An inflection point smooth transition method and system for autoclave temperature control profiles

Abstract

The present application belongs to the technical field of hot press tank temperature control in composite material forming process, and specifically discloses a kind of inflection point smooth transition method and system of hot press tank temperature control curve, comprising: superimposed test disturbance signal and collected response data before reaching temperature control inflection point;Local time constant and steady-state process gain are identified by constructing mathematical model;The parameter is converted into hard constraint boundary, and the dynamic time sequence transition trajectory sequence is generated by combining the slope before and after inflection point solution;Future desired temperature prediction value is generated using state predictor;Predictive control bias is calculated based on trajectory sequence and prediction value, and trajectory change rate is extracted;Accordingly, proportional integral adjustment component and feedforward compensation component are calculated, and are combined into execution control quantity to drive hot press tank.The present application eliminates temperature overshoot and response delay, and realizes inflection point smooth transition.

Description

Technical Field

[0001] This invention belongs to the technical field of autoclave temperature control in composite material molding processes, and relates to a method and system for smoothing the inflection point of the autoclave temperature control curve. Background Technology

[0002] In precision thermal processing such as composite material curing, autoclaves must strictly adhere to a preset temperature-time process curve. This curve typically consists of multiple linear segments with different heating or cooling rates, with temperature control inflection points formed at the junctions of adjacent segments. Current autoclave control systems face challenges in handling these inflection points. When the control target switches from one linear slope to another with a different slope, the system's actual temperature response often struggles to keep pace with this abrupt change. This is because the autoclave, as a complex thermodynamic system with thermal inertia, experiences continuous changes in its dynamic response characteristics due to various factors such as current temperature, workpiece load, and airflow conditions. Simply switching control commands according to the preset slope cannot instantaneously keep up with the target rate of change, leading to overshoot or undershoot in the temperature trajectory near the inflection point—that is, the actual temperature briefly exceeds or falls below the set value. This can not only affect product quality consistency but also trigger unnecessary control system oscillations.

[0003] Currently, the commonly used solutions in the industry mainly include direct switching based on fixed-parameter PID controllers or the use of pre-programmed fixed-shape transition trajectories. In the fixed-parameter PID control scheme, the control system directly switches the target slope from the old value to the new value at the inflection point, and the controller corrects the deviation based on fixed proportional, integral, and derivative gains. In the pre-programmed transition trajectory scheme, the system inserts a pre-designed, fixed-shape transition curve, such as an S-curve, before and after the inflection point to attempt to smooth the slope change. These methods assume that the dynamic characteristics of the system remain constant or change negligibly throughout the process.

[0004] However, these traditional methods have limitations. Control strategies based on fixed parameters typically tune the controller gain parameters according to the system's nominal operating conditions, failing to consider changes in system thermal inertia caused by factors such as workpiece material, loading method, and ambient temperature during actual operation. When the actual dynamic characteristics deviate from the preset parameters, the fixed controller or pre-programmed transition trajectory cannot make adaptive adjustments. As a result, under certain operating conditions, such as when the system's thermal inertia increases significantly, the pre-programmed transition trajectory may be too aggressive, leading to lag and error accumulation in the actual temperature response due to insufficient system capability; while under operating conditions where the system's thermal inertia decreases, the fixed control parameters may appear too conservative, resulting in prolonged transition times and affecting process efficiency. Therefore, existing technical solutions suffer from a mismatch between preset control parameters or trajectories and the system's real-time dynamic characteristics, which may lead to a decline in control quality at inflection points when production conditions change or different workpieces are processed.

[0005] Based on the above problems, the present invention aims to solve the problem that the temperature control at the inflection point of the process curve is overshoot, undershoot, or slow response due to the mismatch between the preset control parameters or fixed transition trajectory and the real-time dynamic characteristics of the autoclave. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art and to achieve the above objectives, the present invention proposes the following technical solution: a method for smoothing the inflection point of the temperature control curve of an autoclave, comprising: S1, extracting the temperature control inflection point of the preset autoclave process curve, superimposing a test disturbance signal into the heating output command before reaching the temperature control inflection point, collecting real-time temperatures in multiple spatial dimensions and normalizing and merging them to generate disturbance response data of multi-level multi-point temperatures.

[0007] S2. Analyze the disturbance response data of multi-level and multi-point temperature, construct a local first-order inertial mathematical model to identify dynamic characteristic parameters, and solve the local first-order inertial mathematical model to generate local thermodynamic dynamic characteristic parameters including local time constant and steady-state process gain.

[0008] S3. Call the local thermodynamic dynamic characteristic parameters as hard constraint boundaries, and combine them with the characteristic slopes of the preset autoclave process curve before and after the temperature control inflection point to generate the objective function of boundary constraint trajectory planning.

[0009] S4. Nonlinearly solve the objective function of boundary constraint trajectory planning to generate a dynamic temporal transition trajectory sequence that fully covers the temperature control inflection point over a time span.

[0010] S5. Inject local thermodynamic dynamic characteristic parameters into the internal state predictor to perform single-step time-forward state calculation and generate the predicted value of the future expected temperature.

[0011] S6. Perform a subtraction algebraic operation between the dynamic time-series transition trajectory sequence and the predicted future temperature to generate the predictive control deviation, and perform a derivative operation on the dynamic time-series transition trajectory sequence to extract the order rate of change.

[0012] S7. Calculate the proportional-integral adjustment component based on the predictive control deviation, calculate the feedforward compensation component based on the order rate of change, and combine the proportional-integral adjustment component and the feedforward compensation component into a smooth transition execution control output to drive the actuator of the autoclave.

[0013] The second aspect of the present invention provides a smooth transition system for the inflection point of the temperature control curve of an autoclave, comprising: a disturbance testing and data acquisition module, which extracts the temperature control inflection point of a preset autoclave process curve, superimposes a test disturbance signal into the heating output command before reaching the temperature control inflection point, and acquires real-time temperatures in multiple spatial dimensions and normalizes and merges them to generate disturbance response data of multi-level multi-point temperatures.

[0014] The dynamic characteristic parameter identification module analyzes the disturbance response data of multi-level and multi-point temperatures, constructs a local first-order inertial mathematical model to identify dynamic characteristic parameters, and solves the local first-order inertial mathematical model to generate local thermodynamic dynamic characteristic parameters including local time constants and steady-state process gains.

[0015] The trajectory optimization objective function generation module calls local thermodynamic dynamic characteristic parameters as hard constraint boundaries and combines them with the characteristic slopes of the preset autoclave process curve before and after the temperature control inflection point to generate the boundary constraint trajectory planning objective function.

[0016] The transition trajectory generation module nonlinearly solves the objective function for boundary constraint trajectory planning, generating a dynamic temporal transition trajectory sequence that fully covers the temperature control inflection point over a time span.

[0017] The temperature prediction module injects local thermodynamic dynamic characteristic parameters into the internal state predictor to perform single-step time-series forward state calculations and generate the expected future temperature prediction value.

[0018] The deviation and rate of change calculation module performs a subtraction algebraic operation between the dynamic time-series transition trajectory sequence and the predicted future temperature to generate the predictive control deviation, and performs derivative operation on the dynamic time-series transition trajectory sequence to extract the order rate of change.

[0019] The control quantity synthesis and output module calculates the proportional-integral adjustment component based on the predicted control deviation and the feedforward compensation component based on the order rate of change. It then synthesizes the proportional-integral adjustment component and the feedforward compensation component into a smooth transition execution control quantity output to drive the actuator of the autoclave.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) By actively injecting a small test disturbance and collecting response data before the temperature control inflection point is reached, the present invention can obtain the dynamic temperature response information of the autoclave at the current specific working point online. This mechanism allows the system to evaluate its thermal inertia and gain based on real-time data rather than fixed assumptions, providing accurate physical boundary basis for subsequent trajectory planning. As a result, the generated transition trajectory can more closely match the instantaneous response capability of the system, reducing the risk of trajectory unachievable or actuator saturation caused by model mismatch.

