Control method of thermostat circuit of sensor

Abstract

The application is a control method of constant temperature circuit of sensor, comprising the following steps: the circuit controller acquires information including initial ambient temperature, sensor structure parameters and historical heating response curve, establishes a temperature rise prediction model based on the information, and outputs temperature rise rate at multiple power levels; in the temperature rise starting stage, the circuit controller selects a target power level according to the prediction model, controls the heating element to be alternately turned on and off in a pulse mode with a first period 1, and realizes stepwise temperature rise; the application has the beneficial effects that: by dividing the sensor heating process into three stages of exploratory preheating, stepwise temperature rise and steady state maintenance, and combining real-time temperature sampling, dynamic modeling and control function feedback control, the sensor temperature can be accurately and stably controlled in the target temperature zone, which is much better than the traditional single-link constant voltage control method, ensures that the sensor works under the best thermal conditions, and improves the detection accuracy and repeatability.

Description

Technical Field

[0001] This invention relates to a control method for a constant temperature circuit, specifically a control method for a sensor's constant temperature circuit. Background Technology

[0002] In the currently published CN105159359B "Heating Method and Control Circuit of Wide-Range Oxygen Sensor", the control circuit design focuses on simplicity and practicality. It adopts a closed-loop adjustment of the heating resistor through comparator and PWM control, which to a certain extent improves the control accuracy of oxygen sensor temperature and the response flexibility of the circuit.

[0003] However, through in-depth analysis of the contradictions between this solution and current industrial needs, it can be found that it has many shortcomings and limitations in several key aspects such as control strategy, dynamic adaptability, isothermal stability, safety of temperature rise process and system scalability, making it difficult to meet the systematic requirements of the new generation of high-performance sensor systems for intelligent isothermal control. First, the core control logic of this scheme relies on the resistance fluctuation of the oxygen sensor's internal resistance as a function of temperature change to determine whether to heat or stop. Essentially, it uses the sensor's internal resistance as the sole indirect temperature estimation parameter. However, in real-world applications, the oxygen sensor's resistance is affected by multiple factors, including but not limited to film aging, ambient humidity, power supply noise, and lead voltage drop. Therefore, the resistance signal is easily amplified by interference, causing the comparator to flip incorrectly, resulting in unstable temperature control accuracy and an inability to guarantee precise temperature maintenance within a small range. Second, the comparator output is only a single flip state, forming a coarse-grained power regulation mechanism with PWM control. It cannot perceive the dynamic characteristics of the temperature rise process, such as the rate of temperature increase, hysteresis, and changes in thermal inertia. It also lacks a temperature change trend prediction mechanism. The control strategy is biased towards static feedback, lacking feedforward regulation and model compensation capabilities. Therefore, it is not timely enough to respond to temperature disturbances in complex working environments, such as cold starts, air convection interference, or sudden changes in sensor load, easily leading to overshoot or regulation lag. Furthermore, the solution does not incorporate the process batch, structural parameters, or historical behavior model of the sensor into the control loop, nor does it design a temperature control reference table or a multi-stage heating strategy. The heating path is singular and cannot be dynamically adjusted according to the specific response characteristics of the sensor. Especially when dealing with sensors with significant differences in electrothermal parameters between batches, the consistency of isothermal behavior is poor, and the system's repeatability control capability is limited.

[0004] Furthermore, while its PWM control duty cycle is adjustable, the adjustment strategy relies solely on the microprocessor processing based on the comparator output waveform characteristics. There is no scalable control algorithm interface, making it unable to support advanced functions such as multi-strategy adaptive algorithms, neural network predictive modeling, or thermal inertia self-learning. This results in weak system scalability, failing to meet the diverse needs of high-precision temperature control systems under different operating conditions. Simultaneously, the lack of a soft-start mechanism during the heating process leads to uncontrollable heating rates during cold starts, easily causing thermal shock and reducing sensor lifespan. This is particularly problematic in applications where ceramic encapsulation materials coexist with highly heat-sensitive films, increasing the risk of localized thermal stress cracking. In the steady-state phase, its PWM control logic only provides constant output adjustment, lacking a fluctuation judgment mechanism based on temperature trend classification and a stepped output function to buffer external abrupt disturbances. It has no suppression mechanism for external environmental changes such as high-frequency airflow disturbances and power supply voltage fluctuations, easily leading to spikes in the constant-temperature output. Finally, while the circuit design simplifies the structure and reduces costs, the control logic is hard-coded, lacks an abstract and adjustable control framework, does not support thermal control function modeling, and lacks adaptive closed-loop optimization capabilities between the sensor, controller, and heating element, thus limiting its industrial expansion and deployment in fields requiring high reliability, high response, and intelligent collaboration. Summary of the Invention

