive signal with an adjustable duty cycle. During operation, the microprocessor controller 4 executes signal acquisition, Kalman filter fusion, isothermal PID control, extended thermodynamic model inverse solution, and fuzzy PID valve control sequentially within a millisecond-level timer interrupt service routine. All computational tasks are completed in a closed loop within an extremely short time slice, ensuring absolute synchronization between measurement and control.

[0086] To more clearly demonstrate the specific implementation process, the following timing sequence can be adopted: At the beginning of each timer interrupt, multiple ADC channels are simultaneously triggered to sample the upstream temperature signal, downstream temperature signal, outer wall temperature signal, pressure signal, and necessary voltage and current detection signals; then, the sampled values ​​are fused using zero-point correction, gain correction, and Kalman filtering to obtain a set of state variables with lower noise and time alignment; then, constant temperature difference PID control is executed based on the temperature difference signal to update the heating element 31 drive; next, the current heating power, real-time pressure signal, and channel outer wall temperature are read, and a preset extended thermodynamic model is called to inversely solve the real-time mass flow rate; then, the real-time mass flow rate is compared with the set flow rate to obtain the flow error and error change rate, and a new valve control signal is generated through a fuzzy PID algorithm; finally, the corresponding duty cycle is output by the PWM generator, and after power amplification, it drives the control valve 5. If the control valve 5 is a current-type actuator, a current sampling closed loop can also be added to the drive circuit to ensure that the control signal is consistent with the actual execution current. By fixing the above actions to be executed within the same interrupt cycle, the problem of inconsistent data timestamps caused by the asynchronous operation of different modules can be avoided.

[0087] In the hardware architecture of the controller 1 described above, the performance of the temperature sensor 32 directly determines the bandwidth and accuracy of the constant temperature difference closed-loop control. In advanced processes such as atomic layer deposition (ALD), the pulse supply time of the precursor gas is typically in the tens of milliseconds or even shorter. Traditional thermistors or ordinary thin-film thermocouples, due to their large thermal mass, often have response times in the hundreds of milliseconds, making it impossible to capture transient changes in fluid temperature. This results in severe lag in constant temperature difference control, which in turn causes distortion in flow measurement.

[0088] To meet the stringent requirements of ultra-high-speed pulsed gas supply, this application proposes a preferred embodiment in which the temperature sensor 32 is a high-frequency response MEMS thermopile structure based on graphene or boron nitride nanosheets, and the response time of the high-frequency response MEMS thermopile structure is less than 10ms.

[0089] MEMS thermopile structures refer to miniature sensing structures that utilize microelectromechanical systems (MEMS) technology to form multiple series thermocouple nodes on tiny suspended membranes or microbridge structures, thereby converting minute temperature differences into measurable voltages. Graphene or boron nitride nanosheets are chosen primarily because these two types of two-dimensional nanomaterials still possess excellent thermal conductivity at extremely thin thicknesses and have very low overall heat capacity, making them suitable as part of high-speed thermal response structures.

[0090] A response time of less than 10ms is not an abstract metric, but can be obtained through pulse thermal excitation testing or gas temperature step testing. For example, a known rapid thermal disturbance is applied to the sensor in a standard test platform, and the time required for its output to reach a stable target change ratio is recorded. If this time is less than 10ms, it is considered to meet the high-frequency response requirement. This metric is particularly critical for millisecond-level pulse gas supply control, because only when the sensor response time is significantly shorter than the gas pulse width can controller 1 have the opportunity to complete effective detection and adjustment during the pulse duration.

[0091] The temperature sensor 32 of this application utilizes the excellent thermal properties of two-dimensional nanomaterials and the miniaturization advantages of microelectromechanical systems (MEMS) technology. Graphene and boron nitride nanosheets have extremely high in-plane thermal conductivity and extremely low specific heat capacity, which means that they can transmit temperature changes instantaneously with minimal heat absorption.

[0092] The thermopile structure is based on the Seebeck effect, which connects multiple miniature thermocouples in series. It can directly convert tiny temperature differences into measurable voltage signals without the need for external excitation current, thus avoiding Joule thermal interference from the sensor itself.