[0021] (2) This invention utilizes the locally identified thermodynamic dynamic characteristic parameters online, reverse-engineers them into the allowable temperature acceleration constraint threshold of the system, and uses this value as the hard constraint boundary of the trajectory optimization problem. This mechanism transforms the abstract identified parameters into specific physical constraints and embeds them into the trajectory generation algorithm. This ensures that the planned adaptive transition trajectory, while meeting the start and end slope requirements, has its rate of change limited by the actual physical capabilities of the system throughout the entire process. Theoretically, this eliminates the possibility of actual temperature overshoot caused by an overly aggressive trajectory, achieving a smooth transition without overshoot.

[0022] (3) This invention constructs an internal state predictor and updates the model of the predictor with the dynamic parameters identified online in real time, while combining the actual temperature at the previous moment for single-step forward prediction. This mechanism can estimate the system's response to the current command in the next control cycle, that is, provide a future expected temperature prediction value. By comparing the target value of the planned trajectory with this prediction value, the leading control deviation can be calculated. This deviation calculation based on model prediction enables the control system to intervene in correction before the error actually appears in the feedback signal, improving the system's ability to resist model errors and external disturbances, and reducing the final following error.

[0023] (4) When generating control commands, this invention not only uses a proportional-integral feedback adjustment component based on prediction deviation to correct low-frequency errors, but also simultaneously calculates the order rate of change of the planned trajectory itself, and scales it using the system gain identified online to generate a feedforward compensation component. This mechanism ensures that the control commands contain an active driving component that directly matches the dynamic rate of change of the target trajectory. By adding the feedback component and the feedforward component in parallel, the synthesized control quantity can more directly drive the system to follow the dynamic changes of the trajectory, compensate for the delay caused by system inertia, and enhance the tracking accuracy of the set trajectory during the inflection point transition. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the implementation steps of the method of the present invention.

[0026] Figure 2 This is a schematic diagram of the system module connections of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Example 1 Please see Figure 1 As shown, the present invention proposes a method for smoothing the inflection point of the temperature control curve of an autoclave, comprising: S1, extracting the temperature control inflection point of the preset autoclave process curve, superimposing a test disturbance signal into the heating output command before reaching the temperature control inflection point, collecting real-time temperatures in multiple spatial dimensions and normalizing and merging them to generate disturbance response data of multi-level multi-point temperatures.

[0029] In a preferred embodiment, step S1 includes: extracting the intersection timestamps of two adjacent temperature control line segments with different slopes in the preset autoclave process curve as the temperature control inflection point, and setting a reserved pre-time threshold earlier than the temperature control inflection point. When the actual processing time is detected to reach the reserved pre-time threshold, the base power output value of the current working point is kept constant, and a set of test disturbance signals that meet the preset target spectrum and amplitude constraints are superimposed on the heating actuator of the autoclave. Simultaneously collect real-time temperatures in multiple spatial dimensions within the workpiece area and airflow circulation area inside the autoclave. Then, align the real-time temperatures in multiple spatial dimensions with the test disturbance signals based on timestamps, normalize and merge them to generate multi-level, multi-point temperature disturbance response data.

[0030] This invention binds the injection time of the test disturbance signal to a reserved pre-time threshold and performs timestamp alignment during real-time temperature acquisition in multiple spatial dimensions, thereby eliminating transmission delay errors caused by the spatial distribution of sensors and improving the temporal fidelity of disturbance response data for multi-level and multi-point temperatures.

[0031] In a further preferred embodiment, a set of test disturbance signals that meet preset target spectrum and amplitude constraints are superimposed onto the heating actuator of the autoclave, including: activating the internal signal generator module to generate a predefined pseudo-random binary sequence; The minimum pulse width of the pseudo-random binary sequence is set according to the nominal dominant time constant of the autoclave system to determine the target spectrum; The amplitude constraint of the pseudo-random binary sequence is set to a preset percentage range of the maximum heating power of the autoclave to generate a test disturbance signal.

[0032] This invention improves the signal-to-noise ratio of local thermodynamic dynamic characteristic parameters by matching the minimum pulse width of the pseudo-random binary sequence to the nominal dominant time constant and limiting the amplitude constraint, ensuring that the energy of the test disturbance signal fully covers the main frequency band of the system's dynamic response without disrupting the thermodynamic balance of the actual curing process of the workpiece.

[0033] Specifically, the first implementation step S1 of this method is executed by a process control processor deployed within the autoclave's main control system, aiming to acquire input and output data for subsequent dynamic characteristic identification. First, the process control processor loads and parses a preset digital autoclave process curve, which is stored as a structured data array containing multiple sequentially executed process segments. Each process segment is defined by four core parameters: start time, end time, start temperature, and target temperature. The processor sequentially traverses this data array, calculating and comparing the temperature set slopes of two adjacent process segments. When the first... Section and the When the slopes of the segments are inconsistent, the processor will... The end timestamp of the segment is identified as the temperature control inflection point. At the same time, the processor reads a preset reserved pre-time threshold and subtracts the threshold from the timestamp of the temperature control inflection point to calculate the start trigger time of the disturbance injection.

[0034] Subsequently, the process control processor continuously monitors the system's real-time clock. When the real-time clock reaches the initial trigger point of the disturbance injection, the system immediately performs the following parallel operations. On one hand, the processor locks the heating power output command of the current main feedback control loop and caches it as the base power output value to keep it constant. Next, the signal generator module inside the processor is activated to generate a predefined pseudo-random binary sequence as a test disturbance signal. The amplitude and minimum pulse width of this sequence meet the preset amplitude constraints and target spectrum requirements. The processor algebraically sums the real-time generated test disturbance signal with the cached base power output value to synthesize the final heating output command, and sends it to the heating actuator of the autoclave, such as a silicon controlled rectifier (SCR) power regulator, through a digital-to-analog conversion interface. On the other hand, throughout the entire period of disturbance injection, the multi-channel synchronous data acquisition module concurrently reads and records the multi-dimensional real-time temperature measured by multiple thermocouple sensors deployed in the workpiece area and airflow circulation area inside the autoclave at a fixed high-frequency sampling rate. Each temperature sample value is strongly correlated with a high-precision timestamp.

[0035] Upon arrival of the time stamp at the temperature control inflection point, i.e., after the perturbation injection phase ends, the processor normalizes the acquired raw data. Using the time series of the test perturbation signal as a benchmark, the processor aligns the time series of multi-dimensional real-time temperatures acquired from different physical locations using interpolation or resampling. Next, for each temperature measurement point's time series, the processor subtracts its initial temperature value at the start of the perturbation injection to normalize the data, thereby extracting the temperature fluctuation caused by the perturbation input. Finally, the processor merges the aligned and normalized input perturbation sequence and the multiple temperature response sequences into a structured dataset, which is the perturbation response data for multi-level, multi-point temperatures, and stores it in a shared memory area for subsequent steps.

[0036] It should be noted that the preset autoclave process curve refers to a series of temperature and time correspondences pre-arranged for the curing process of specific composite material parts. In the time domain, it is composed of multiple straight line segments with different heating rates, holding rates, or cooling rates.

[0037] The temperature control inflection point is the connection point of two adjacent straight line segments with different slopes on the preset process curve, representing the moment when the control target undergoes a discontinuous change.

[0038] The pre-set time threshold is a time length empirically set based on the thermal inertia of the autoclave system, with a typical value range of [value missing]. to The basis for its setting is to ensure that there is a sufficient time window to stimulate the system and collect complete dynamic response data without affecting the normal start of the next process segment.