[0005] The purpose of this invention is to provide a control method for the constant temperature circuit of a sensor, thereby solving some of the drawbacks and shortcomings pointed out in the background art.

[0006] The technical solution of this invention: a control method for a temperature-controlled circuit of a sensor, applied to a temperature-controlled circuit including a heating element, a temperature acquisition unit, and a power supply module, wherein the temperature-controlled circuit is driven by a circuit controller, characterized by comprising the following steps:

[0007] The circuit controller acquires information including the initial ambient temperature. Based on the sensor's structural parameters and historical heating response curves, a temperature rise prediction model is established, and the output is displayed at multiple power levels. The rate of temperature rise at the following conditions ;

[0008] During the heating start-up phase, the circuit controller selects the target power level according to the prediction model and controls the heating element to operate in a first cycle. The pulse mode alternates on and off to achieve stepped heating;

[0009] The circuit controller in the second cycle Internally based on real-time temperature Fine-tuning the power level with the temperature rise prediction model Calculate the rate of temperature change ΔT / Δt; when |ΔT / Δt|≤ Record the current temperature. As the first reference temperature point ;

[0010] Near the first reference temperature point, the circuit controller in each control cycle Internal analysis of the temperature change trend in the previous cycle And the predicted temperature rise Δt is obtained by combining the prediction delay δt. If Δ > Then reduce or pause power output, if <– This increases power output;

[0011] When | – |< And when the fluctuation trend falls within the noise judgment region N, maintain the current power output; when | – |≥ When sampling is continuously offset n times in the same direction, a compensation power ΔP is applied to adjust the real-time temperature. Return to the first reference temperature point ± The noise judgment region N is the range of random fluctuations in the sensor temperature, and the specific judgment criteria are as follows:

[0012] Real-time temperature First reference temperature point The deviation satisfies | – |< The temperature change direction is inconsistent within 10 consecutive sampling periods; the temperature change in a single sampling period is |ΔT| ≤ 0.05°C; when all three conditions are met, the temperature fluctuation is determined to belong to the noise judgment zone N.

[0013] The compensation power ΔP is calculated based on the temperature offset and the thermal inertia coefficient, and the specific formula is: ΔP = k × | - | / ×C

[0014] Where k is the power compensation coefficient, taken as 0.8 W·s / °C, obtained from the sensor's thermal resistance and thermal capacity calibration; | - | represents the absolute offset between the real-time temperature and the reference temperature point, in °C; The steady-state control period is expressed in milliseconds (ms) and needs to be converted to seconds for calculation. C represents the sensor's heat capacity, expressed in J / °C, and is obtained from the sensor's structural parameters.

[0015] The compensation power ΔP is achieved by adjusting the pulse width, and the pulse width adjustment amount is... The duty cycle is adjusted synchronously with the pulse width, where U is the power supply voltage of the heating element and R is the resistance of the heating element.

[0016] The temperature acquisition unit indirectly obtains the sensor temperature by detecting the resistance change of the heating element. The resistance-temperature calibration relationship is provided by the sensor manufacturer, and the calibration accuracy is ±0.1°C.

[0017] Furthermore: the first cycle The pulse width is 1ms to 20ms, the second cycle The duration is 50ms to 500ms; where... The preset threshold is fixed at the factory or adaptively set according to the sensor model and historical data each time it is started. It can be 0.05°C / s, 0.1°C and 0.2°C respectively. In each control cycle, the pulse width is adjusted in steps of 0.1ms according to the deviation between the current temperature and the target temperature until the temperature change rate falls within the preset threshold range. The value is determined based on the sensor's temperature detection accuracy and thermal inertia characteristics. For heated metal-oxide-semiconductor sensors, the temperature detection accuracy is ±0.05°C, and the temperature hysteresis due to thermal inertia is 0.1°C. Therefore, the value is set... Twice the detection accuracy It is twice the amount of thermal inertia hysteresis. The thermal inertial hysteresis rate per unit time; the temperature rise prediction model adopts first-order linear regression or nonlinear fitting based on neural networks.