[0093] Furthermore, graphene is better suited for design paths emphasizing high thermal conductivity and rapid thermal diffusion, while boron nitride nanosheets have advantages in electrical insulation and thermal stability. Therefore, the choice between the two can be made based on the specific application environment. For example, in scenarios with higher requirements for electrical insulation and a strong driving electric field, boron nitride nanosheets can be preferentially used to construct the thermistor layer; while in scenarios that emphasize extreme response speed, graphene solutions can be given priority.

[0094] The advantage of thermopile structure is that its output is directly related to the temperature difference, and there is no need to continuously inject detection current. Therefore, it will not disturb the measured thermal field due to self-heating effect like traditional resistive temperature sensors.

[0095] Furthermore, the temperature sensor 32 includes at least an upstream temperature sensor, a downstream temperature sensor, and a reference temperature sensor; The upstream temperature sensor and the downstream temperature sensor are set inside the main channel 2 and located on both sides of the heating element 31. They are used to collect the upstream temperature value and the downstream temperature value of the fluid, respectively, to calculate the temperature difference signal before and after the fluid is heated. A reference temperature sensor is located outside the main channel 2 to measure the outer wall temperature; Pressure sensor 33 is a MEMS pressure sensor integrated inside the main channel 2, used to collect real-time pressure signals of the fluid.

[0096] In practical implementation, the upstream and downstream temperature sensors are preferably arranged symmetrically upstream and downstream of the heating element 31 along the fluid flow direction. This symmetrical arrangement can effectively eliminate common-mode interference from ambient baseline temperature drift on temperature difference calculation. When the fluid flows through the heating element 31, the fluid carries away heat and transfers it downstream, causing the downstream temperature to rise, while the upstream temperature remains basically at the initial temperature of the fluid.

[0097] The microprocessor controller 4 can obtain the temperature difference signal caused purely by fluid convection heat transfer by synchronously acquiring the outputs of the two sensors and performing differential calculations. In order to avoid the heat radiation of the heating element 31 directly irradiating the sensor and causing measurement distortion, the upstream and downstream temperature sensors and the heating element 31 must maintain a reasonable distance calculated by fluid dynamics.

[0098] The reference temperature sensor is tightly attached to the non-directly heated area of ​​the outer wall of the main channel 2 using thermally conductive silicone grease or silver paste with high thermal conductivity, and is covered with thermal insulation foam to ensure that the temperature it measures can truly represent the heat exchange boundary conditions between the main channel 2 and the external environment, thereby providing accurate outer wall temperature compensation input for the extended thermodynamic model.

[0099] In addition, the pressure sensor 33 employs a MEMS (Micro-Electro-Mechanical Systems) piezoresistive or capacitive pressure sensor. To integrate it inside the main flow channel 2 without disrupting the smoothness of the flow field, a very small pressure-guiding micro-hole can be fabricated on the sidewall of the main flow channel 2. The sensitive diaphragm of the MEMS pressure sensor communicates with the fluid inside the main flow channel 2 through this pressure-guiding micro-hole.

[0100] Through the above technical solution, the technical effects of this application are as follows: the built-in symmetrical layout of the upstream and downstream temperature sensors allows the temperature difference signal to reflect the actual situation of heat carried away by the fluid most directly and quickly, avoiding the signal lag and attenuation caused by the thermal resistance of the pipe wall in traditional external sensors; the independent external placement of the reference temperature sensor ensures the purity of the environmental thermal boundary information and prevents interference from the internal heating process. More importantly, the use of a MEMS pressure sensor and its integration via micro-orifice pressure tapping not only significantly reduces the dead volume of pressure measurement, allowing the dynamic response frequency of the pressure signal to match that of the temperature sensor 32, but also avoids the eddies and flow resistance generated in the flow channel by traditional large-volume pressure transmitters. This synergy of layout and selection provides extremely solid and reliable underlying data support for subsequent isothermal difference control and extended thermodynamic model calculations.