[0039] The test disturbance signal is an excitation signal used for system identification. In this embodiment, a pseudo-random binary sequence is used, and its amplitude constraint is typically set to the maximum heating power of the autoclave. to This setting assumes an application environment with stringent product quality requirements, aiming to ensure that the injected energy fluctuations are sufficiently small to avoid irreversible impacts on the actual curing process of the workpiece. Its target spectrum is determined by the minimum pulse width of the sequence, which is set according to the nominal dominant time constant of the autoclave system to ensure that the energy of the excitation signal covers the main frequency band of the system's dynamic response.

[0040] Multi-dimensional real-time temperature refers to temperature data collected from at least two different physical regions, such as sensor data directly attached to the surface of the workpiece and sensor data located in the main air duct of the hot air circulation, in order to capture the heat transfer characteristics of different parts of the system.

[0041] The perturbation response data of multi-level and multi-point temperature is a time series matrix, whose column vectors contain the applied perturbation power signal and the normalized temperature change caused by the perturbation at multiple different measurement points.

[0042] For example, suppose a pre-defined autoclave process curve contains two consecutive heating stages, the first stage starting from... time linear heating to time Its slope is The second paragraph begins time linear heating to time Its slope is The system first determines the temperature control inflection point as... Set the reserved lead time threshold to... Therefore, the perturbation test will be conducted. Starts at that time. At that time, the system detected the current base power output value as The system then maintained this. The output remains unchanged, and a value with magnitude is added. The minimum pulse width is A pseudo-random binary sequence is used as the test perturbation signal. For example, in to During this period, the output signal is The total heating command is: ;exist to During this period, the output signal is switched to The total heating command is: At the same time, the system collects temperature data from sensors located in the workpiece area. Temperature sensor in airflow circulation area The data. Assuming in At that time, it was measured The initial temperature is , The initial temperature is In subsequent disturbances, such as during At that time, it was measured for , for The system processes data after the disturbance ends. At this point in time, the corresponding normalized temperature biases are as follows: , Ultimately, the system will... The disturbance signal values ​​of all sampling points during the disturbance period Value and The values ​​are compiled into a time series data table, forming multi-level, multi-point temperature disturbance response data for use in the next step.

[0043] S2. Analyze the disturbance response data of multi-level and multi-point temperatures, construct a local first-order inertial mathematical model to identify dynamic characteristic parameters, and solve the local first-order inertial mathematical model to generate local thermodynamic dynamic characteristic parameters including local time constants and steady-state process gains.

[0044] In a preferred embodiment, step S2 includes: extracting the time-series scatter set of temperature bias values ​​generated by superimposed test disturbance signals from the disturbance response data of multi-level multi-point temperatures; Using the test disturbance signal as the input variable and the time-series scatter set of temperature bias as the output variable, a first-order differential equation is constructed to map the dynamic relationship between heating power change and temperature response, serving as a local first-order inertial mathematical model. Based on the local first-order inertial mathematical model, dynamic characteristic parameters are identified, and the local time constant characterizing the transient thermal inertia of the autoclave and the steady-state process gain characterizing the input-output ratio are calculated and extracted, and then combined and packaged into local thermodynamic dynamic characteristic parameter output.

[0045] This invention combines the time-series scatter set of temperature bias with the basic power output value to fit a local first-order inertial mathematical model, separating the local time constant and steady-state process gain. This avoids the calculation divergence problem of global nonlinear modeling and improves the online identification and convergence speed of local thermodynamic dynamic characteristic parameters.

[0046] Specifically, after completing step S1, the process control processor then executes step S2 to extract the dynamic characteristics of the system from the acquired data. The processor retrieves multi-level, multi-point temperature disturbance response data from the shared memory region. This data structure contains the time series of the test disturbance signal as the system input, and the multi-channel normalized temperature response time series as the system output. The processor first separates the input signal sequence and the output temperature response sequence after averaging one or more specified channels from it. These two together constitute a time-series scatter plot of the temperature bias, which is a two-dimensional dataset that records the difference between the input disturbance power value and the corresponding output temperature deviation from the initial value at each sampling time during the disturbance test, providing a direct data basis for subsequent parameter identification.

[0047] Next, the processor constructs a mathematical model to describe the local dynamic response of the system. In this embodiment, this model is determined to be a local first-order inertial mathematical model. This model characterizes the dynamic relationship between small changes in heating power and the temperature response of the workpiece region. The processor invokes an online parameter identification algorithm, such as recursive least squares, to fit the parameters of the local first-order inertial mathematical model. This algorithm uses the time series of the test disturbance signal as the input variable and the temperature change sequence in the scatter plot of the temperature bias as the output variable. It iteratively calculates and minimizes the sum of squared errors between the model's predicted output and the actual measured output, thereby approximating and determining the unknown parameters in the model.

[0048] The local first-order inertial mathematical model can be represented in the continuous time domain by the following first-order differential equation: The goal of parameter identification algorithms is to find an optimal set of parameters. and This minimizes the error between the temperature bias predicted by the model and the actual temperature bias measured. This is typically achieved by minimizing the cost function. To achieve: in, This represents the local time constant, and its unit is seconds. This parameter reflects the time required for the system to reach a new steady state. The corresponding time length, the larger the value, the greater the thermal inertia of the system and the slower the response, which is obtained by the identification algorithm based on the input and output data fitting. Represents the steady-state process gain, expressed in degrees Celsius per percentage of power. This parameter reflects the ratio between the system input and output in steady state. The larger the value, the more sensitive the system is to the power input. It is also obtained by fitting the identification algorithm. Represents consecutive time points The predicted temperature bias calculated by the mathematical model, in degrees Celsius (°C). Represents a point in time The amplitude of the test disturbance signal corresponds to the input sequence in the multi-level, multi-point temperature disturbance response data, expressed as a power percentage. ). At discrete sampling time points The actual measured and normalized temperature bias. At discrete sampling time points Given parameters and The predicted temperature bias is calculated using a mathematical model. It is the total number of data points in the time series scatter plot of the temperature bias.

[0049] After the identification algorithm converges or reaches the preset computation cycle, the processor calculates two key physical parameters from the fitted local first-order inertial mathematical model. The first is the local time constant, which quantitatively describes the response delay and inertial characteristics of the autoclave system to heat input at the current operating point. The second is the steady-state process gain, which represents the change in temperature output caused by a unit change in heating power input under steady-state conditions. The processor integrates these two calculated values, namely the local time constant and the steady-state process gain, into a data structure, packages them into local thermodynamic dynamic characteristic parameters, and stores this parameter set for use in subsequent trajectory optimization steps.

[0050] For example, following the example from the previous step, the system now begins processing the generated multi-level, multi-point temperature disturbance response data. This data includes data from... arrive During this period, a sequence of input disturbance signals is recorded at a specific sampling period, i.e., a sequence of signals within a given period. and The pseudo-random sequence that varies between these values, and the corresponding temperature offset in the workpiece area. Temperature offset of sequence and airflow zone Sequence. This example uses the workpiece area. The sequence is used as the output for identification. Subsequently, the system calls a recursive least squares fitting algorithm to combine the input perturbation sequence with... The sequence is used as input to perform online parameter identification of a local first-order inertial mathematical model. This is achieved through... Iterative calculations are performed within a data window, and the fitting process aims to find a set of... and Values ​​to minimize the temperature bias predicted by the model and the actual measured values. The root mean square error between the values. After calculation, the system obtains the model parameters at this operating point, such as the local time constant. The identification result is steady-state process gain The identification result is Ultimately, the system packages these two parameters into local thermodynamic dynamic characteristic parameters, that is, a set containing... The data object is stored in a specified memory address for later use in subsequent steps.