[0018] The nonlinear fitting based on neural networks employs a three-layer perceptron neural network. The specific structure and training method are as follows: the input layer has two nodes, representing the heating power... The system includes a cumulative running time τ; a hidden layer of 10 nodes with the sigmoid activation function σ(x) = 1 / (1+eX); an output layer of 1 node, representing the predicted temperature rise rate dT / dt; power, time, and temperature rise rate data from the sensor's historical heating response curves, with a sample size of no less than 1000 sets; a backpropagation algorithm with mean squared error (MSE) as the loss function and a learning rate of 0.01, training is stopped when the loss function is less than 0.001; historical heating response curves are stored in the non-volatile memory of the circuit controller, and the controller automatically retrieves the historical data of the corresponding batch through the sensor batch identifier.

[0019] Furthermore: the step-by-step heating method includes the following steps:

[0020] (S1) The circuit controller generates a step comparison table corresponding to different power levels and target temperature rise increments;

[0021] (S2) Select the first power level according to the step reference table, control the heating element to alternately turn on and off in the pulse mode of the first cycle, collect the current temperature after each pulse and update the next pulse width or duty cycle, so that the temperature rise curve rises in a step-like manner.

[0022] (S3) If the real-time temperature rise rate of this step is detected to be lower than the predicted range, extend the power-on time of the next pulse; if it is higher than the predicted range, shorten the power-on time to eliminate the temperature rise error.

[0023] (S4) When the target temperature rise increment corresponding to this step reaches the preset value, the circuit controller immediately switches to the next power level in the step lookup table and repeats steps (S2) to (S3) until the temperature approaches the target constant temperature zone.

[0024] (S5) After entering the target constant temperature zone, the circuit controller will switch the pulse period between long and short, and determine whether to perform small-amplitude power compensation based on the temperature trend obtained by continuous sampling in order to maintain temperature stability.

[0025] Furthermore: the step reference table is regenerated based on real-time initial conditions each time it is started, in order to adapt to differences in sensor batches and changes in the external environment; the temperature rise error determination in step (S3) is based on the temperature change rate range.

[0026] Further: the step lookup table generation method includes,

[0027] The circuit controller controls the heating element to perform short-term exploratory heating at low power and continuously collects temperature change data to analyze the current thermal response behavior of the sensor. Based on the temperature rise rate, response delay and thermal inertia during the preheating period, the circuit controller dynamically generates a step lookup table for temperature rise control. The lookup table includes temperature increment steps, with each step corresponding to a single heating power level and pulse control parameters.

[0028] During the heating process, the circuit controller adjusts the power output and pulse rhythm of the heating element step by step according to the step lookup table to ensure a smooth temperature rise. If a sudden change in the environment or a temperature rise deviation is detected during the heating process, the circuit controller re-corrects the step lookup table or adjusts the current step parameters to adapt to external disturbances or device differences.

[0029] Specifically, the above technical solution involves the controller driving the heating element with low power for a short-term exploratory heating during the startup phase, while simultaneously acquiring the temperature change sequence output by the sensor. Based on the observed temperature rise rate, response delay, and thermal inertia, a temperature rise comparison model is fitted and generated. This model uses the following step response control function as the control benchmark:

[0030]

[0031] in:

[0032] Let be the heating power adjustment function, representing the function at a temperature of . Running until time The heating control intensity that should be output at that time; For the first The reference power coefficient for each temperature step is used to adjust the target energy input, and its value ranges from 0.3 to 0.7. The nonlinear adjustment factor for compensating for the thermal inertia of the step ranges from 0.1 to 0.25. : Represents the thermal slow response index, reflecting the hysteresis characteristics of the heating element within this range, with a value range of 0.015 to 0.04; : indicates the first The temperature span of each temperature control step; For definition in The switching function within the interval takes a value of 1 only if the current time falls within the step; otherwise, it takes a value of 0. The current sensor temperature (indirectly reflected by voltage or resistance); This represents the cumulative time the system has been running.