[0101] Even with an ultra-fast-response sensor network and a high-precision inverse kinematic model, the final execution stage of flow regulation still requires extremely high dynamic performance. While the aforementioned adaptive control algorithm can output precise control signals at the software level, the overall dynamic regulation effect of the system will still be significantly reduced if the physical actuator exhibits severe mechanical hysteresis or slow response. Especially in processes such as atomic layer deposition (ALD) requiring millisecond-level ultra-short pulse gas supply, traditional electromagnetic proportional valves are often limited by the current settling time due to coil inductance and hysteresis, making it difficult to meet the stringent requirements for rapid opening and closing. Therefore, this application further proposes that control valve 5 be a piezoelectric ceramic driven valve.

[0102] Please refer to Figure 6 In its implementation, the piezoelectric ceramic driven valve utilizes the inverse piezoelectric effect of piezoelectric materials, meaning that when an external electric field is applied, the piezoelectric ceramic material undergoes minute mechanical deformation. The actuator of this control valve 5 is typically composed of a piezoelectric stack made up of multiple layers of piezoelectric ceramic sheets. The control signal (such as a PWM signal or digital instruction) output by the microprocessor controller 4 is first converted into an analog drive voltage of 0 to several hundred volts by a dedicated high-voltage drive circuit, and then applied to both ends of the piezoelectric stack. As the drive voltage increases instantaneously, the piezoelectric stack undergoes longitudinal elongation within microseconds. This minute displacement is amplified by a mechanical amplification mechanism (such as a flexible hinge or lever structure) and directly pushes the valve core away from the valve seat, overcoming the spring preload, thereby opening the fluid passage or adjusting the flow area. When the drive voltage decreases or is removed, the piezoelectric stack quickly returns to its original shape, and the valve core instantly falls back under the action of the return spring, cutting off the fluid flow. To achieve high-precision flow regulation, the high-voltage drive circuit needs to have extremely low ripple and extremely high voltage resolution, so that the elongation of the piezoelectric stack can be continuously and smoothly controlled, thereby realizing stepless adjustment of the valve opening.

[0103] Furthermore, the heating element 31 is a micron-sized platinum thin film resistor, which is sputtered or electroplated onto the inner wall of the main channel 2.

[0104] In practical implementation, platinum (Pt) is chosen as the ideal heating material due to its excellent chemical stability, high-temperature resistance, and highly linear temperature coefficient of resistance. To directly fabricate platinum thin-film resistors on the inner wall of the main flow channel 2, specific heating areas of the inner wall of the main flow channel 2 first need to be insulated. For example, a dense and extremely thin silicon oxide or silicon nitride insulating layer can be grown using chemical vapor deposition (CVD) or atomic layer deposition (ALD) techniques to prevent heating current leakage into the metal flow channel body. Subsequently, a platinum thin film with a thickness of only micrometers or even submicrometers is deposited on the insulating layer using magnetron sputtering or precision electroplating. Next, the platinum thin film is etched into a serpentine or grid-like resistor pattern with specific resistance values ​​and heating areas using flexible photolithography or laser direct writing techniques. Finally, an extremely thin corrosion-resistant passivation layer (such as aluminum oxide or silicon carbide) is coated on the surface of the platinum thin film to protect the platinum resistor from physical erosion or chemical corrosion by the active components in the fluid. During system operation, the microprocessor controller 4 applies voltage or current directly to the platinum thin film resistor through the power drive circuit, and the generated Joule heat can be instantly transferred to the flowing fluid by simply penetrating the extremely thin passivation layer.

[0105] Through the above technical solution, the shape and heat transfer path of the heating element 31 have undergone revolutionary changes. Its unique technical effect lies in the extremely small volume and mass of the micron-sized platinum thin-film resistor, whose own heat capacity is almost negligible. This means that when the heating power changes, the platinum film can instantly heat up or cool down without any heat trailing phenomenon. Simultaneously, by directly sputtering or electroplating it onto the inner wall of the main channel 2, the tube wall thermal resistance and contact thermal resistance that must be overcome in traditional external heaters are completely eliminated, achieving zero-distance direct injection of heat into the fluid. This combination of extremely low thermal inertia and extremely high thermal conductivity enables the constant temperature difference closed-loop control system to dynamically adjust the heat input at a very high frequency and quickly establish a new thermal equilibrium.