[0051] S3. Call the local thermodynamic dynamic characteristic parameters as hard constraint boundaries, and combine them with the characteristic slopes of the preset autoclave process curve before and after the temperature control inflection point to generate the objective function of boundary constraint trajectory planning.

[0052] In a preferred embodiment, step S3 includes: inversely calculating the local time constant contained in the local thermodynamic dynamic characteristic parameters into the temperature acceleration constraint threshold of the physical system's permissible response. Extract the first set temperature change slope of the line segment before the temperature control inflection point in the preset autoclave process curve, and the second set temperature change slope of the line segment after the temperature control inflection point. The goal is to minimize the transition time from the first set temperature change slope to the second set temperature change slope and ensure the trajectory converges monotonically, using this as the optimization benchmark for the closed-loop cost function. A temperature acceleration constraint threshold is introduced as a hard constraint boundary to generate the objective function for boundary constraint trajectory planning.

[0053] This invention transforms the local time constant into a temperature acceleration constraint threshold and uses it as a hard constraint boundary. It combines a first set temperature change slope and a second set temperature change slope to construct a closed-loop cost function optimization benchmark. Under the premise of ensuring that the actual temperature follow-up overshoot is always zero, the transition time is compressed, thereby achieving the synergy between physical boundary constraints and time-optimal objectives.

[0054] Specifically, after online identification is completed in step S2, the process control processor immediately proceeds to step S3, which aims to construct a mathematical programming problem for trajectory optimization. The processor first reads the local thermodynamic dynamic characteristic parameters generated in step S2 from its internal memory and extracts the local time constant. Subsequently, the processor performs a reverse operation, converting the local time constant into a constraint index representing the system's physical response capability—namely, the temperature acceleration constraint threshold—through a preset empirical or physical model. This constraint threshold is a dynamic boundary calculated based on the system's current thermodynamic characteristics; it reflects the rate of change of the maximum temperature change that the autoclave heating or cooling system can achieve without causing actuator saturation or drastic temperature fluctuations.

[0055] In parallel, the process control processor accesses the structured data of the preset autoclave process curve stored in memory again. The processor locates the temperature control inflection point determined in step S1 and extracts the parameters of the two linear process segments immediately before and after the inflection point. By calculating the temperature change per unit time, the processor obtains the first preset temperature change slope of the segment before the temperature control inflection point and the second preset temperature change slope of the segment after the temperature control inflection point.

[0056] Finally, the processor integrates the calculated and extracted information to generate a boundary-constrained trajectory planning objective function in a programmed manner. This objective function is a formalized mathematical description that encapsulates the engineering objective (shortest time, no overshoot) and physical constraints (system response capability) into a standard optimization problem, laying the foundation for subsequent numerical solutions. No overshoot means that the actual temperature response value does not exceed a set value. The optimization objective is set to minimize the total time required to transition from the initial slope to the final slope. The constraints of this problem include: first, the initial tangent slope of the trajectory must be equal to the first set temperature change slope; second, the final tangent slope of the trajectory must be equal to the second set temperature change slope; third, during the entire transition period, the absolute value of the second derivative of the trajectory, i.e., the acceleration, must not exceed the temperature acceleration constraint threshold. The optimization benchmark for the closed-loop cost function is set by minimizing the time consumption for a smooth transition from the first set temperature change slope to the second set temperature change slope while ensuring that the actual temperature overshoot is always zero. The temperature acceleration constraint threshold is introduced as a hard constraint boundary, ultimately forming a complete boundary-constrained trajectory planning objective function that can be processed by the numerical solver, and then passed to the next step.

[0057] The objective function of boundary-constrained trajectory planning is a time-optimal control problem, which can be expressed as finding a temperature trajectory. This makes the transition time Minimize: , Let represent the control variable (acceleration). This optimization problem is subject to the following constraints: Among them, the temperature acceleration constraint threshold From the local time constant The calculation shows that: in, This is the total transition time to be optimized, in seconds. ). It is the transition temperature trajectory function to be planned, which describes the temperature change path from the moment before the inflection point to the moment after the inflection point, and is generated by the solver. and These represent the first derivatives of the transition trajectory at the start and end times, respectively, i.e., the instantaneous slope, in degrees Celsius per second. ). Represents the first set temperature change slope, in units of . This represents the second set temperature change slope, in units of . Represents the transition trajectory at a certain point in time. The second derivative, i.e., the instantaneous temperature acceleration, is expressed in degrees Celsius per second squared. ). It is the temperature acceleration constraint threshold, calculated from the system identification parameters, and serves as a hard constraint boundary, with units of . . It is a local time constant. The system characteristic rate coefficient is pre-calibrated based on the ratio of the heat exchanger's maximum power to the tank's total heat capacity. It is used to establish the relationship between the time constant and the maximum acceleration, and its unit is degrees Celsius per second. For example, the value is set to 0.1 ( ).

[0058] For example, continuing from the previous step, the system has obtained local thermodynamic dynamic characteristic parameters, including the local time constant. for The system first calculates the temperature acceleration constraint threshold. It assumes that the system characteristic rate coefficient is determined based on the equipment's factory calibration and historical operating data. for Therefore, the processor calculates... Next, the system extracts the slope before and after the inflection point from the preset autoclave process curve to obtain the first preset temperature change slope. And the second set temperature change slope Finally, the system substitutes these specific values ​​to generate a boundary constraint trajectory planning objective function to be solved. This objective function is specifically expressed as: finding a temperature trajectory. So that it satisfies the initial slope Termination slope And acceleration throughout the entire process No more than Under these conditions, the total time used The minimum is reached. This complete optimization problem with specific numerical boundaries is then constructed and passed to the next nonlinear programming solver.

[0059] S4. Nonlinearly solve the objective function of boundary constraint trajectory planning to generate a dynamic temporal transition trajectory sequence that fully covers the temperature control inflection point over a time span.

[0060] In a preferred embodiment, step S4 includes: running a nonlinear programming algorithm within the current prediction triggering period to perform a finite-time-space iterative solution for the objective function of the boundary constraint trajectory planning; Obtain the time series curve of the planning acceleration that follows the hard constraint boundary, and perform a quadratic cumulative integral calculation in the discrete time domain; The discrete numerical results calculated by the second cumulative integral are shifted and connected based on the current measured initial physical temperature to generate a dynamic time-series transition trajectory sequence in which the initial tangent slope matches the first set temperature change slope and the final tangent slope matches the second set temperature change slope. Among them, the second-order cumulative integral calculation in the discrete time domain includes: taking the first set temperature change slope as the initial value of the integral, multiplying the acceleration value of the planned acceleration time series curve of the current step by the time step and accumulating it to the slope value of the previous step at each discrete control time step to generate the temperature change slope trajectory. Using the initial zero value as the initial value for integration, at each discrete control time step, the average slope value of the temperature change slope trajectory of the current step is multiplied by the time step and accumulated to the relative temperature value of the previous step to generate the relative temperature transition trajectory.

[0061] This invention seamlessly transforms the control strategy in the acceleration domain into a dynamic temporal transition trajectory sequence in the temperature domain by performing a double-integral calculation on the planned acceleration time-series curve and shifting it to the current measured initial physical temperature value, thus eliminating trajectory drift caused by deviations in the initial integral value. By sequentially reducing the dimension of the planned acceleration time-series curve to a temperature-varying slope trajectory and a relative temperature transition trajectory, the discrete acceleration command output by the nonlinear programming algorithm is smoothly transformed into a continuous temperature setting benchmark, eliminating the mechanical impact of the underlying hardware execution level nodes on the step acceleration signal.

[0062] Specifically, step S4 is the core computational stage of the entire smooth transition method, executed immediately after the process control processor receives the objective function generated in step S3. The process control processor first calls a built-in nonlinear programming algorithm library, such as a solver based on sequential quadratic programming or the interior-point method, within the current prediction trigger cycle. A nonlinear programming algorithm is a mathematical optimization algorithm capable of handling objective functions or constraints containing nonlinear terms. In this scheme, its role is to find an acceleration control strategy that minimizes the transition time while satisfying all physical and technological constraints. The current prediction trigger cycle is a maximum allowable computation time set for the solution algorithm, for example... This ensures the real-time generation of control trajectories in dynamically changing industrial environments. The processor passes the objective function for boundary-constrained trajectory planning constructed in step S3, along with its complete constraints, including the initial slope, termination slope, and maximum acceleration limit, as input parameters to the solver. The solver performs high-speed iterative calculations within a finite time domain, aiming to minimize the transition time, and calculates an acceleration control strategy whose second derivative strictly satisfies the hard constraint boundaries at any given time. The time-series data set of this strategy constitutes the planned acceleration time-series curve and is stored in a temporary cache.

[0063] Next, the process control processor performs a second-order cumulative integration in the discrete time domain on the planned acceleration time-series curve output by the solver. This second-order cumulative integration is a standard numerical integration method used to reconstruct velocity and displacement signals from acceleration signals; in this scheme, it reconstructs the temperature gradient slope and temperature trajectory from temperature acceleration. First, the processor performs a first-order cumulative integration, converting the acceleration sequence into a temperature gradient slope sequence. This process uses a first set temperature gradient slope as the initial value for integration. At each discrete control time step, the acceleration value of the current step is multiplied by the time step, and then accumulated to the slope value of the previous step, thereby generating a dynamically changing temperature gradient slope trajectory. Subsequently, the processor performs a second-order cumulative integration, converting the temperature gradient slope sequence into a relative temperature trajectory sequence. This process uses relative zero degrees as the initial value for integration. At each time step, the average slope value of the current step is multiplied by the time step, and then accumulated to the relative temperature value of the previous step, resulting in a relative temperature transition trajectory with optimal shape, starting from zero.

[0064] Finally, to apply this relative trajectory to the actual physical system, the processor samples the initial value of the current measured physical temperature inside the autoclave at the moment of planning initiation. This initial measured physical temperature serves as the baseline for trajectory generation, ensuring a smooth and seamless connection between the planned trajectory and the actual state. The processor uses this measured temperature value as a DC bias and performs algebraic addition with each data point in the previously calculated relative temperature trajectory sequence. Through this translation and continuation operation, a final trajectory is generated where the starting point perfectly matches the current actual temperature, and the initial tangent slope equivalently corresponds to the first set temperature change slope, and the ending tangent slope equivalently corresponds to the second set temperature change slope. This discrete numerical sequence, containing a series of high-density timestamps and corresponding temperature setpoints for the future, is the dynamic temporal transition trajectory sequence. It is loaded into a first-in-first-out queue, ready to be read and executed point by point by the lower-level control loop.

[0065] The second-order cumulative integral is calculated in the discrete-time domain using the following iterative formula. Assume the control period is... In the Each sampling time: First, calculate the instantaneous temperature change slope. : Secondly, calculate the relative temperature. : Finally, points in the dynamic temporal transition trajectory sequence are generated. : in, It is in the At each sampling time point, the set acceleration value is given by the planned acceleration time series curve, in units of... . and They are respectively in the 1st The sampling time and the first sampling time The instantaneous temperature change slope calculated at -1 sampling time point, with its initial value Set as the first set temperature change slope The unit is . It is the discrete sampling time step of the control system, for example . and They are respectively in the 1st The sampling time and the first sampling time The relative temperature value calculated at -1 sampling time point, its initial value Set as . It is at the instant the transition trajectory begins execution ( The initial value of the currently measured physical temperature (at what time) is collected, in units of . It is in the The set temperature value of the dynamic temporal transition trajectory sequence finally generated at each sampling time, in units of... Its value is within the absolute process temperature range, for example, the value is set to... .

[0066] For example, following the example from the previous step, the system has already constructed the objective function, where , ,as well as The nonlinear programming solver receives this task and begins solving. The solver determines which slope to change from... Upgraded to The fastest way is to maintain maximum acceleration continuously. To accelerate, the required time is Therefore, the time-series curve of the planned acceleration output by the solver is: at time... to Inside, ,after .

[0067] Next, the system will use The step size is used to perform a double-integration. At the instant the trajectory starts, the system measures the initial value of the current measured physical temperature. The integral calculation process is as follows: exist Initial slope initial relative temperature Final temperature .

[0068] exist : . . .

[0069] exist : . . .

[0070] exist : . . .

[0071] exist After that, the acceleration is The slope remains Through this process, the system generates a dynamic temporal transition trajectory sequence, i.e. After a series of set points, the sequence is sent to the execution queue.

[0072] S5. Inject local thermodynamic dynamic characteristic parameters into the internal state predictor to perform single-step time-series forward state calculation and generate the predicted value of the future expected temperature.

[0073] In a preferred embodiment, step S5 includes: extracting the local time constant and steady-state process gain contained in the local thermodynamic dynamic characteristic parameters, and updating them to the state equation coefficient variable array of the internal state predictor; The average feedback temperature of the actual space inside the autoclave in the previous sampling control cycle is retrieved and used as the recursive initial state reference for the internal state predictor to perform state recursion. Based on the state equation coefficient variable array, the recursive initial state reference, and the current control command, the internal state predictor is triggered to perform a single-step timing forward state calculation to generate the expected future temperature prediction value for the next sampling control cycle.

[0074] This invention overwrites the local time constant and steady-state process gain into the state equation coefficient variable array without delay, and uses the average feedback temperature of the actual space inside the autoclave in the previous sampling control cycle as the recursive initial state reference, enabling the internal state predictor to track the dynamic changes of the autoclave in real time and improve the advance observation accuracy of the future expected temperature prediction value.

[0075] Specifically, step S5 is executed synchronously with other computational tasks in the control loop, aiming to provide a high-rate future temperature prediction. The process control processor first reads the local thermodynamic dynamic characteristic parameters, identified online and updated in real time by step S2, from its high-speed working memory. The processor immediately parses the local time constant and steady-state process gain from these parameters. Next, the processor substitutes these two parameters into a preset discretization transformation formula to calculate the state equation coefficients required by the internal state predictor. The calculated coefficients are immediately written to a specific memory address where the state equation coefficient variable array of the internal state predictor resides; this overwrite operation has no delay, ensuring that the prediction model always reflects the latest system dynamics.

[0076] Simultaneously, the process control processor retrieves the average feedback temperature of the actual internal space of the autoclave from the circular buffer of the data acquisition interface for the previous sampling control cycle. This temperature value represents the latest known state of the system at the start of the current predictive calculation. The processor loads this temperature value into the internal state predictor as the initial state reference for its recursive calculation of the state model.

[0077] Finally, the processor-driven internal state predictor, whose parameters have been synchronously overwritten, performs a single-step timing-forward state calculation. This calculation not only uses the aforementioned initial state reference but also requires the control command value issued in the previous control cycle as input. The predictor outputs a scalar value by performing an iterative calculation of its internal discrete state equations. This value is the predicted future expected temperature of the system at the end of the next absolute time sampling cycle. This predicted value is the final output of this step, providing the control system with forward-looking information about the upcoming state changes of the controlled object. It is immediately cached for deviation calculation in step S6.