[0106] In addition, the excellent corrosion resistance and long-term stability of platinum material itself, combined with the protection of a dense passivation layer, ensure that the heating element 31 can maintain a stable resistance value in complex and harsh fluid environments such as semiconductor precursors or fuel cell reaction gases, and will not age and drift due to long-term use. This provides a reliable hardware guarantee for the long-term high-precision inverse solution of the extended thermodynamic model.

[0107] To better illustrate the implementation methods and beneficial effects of this application, the following example uses Ar gas flow control for semiconductor coating processes. A controller 1 of a temperature and pressure control combination claimed in this invention is used, with a range of 0-500 sccm (standard milliliters per minute), and is connected to the Ar gas source and the reaction chamber.

[0108] Under standard pressure (101.325 kPa) and temperature (20°C), full-range calibration was performed using Ar gas, and the model parameters A, B, C, n, k, m were fitted and stored in the microprocessor controller 4.

[0109] The working process is as follows: First, the flow rate is set to Qset = 250 sccm. The microprocessor controller 4 controls the heating element 31 to start, quickly establishing a stable constant temperature difference ΔTset in less than 100 ms. Then, the pressure sensor 33 monitors the real-time pressure signal P of the flow channel (assuming that the pressure drops from 100 kPa to 95 kPa due to gas source fluctuations). The upstream and downstream temperature sensors sample at a frequency of 1 kHz to keep the temperature difference signal stable at a constant temperature difference ΔTset. The reference temperature sensor also samples at a frequency of 1 kHz to obtain the outer wall temperature. .

[0110] The microprocessor controller 4 calculates the current heating power in real time. The microprocessor controller 4 will control the current heating power. Real-time pressure signal P=95 kPa and Substituting into the extended thermodynamic model, the real-time mass flow rate is calculated in real time. .

[0111] Due to the pressure drop, there is an error of -2 sccm in the flow rate. The adaptive PID algorithm immediately adjusts the control voltage output to the piezoelectric valve based on this error and the rate of change of the error.

[0112] The piezoelectric valve is finely adjusted to increase the flow rate. The entire adjustment process is completed within 50ms, ultimately stabilizing the flow rate within the range of 250±0.5 sccm, far exceeding the ±1% accuracy of traditional mass flow controllers.

[0113] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A flow control method combining temperature and pressure control, characterized in that, The method includes the following steps: S1: Control the heating element (31) to heat the fluid flowing through the channel, and collect the temperature difference signal before and after the fluid is heated, the real-time pressure signal of the fluid in the channel, and the temperature of the outer wall of the channel. S2: Based on the temperature difference signal, the heating power of the heating element (31) is controlled through closed-loop feedback to maintain the temperature difference signal at a preset constant temperature difference; S3: Under constant temperature difference conditions, monitor the heating power of the heating element (31) in real time, substitute the heating power, the real-time pressure signal and the temperature of the outer wall of the channel into the preset extended thermodynamic model, and solve the real-time mass flow rate of the fluid. S4: Compare the real-time quality flow rate with the preset set flow rate, and calculate the flow rate error and the error change rate; S5: Input the flow error and the error change rate into the adaptive control algorithm to generate a control signal and drive the control valve (5) to adjust the fluid flow.

2. The flow control method of temperature control and pressure control combination according to claim 1, characterized in that, Step S1 is followed by: S11: Synchronously sample the temperature difference signal and the real-time pressure signal; S12: The Kalman filter algorithm is used to perform multimodal fusion processing on the synchronously sampled temperature difference signal and the real-time pressure signal to obtain the optimal estimated temperature difference signal and the optimal estimated real-time pressure signal after noise suppression. S13: The optimal estimated temperature difference signal is used as the temperature difference signal to facilitate subsequent control of the heating power of the heating element (31), and the optimal estimated real-time pressure signal is used as the real-time pressure signal to facilitate subsequent calculation in the preset thermodynamic model.