[0078] The internal state predictor is a discretized implementation based on the first-order inertial element model identified in step S2. Its single-step prediction state equation is as follows: Among them, the coefficients in the coefficient variable array of the state equation and Calculated from local thermodynamic dynamic characteristic parameters: in, It is the predicted future temperature value calculated in the current cycle, that is, in The predicted temperature at any given time, in degrees Celsius. ). It is the average feedback temperature of the actual space inside the autoclave in the previous sampling control cycle, that is, in The measured temperature at any given time, in units of . It is the previous control cycle. Control command values ​​are constantly sent to the actuator, in units of power percentage ( ). The baseline power representing the current operating point is obtained based on the steady-state actual output power locked just before the disturbance test (i.e., before reaching the temperature control inflection point). Its unit is power percentage (%), and its value is within the physical execution range of 0-100%, for example, the value is set to 60%. The baseline temperature representing the current operating point is obtained based on the steady-state actual temperature locked just before the disturbance test, and its unit is... Its value is within the process set temperature range, for example, the value is set to... . The dimensionless discretization coefficients in the array of coefficient variables of the state equation are obtained by calculating the local time constant and sampling period of the system in the continuous domain. They are used to characterize the natural decay or retention characteristics of the system based on its own thermal inertia. Their values ​​are in the range of (0,1), for example, the value is set to 0.995. The dimensional discretized coefficients in the variable array of the state equations are obtained by calculating the steady-state process gain, local time constant, and sampling period of the system in the continuous domain. They are used to characterize the temperature increment response of the system under incremental control input, and their units are... / %, for example, the value is set to 0.0125. / %. It is a local time constant. It is the steady-state process gain. It is the sampling period of the control system, in seconds. For example, the value is set to 1s. e is the natural constant.

[0079] It should be noted that the internal state predictor is a software module that runs in the process control processor. Its core function is to calculate the system state at the next moment based on the known system model, the actual state at the previous moment, and the control input.

[0080] The state equation coefficient variable array is a memory area used to store model parameters. A delay-free overwrite mechanism ensures real-time synchronization between the prediction model and the actual dynamics of the system.

[0081] The average feedback temperature of the actual space inside the autoclave in the previous sampling control cycle is the starting point for the prediction iteration and a benchmark to ensure that the prediction results are closely related to the actual process.

[0082] For example, following the previous steps, the system continuously executes a control loop, with a sampling period assumed to be... The local thermodynamic dynamic characteristic parameters obtained by the system from S2 are as follows: Simultaneously, the system retrieves the baseline data for the current operating point: the baseline temperature. Basic power reference First, the processor calculates the coefficients of the state equation based on this parameter. The calculation process is as follows: ,as well as These two coefficients are immediately overwritten into the state equation coefficient variable array of the internal state predictor. Next, the system retrieves feedback data to obtain the average feedback temperature of the actual space inside the autoclave from the previous sampling control cycle. At the same time, the system recorded the control commands issued in the previous cycle as follows: Finally, the processor-driven predictor performs a single-step timing advance state calculation, the calculation process of which is as follows: The calculation result is Therefore, the system generates a future expected temperature prediction value for the next sampling period. This data is then passed to the next step to calculate the control deviation.

[0083] S6. Perform a subtraction algebraic operation between the dynamic time-series transition trajectory sequence and the predicted future temperature to generate the predictive control deviation, and perform a derivative operation on the dynamic time-series transition trajectory sequence to extract the order rate of change.

[0084] In a preferred embodiment, step S6 includes: acquiring the current single-step target temperature setpoint in the dynamic time-series transition trajectory sequence within each sampling control cycle; The current single-step target temperature setpoint is subtracted from the future expected temperature prediction value to generate the predictive control deviation. Perform a first-order finite difference operation on the dynamic time-series transition trajectory sequence to extract the order rate of change.

[0085] This invention generates a predictive control deviation by performing algebraic subtraction between the current single-step target temperature setpoint and the predicted future expected temperature, and simultaneously extracts the order rate of change. This decouples the static position error of the trajectory from the dynamic trend characteristics, providing an orthogonal control input source for subsequent dual-channel compensation.

[0086] Specifically, upon completing step S5, the process control processor enters the calculation flow of step S6. The core task of this step is to generate two key signals for feedback and feedforward control. First, the processor locates the dynamic timing transition trajectory sequence stored in the first-in-first-out queue based on its real-time control cycle time. The control cycle time is the main loop clock of the control system software, which specifies the execution frequency of all periodic calculation tasks, for example, every Execute once. The processor pulls and reads the current single-step target temperature setpoint corresponding to the current control cycle from the head of the queue. The current single-step target temperature setpoint is the discrete value of the dynamic time-series transition trajectory sequence at the current time point, and is the instantaneous target of feedback control. This setpoint is the ideal temperature that the optimal path planned in step S4 should reach at the current moment.

[0087] Next, the processor performs an algebraic subtraction operation between the current single-step target temperature setpoint and the predicted future temperature generated in step S5. The result of this subtraction operation is the predictive control deviation. This deviation is not the error between the setpoint and the actual value used in traditional PID controllers, but rather the difference between the setpoint and the predicted value. It anticipates the following error that may occur in the system at the next moment, providing a basis for the control system's advance compensation and disturbance rejection.

[0088] Simultaneously, the processor performs a differential operation on the dynamic time-series transition trajectory sequence to extract its dynamic characteristics. The processor reads not only the current setpoint from the queue but also the setpoint for the next control cycle. By subtracting the current setpoint from the next cycle's setpoint and dividing by the control system's time step, the processor performs a first-order finite-difference operation. The result of this operation directly quantifies the rate of ascent or descent of the target trajectory at the current point, i.e., the order rate of change. This rate of change characterizes the target rate of temperature change that the system needs to achieve to accurately follow the trajectory and is the core input for calculating the feedforward compensation. Finally, the processor outputs the calculated predictive control deviation and order rate of change to different processing channels for use in the final fusion control step.

[0089] Predictive control deviation The calculation formula is as follows: Rate of change of order The calculation is performed using first-order finite difference operations: in, This is the current control cycle. The predictive control deviation, expressed in degrees Celsius. ). In the current control cycle The current single-step target temperature setpoint read from the dynamic time-series transition trajectory sequence, in units of... . This is the predicted future temperature value generated in step S5 during the current cycle, in units of... . In the current control cycle Extracted order rate of change, in degrees Celsius per second (°C) ). It is the next control cycle pre-read from the dynamic timing transition trajectory sequence. The target temperature setpoint, in units of . It is the discrete sampling time step of the control system, in seconds. ).

[0090] For example, following the previous steps, the control cycle is... In the current control cycle The system first reads the current single-step target temperature setpoint from the dynamic time-series transition trajectory sequence. Following the example in S4, assuming the current time point is the second time point of the trajectory, then... At the same time, the system also pre-fetched the target value for the next time point. In step S5, the system has calculated the predicted future temperature value. .

[0091] Next, the system performs a subtraction operation to generate the predictive control deviation: .

[0092] Simultaneously, the system performs first-order finite difference operations to extract the order rate of change: This result corresponds to the calculation of the target trajectory in S4. arrive The effective average temperature slope within the sampling step size verifies the self-consistency of the discrete trajectory differentiation calculation. Finally, the system outputs the predicted control deviation and the order rate of change for use in the synthesized control law.

[0093] S7. Calculate the proportional-integral adjustment component based on the predictive control deviation, calculate the feedforward compensation component based on the order rate of change, and combine the proportional-integral adjustment component and the feedforward compensation component into a smooth transition execution control output to drive the actuator of the autoclave.

[0094] In a preferred embodiment, step S7 includes: directly introducing the predicted control deviation into the proportional-integral control loop to perform calculations and solve for the proportional-integral adjustment component. The order change rate is multiplied by the local time constant in the local thermodynamic dynamic characteristic parameters and divided by the steady-state process gain execution scale dynamic scaling and leveling to generate the feedforward compensation component. The proportional-integral control component and the feedforward compensation component are algebraically summed to synthesize a smooth transition execution control quantity, which is then output to drive the heating thyristor regulator and cooling flow control valve of the autoclave.

[0095] This invention achieves complementary advantages of steady-state error elimination and dynamic trajectory tracking by combining the proportional-integral adjustment component that suppresses extremely low-frequency oscillation disturbances with the feedforward compensation component that matches the transient power follow command in parallel algebraically, thus avoiding the phase lag at the inflection point of single feedback control.

[0096] Specifically, step S7 is the final synthesis and output stage of the control command. The process control processor integrates the two core signals generated in step S6 to drive the physical actuator of the autoclave. First, the processor sends the predicted control deviation into a built-in proportional-integral control loop. In this loop, the deviation is multiplied by a proportional gain. The proportional term is obtained, and then the summation and integration are multiplied by an integral gain. The integral term is obtained. These two terms are added together to calculate the proportional-integral adjustment component. The main function of this component is to provide feedback correction based on the prediction of future errors, thereby eliminating low-frequency disturbances and steady-state errors caused by inaccurate or unmodeled dynamics.

[0097] In parallel, the processor performs the calculation of the feedforward compensation. It reads the order rate of change and retrieves the steady-state process gain from the local thermodynamic dynamic characteristic parameters obtained in step S2. and local time constant The processor performs a scale dynamic scaling and leveling operation, multiplying the order rate of change by... Divide by This operation is essentially the inverse operation of the system model, and its result directly provides the theoretical power increment required to achieve the target temperature change rate, i.e., the feedforward compensation component. This component can actively and quickly respond to the dynamic changes in the target trajectory and is the core of achieving high-precision tracking.

[0098] Finally, the processor synthesizes the proportional-integral adjustment component and the feedforward compensation component obtained from the two parallel computing channels. On the internal bus of the digital controller, through a parallel algebraic addition synthesis operation of a controller digital space channel, the two component values ​​are added together and superimposed with the base power output value locked in step S1 to form the final smooth transition execution control quantity. This control quantity is a specific power percentage value, which is immediately sent through the digital output interface to directly drive the physical actuators at the bottom of the autoclave, such as the thyristor power regulator controlling the bottom heater, or the constant flow control valve controlling the circulating fluid cooling section during the cooling stage, thereby completing the precise regulation of the heat exchange of the physical medium and closing the entire control loop.

[0099] The final output is a smooth transition control quantity. It is synthesized by the following formula: Among them, the proportional-integral adjustment component Calculated by the following formula: Feedforward compensation component Calculated by the following formula: in, This is the current control cycle. The final output is the smooth transition control quantity, expressed as a percentage of power. ). The base power reference represents the current operating point. This is the current control cycle. The proportional-integral adjustment component, in units of . This is the current control cycle. The feedforward compensation component, in units of . and These are the proportional gain and integral gain parameters of the proportional-integral control loop, which are tuned by engineers based on system characteristics, for example, set to 5% / and 0.1% / ( ). This is the current control cycle. The predictive control deviation, in units of . This is the current control cycle. The order of change, in units of . It is a local time constant. It is the steady-state process gain. It is the sampling period of the control system, in units of .

[0100] For example, continuing from the previous step, the system has already calculated the predictive control deviation. and order rate of change Base power output value The system parameters are: The PI controller parameters are assumed to be tuned to... , .

[0101] First, calculate the proportional-integral adjustment component. Assume the integral accumulator term is currently... ,but .

[0102] Next, calculate the feedforward compensation component: .

[0103] Finally, the system performs parallel algebraic addition to synthesize the final smooth transition control quantity: .Should The power command is immediately sent to the thyristor power controller, driving the heating element to operate at a higher power, thereby compensating for the delay caused by the thermal inertia of the system, ensuring that the temperature response can strictly follow the rapid rise requirements of the dynamic time-series transition trajectory sequence, and achieving a smooth, overshoot-free transition at the inflection point.

[0104] Example 2 Please see Figure 2 As shown, based on Embodiment 1, the second aspect of the present invention provides a smooth transition system for the inflection point of the temperature control curve of an autoclave, comprising: a disturbance testing and data acquisition module, a dynamic characteristic parameter identification module, a trajectory optimization objective function generation module, a transition trajectory generation module, a temperature prediction module, a deviation and rate of change calculation module, and a control quantity synthesis and output module.

[0105] The disturbance testing and data acquisition module extracts the temperature control inflection point of the preset autoclave process curve, superimposes the test disturbance signal into the heating output command before reaching the temperature control inflection point, and collects real-time temperature data in multiple spatial dimensions and normalizes and merges it to generate multi-level, multi-point temperature disturbance response data.

[0106] The dynamic characteristic parameter identification module analyzes the disturbance response data of multi-level and multi-point temperatures, constructs a local first-order inertial mathematical model to identify dynamic characteristic parameters, and calculates the local first-order inertial mathematical model to generate local thermodynamic dynamic characteristic parameters including local time constants and steady-state process gains.

[0107] The trajectory optimization objective function generation module calls local thermodynamic dynamic characteristic parameters as hard constraint boundaries and combines them with the characteristic slopes of the preset autoclave process curve before and after the temperature control inflection point to generate a boundary constraint trajectory planning objective function.

[0108] The transition trajectory generation module nonlinearly solves the objective function of boundary constraint trajectory planning and generates a dynamic temporal transition trajectory sequence that fully covers the temperature control inflection point over a time span.

[0109] The temperature prediction module injects local thermodynamic dynamic characteristic parameters into the internal state predictor to perform single-step time-series forward state calculations and generate future expected temperature prediction values.

[0110] The deviation and rate of change calculation module performs a subtraction algebra operation between the dynamic time-series transition trajectory sequence and the predicted future temperature to generate the predictive control deviation, and performs differentiation operation on the dynamic time-series transition trajectory sequence to extract the order rate of change.

[0111] The control quantity synthesis and output module calculates the proportional-integral adjustment component based on the predicted control deviation and the feedforward compensation component based on the order rate of change. It then synthesizes the proportional-integral adjustment component and the feedforward compensation component into a smooth transition execution control quantity output to drive the actuator of the autoclave.

[0112] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for smoothing the inflection point of an autoclave temperature control curve, characterized in that, include: S1. Extract the temperature control inflection point of the preset autoclave process curve, and superimpose the test disturbance signal into the heating output command before reaching the temperature control inflection point. Collect real-time temperature in multiple spatial dimensions and normalize and merge to generate multi-level multi-point temperature disturbance response data. S2. Analyze the disturbance response data of multi-level and multi-point temperature, construct a local first-order inertial mathematical model to identify dynamic characteristic parameters, and solve the local first-order inertial mathematical model to generate local thermodynamic dynamic characteristic parameters including local time constant and steady-state process gain. S3. Call the local thermodynamic dynamic characteristic parameters as hard constraint boundaries, and combine them with the characteristic slopes of the preset autoclave process curve before and after the temperature control inflection point to generate the objective function for boundary constraint trajectory planning. S4. Nonlinearly solve the objective function of boundary constraint trajectory planning to generate a dynamic time-series transition trajectory sequence that fully covers the temperature control inflection point over a time span. S5. Inject local thermodynamic dynamic characteristic parameters into the internal state predictor to perform single-step time-forward state calculation and generate the future expected temperature prediction value. S6. Perform a subtraction algebraic operation between the dynamic time-series transition trajectory sequence and the predicted future temperature to generate the predictive control deviation, and perform a derivative operation on the dynamic time-series transition trajectory sequence to extract the order rate of change. S7. Calculate the proportional-integral adjustment component based on the predictive control deviation, calculate the feedforward compensation component based on the order rate of change, and combine the proportional-integral adjustment component and the feedforward compensation component into a smooth transition execution control output to drive the actuator of the autoclave.