3. The flow control method of temperature control and pressure control combination according to claim 1, characterized in that, In step S3, the heating power is calculated by detecting the real-time voltage and real-time current of the heating element (31).

4. The flow control method of temperature control and pressure control combination according to claim 1, characterized in that, In step S3, the extended thermodynamic model is: ,in, The heating power, The nth power of the real-time quality flow rate For real-time pressure signals, For standard pressure, The outer wall temperature, The standard temperature is represented by A, B, n, k, m, and C, which are instrument coefficients. A represents the zero-flow power bias coefficient, B represents the mass flow rate to heating power ratio coefficient, C represents the system power offset constant, n represents the flow nonlinearity index, k represents the pressure compensation index, and m represents the temperature compensation index.

5. The flow control method of temperature control and pressure control combination according to claim 1, characterized in that, Step S5 includes: S51: Based on the flow error and the error change rate, determine the proportional coefficient of the adaptive control algorithm online using fuzzy inference rules. Integral coefficient and differential coefficients ; S52: Adjust the aforementioned proportional coefficient Integral coefficient and differential coefficients Substitute into adaptive control algorithm The control signal is obtained; among which, For control signals, The flow error is... The error change rate is given.

6. A controller for a temperature and pressure control combination, characterized in that, The controller (1) is used to implement the steps of the method according to any one of claims 1-5 above, and the controller (1) includes at least: a main channel (2), a control valve (5) disposed on the main channel (2), a flow sensing unit (3) and a microprocessor controller (4); The flow sensing unit (3) includes a heating element (31), a temperature sensor (32), and a pressure sensor (33); The microprocessor controller (4) is electrically connected to the heating element (31), temperature sensor (32), pressure sensor (33) and control valve (5); The microprocessor controller (4) is used to control the heating element (31) to heat the fluid flowing through the channel, and to collect the temperature difference signal before and after the fluid is heated, the real-time pressure signal of the fluid in the channel, and the temperature of the outer wall of the channel. Based on the temperature difference signal, the heating power of the heating element (31) is controlled through closed-loop feedback to maintain the temperature difference signal at a preset constant temperature difference; Under constant temperature difference conditions, the heating power of the heating element (31) is monitored in real time. The heating power, the real-time pressure signal and the temperature of the outer wall of the channel are substituted into the preset extended thermodynamic model and the real-time mass flow rate of the fluid is obtained by inverse solution. The real-time quality flow rate is compared with the preset set flow rate, and the flow rate error and error change rate are calculated. The flow error and the rate of change of the error are input into the adaptive control algorithm to generate a control signal and drive the control valve (5) to adjust the fluid flow.

7. The controller for a temperature and pressure control combination according to claim 6, characterized in that, The temperature sensor (32) is a high-frequency response MEMS thermopile structure based on graphene or boron nitride nanosheets, and the response time of the high-frequency response MEMS thermopile structure is less than 10ms.

8. The controller for a temperature and pressure control combination according to claim 7, characterized in that, The temperature sensor (32) includes at least an upstream temperature sensor, a downstream temperature sensor, and a reference temperature sensor; The upstream temperature sensor and the downstream temperature sensor are disposed inside the main channel (2) and located on both sides of the heating element (31), respectively for collecting the upstream temperature value and the downstream temperature value of the fluid, so as to calculate the temperature difference signal of the fluid before and after being heated; The reference temperature sensor is disposed outside the main channel (2) and is used to measure the temperature of the outer wall. The pressure sensor (33) is a MEMS pressure sensor, integrated inside the main channel (2), used to collect the real-time pressure signal of the fluid.

9. A controller for a temperature and pressure control combination according to claim 6, characterized in that, The control valve (5) is a piezoelectric ceramic driven valve.

10. A controller for a temperature and pressure control combination according to claim 6, characterized in that, The heating element (31) is a micron-sized platinum thin film resistor, which is sputtered or electroplated onto the inner wall of the main channel (2).

Images