2. The method for smoothing the inflection point of the temperature control curve of an autoclave according to claim 1, characterized in that, Step S1 includes: Extract the intersection timestamps of two adjacent temperature control line segments with different slopes in the preset autoclave process curve as the temperature control inflection point, and set a reserved pre-time threshold earlier than the temperature control inflection point. When the actual processing time is detected to reach the reserved pre-time threshold, the base power output value of the current working point is kept constant, and a set of test disturbance signals that meet the preset target spectrum and amplitude constraints are superimposed on the heating actuator of the autoclave. Simultaneously collect real-time temperatures in multiple spatial dimensions within the workpiece area and airflow circulation area inside the autoclave. Then, align the real-time temperatures in multiple spatial dimensions with the test disturbance signals based on timestamps, normalize and merge them to generate multi-level, multi-point temperature disturbance response data.

3. The method for smoothing the inflection point of the temperature control curve of an autoclave according to claim 2, characterized in that, A set of test disturbance signals that meet preset target spectrum and amplitude constraints are superimposed onto the heating actuator of the autoclave, including: Activate the internal signal generator module to generate a predefined pseudo-random binary sequence; The minimum pulse width of the pseudo-random binary sequence is set according to the nominal dominant time constant of the autoclave system to determine the target spectrum; The amplitude constraint of the pseudo-random binary sequence is set to a preset percentage range of the maximum heating power of the autoclave to generate a test disturbance signal.

4. The method for smoothing the inflection point of the temperature control curve of an autoclave according to claim 1, characterized in that, Step S2 includes: Extract the time-series scatter set of temperature bias values ​​generated by superimposed test disturbance signals from the multi-level, multi-point temperature disturbance response data; Using the test disturbance signal as the input variable and the time-series scatter set of temperature bias as the output variable, a first-order differential equation is constructed to map the dynamic relationship between heating power change and temperature response, serving as a local first-order inertial mathematical model. Based on the local first-order inertial mathematical model, dynamic characteristic parameters are identified, and the local time constant characterizing the transient thermal inertia of the autoclave and the steady-state process gain characterizing the input-output ratio are calculated and extracted, and then combined and packaged into local thermodynamic dynamic characteristic parameter output.

5. The method for smoothing the inflection point of the temperature control curve of an autoclave according to claim 1, characterized in that, Step S3 includes: The local time constants contained in the local thermodynamic dynamic characteristic parameters are inversely calculated to obtain the temperature acceleration constraint threshold of the physical system's allowable response; Extract the first set temperature change slope of the line segment before the temperature control inflection point in the preset autoclave process curve, and the second set temperature change slope of the line segment after the temperature control inflection point. The goal is to minimize the transition time from the first set temperature change slope to the second set temperature change slope and ensure the trajectory converges monotonically, using this as the optimization benchmark for the closed-loop cost function. A temperature acceleration constraint threshold is introduced as a hard constraint boundary to generate the objective function for boundary constraint trajectory planning.

6. The method for smoothing the inflection point of the temperature control curve of an autoclave according to claim 5, characterized in that, Step S4 includes: Within the current prediction triggering period, a nonlinear programming solution algorithm is run to perform a finite-time-space iterative solution for the objective function of the boundary-constrained trajectory planning; Obtain the time series curve of the planning acceleration that follows the hard constraint boundary, and perform a quadratic cumulative integral calculation in the discrete time domain; The discrete numerical results calculated by the second cumulative integral are shifted and connected based on the current measured initial physical temperature to generate a dynamic time-series transition trajectory sequence in which the initial tangent slope matches the first set temperature change slope and the final tangent slope matches the second set temperature change slope. Among them, the second-order cumulative integral calculation in the discrete time domain includes: taking the first set temperature change slope as the initial value of the integral, multiplying the acceleration value of the planned acceleration time series curve of the current step by the time step and accumulating it to the slope value of the previous step at each discrete control time step to generate the temperature change slope trajectory. Using the initial zero value as the initial value for integration, at each discrete control time step, the average slope value of the temperature change slope trajectory of the current step is multiplied by the time step and accumulated to the relative temperature value of the previous step to generate the relative temperature transition trajectory.

7. The method for smoothing the inflection point of the temperature control curve of an autoclave according to claim 1, characterized in that, Step S5 includes: The local time constant and steady-state process gain contained in the local thermodynamic dynamic characteristic parameters are extracted and updated into the state equation coefficient variable array of the internal state predictor. The average feedback temperature of the actual space inside the autoclave in the previous sampling control cycle is retrieved and used as the recursive initial state reference for the internal state predictor to perform state recursion. Based on the state equation coefficient variable array, the recursive initial state reference, and the current control command, the internal state predictor is triggered to perform a single-step timing forward state calculation to generate the expected future temperature prediction value for the next sampling control cycle.

8. The method for smoothing the inflection point of the temperature control curve of an autoclave according to claim 1, characterized in that, Step S6 includes: Within each sampling control cycle, acquire the current single-step target temperature setpoint in the dynamic time-series transition trajectory sequence; The current single-step target temperature setpoint is subtracted from the future expected temperature prediction value to generate the predictive control deviation. Perform a first-order finite difference operation on the dynamic time-series transition trajectory sequence to extract the order rate of change.

9. The method for smoothing the inflection point of the temperature control curve of an autoclave according to claim 1, characterized in that, Step S7 includes: The predictive control deviation is directly introduced into the proportional-integral control loop for calculation, and the proportional-integral adjustment component is calculated. The order change rate is multiplied by the local time constant in the local thermodynamic dynamic characteristic parameters and divided by the steady-state process gain execution scale dynamic scaling and leveling to generate the feedforward compensation component. The proportional-integral control component and the feedforward compensation component are algebraically summed to synthesize a smooth transition execution control quantity, which is then output to drive the heating thyristor regulator and cooling flow control valve of the autoclave.

10. A system for smoothing the inflection point of an autoclave temperature control curve, characterized in that, include: The disturbance testing and data acquisition module extracts the temperature control inflection point of the preset autoclave process curve, superimposes the test disturbance signal into the heating output command before reaching the temperature control inflection point, and collects real-time temperature in multiple spatial dimensions and normalizes and merges it to generate multi-level multi-point temperature disturbance response data. The dynamic characteristic parameter identification module analyzes the disturbance response data of multi-level and multi-point temperature, constructs a local first-order inertial mathematical model to identify dynamic characteristic parameters, and solves the local first-order inertial mathematical model to generate local thermodynamic dynamic characteristic parameters including local time constants and steady-state process gains. The trajectory optimization objective function generation module calls local thermodynamic dynamic characteristic parameters as hard constraint boundaries and combines them with the characteristic slopes of the preset autoclave process curve before and after the temperature control inflection point to generate the boundary constraint trajectory planning objective function. The transition trajectory generation module nonlinearly solves the objective function for boundary constraint trajectory planning and generates a dynamic temporal transition trajectory sequence that fully covers the temperature control inflection point over a time span. The temperature prediction module injects local thermodynamic dynamic characteristic parameters into the internal state predictor to perform single-step time-series forward state calculations and generate future expected temperature prediction values. The deviation and rate of change calculation module performs a subtraction algebra operation between the dynamic time-series transition trajectory sequence and the predicted future temperature to generate the predictive control deviation, and performs differentiation operation on the dynamic time-series transition trajectory sequence to extract the order rate of change. The control quantity synthesis and output module calculates the proportional-integral adjustment component based on the predicted control deviation and the feedforward compensation component based on the order rate of change. It then synthesizes the proportional-integral adjustment component and the feedforward compensation component into a smooth transition execution control quantity output to drive the actuator of the autoclave.

Images