[0033] Implementation process description:

[0034] 1. Trial heating and data acquisition stage: The controller applies a low-power pulse to the heating element and simultaneously acquires temperature change data, recording the temperature rise rate, start-up delay, and thermal response trend.

[0035] 2. Control Function Modeling and Step Table Generation: Based on the temperature rise data mentioned above, the system automatically determines multiple heating steps. Each step is dynamically established based on the temperature span, required response rate, and power. Control Function coefficients in The controller fits and generates a heating output model for temperature rise; the control function... coefficients in Nonlinear fitting is performed using the least squares method. The specific fitting steps are as follows:

[0036] Temperature-time series data {T} collected during the exploratory heating phase k ,τ k} (k=1,2,...,m, where m is the number of sampling points, not less than 50) are the fitted samples; the objective function for fitting is to minimize the sum of squared residuals: min∑ m k=1 [Φ(T k ,τ k )-Tk] 2 The constraints are The gradient descent method is used to iteratively solve for each temperature step. The iterative convergence threshold is set to 10. -6 ,

[0037] The maximum number of iterations is 1000; after fitting, the correlation coefficient R is used to determine the result. 2 To verify the fit, R 2 A value ≥0.95 indicates a valid fit; otherwise, data should be recollected and the number of sampling points increased.

[0038] 3. Gradual heating and real-time correction mechanism: During the heating process, the controller monitors temperature changes in real time and adjusts accordingly. The defined power output value is used for step-by-step heating control; if the system detects an abnormal temperature rise trend (such as slow heating or sudden external cooling), the parameters of the current step are immediately refitted or a new step is dynamically inserted to make the lookup table highly adaptable.

[0039] 4. Steady-state entry and isothermal maintenance: When the target temperature is approached, the controller utilizes the function decay segment... The control output tends to be stable, avoiding overshoot, and smoothly transitioning to a constant temperature steady state.

[0040] Introducing control functions The temperature control process is dynamically described, making the control process mathematically adjustable. The sine square term in the control function is used to smooth the step switching, and the exponential decay term is used to suppress thermal inertia error, reflecting in-depth modeling of thermal behavior. The piecewise step function combination strategy is used to discretize and controllable the heating process, improving the system's adaptability to device differences and changes in the external environment. The control system has self-evolution capability through real-time fitting and reconstruction of parameters within the function, making it suitable for industrial-grade sensors in complex environments.

[0041] Furthermore: the power duty cycle of the exploratory preheating is 1% to 5%, with the lower limit of 1% being the minimum duty cycle to ensure the collection of effective temperature change data, and the duration is 1 to 2 seconds; the step lookup table is generated in real time after each power-on start-up of the device; the power output corresponding to each step in the lookup table includes the heating pulse width, cycle frequency, and expected temperature rise increment.

[0042] Furthermore: the thermal response behavior includes one of the following indicators: temperature change rate, heating delay time, and heat transfer inertia judgment result; when the batch identifier of the sensor is available, the circuit controller combines the historical heating response data of the batch to perform model fitting and comparison table optimization; the adaptive adjustment process includes identifying the disturbance trend in the heating process and adjusting the current step power or switching to the next step in advance according to the disturbance amplitude and duration.

[0043] Furthermore: the method for maintaining temperature stability includes,

[0044] When the temperature approaches the target constant temperature value, the circuit controller switches the heating control mode from short-cycle pulse mode to long-cycle pulse mode; during the constant temperature maintenance phase, the circuit controller continuously collects multiple sets of temperature data at fixed time intervals, and analyzes the temperature change trend type based on the sampling results, including stable fluctuation, continuous deviation or sudden change; when the trend analysis result shows that the temperature is near the target value and the fluctuation is within the set threshold range, the current power output is maintained and no adjustment is made.

[0045] When the trend analysis result shows that the temperature continues to shift upward or downward and continues to exceed a preset number of times, the circuit controller triggers a small power compensation, which includes appropriately increasing or decreasing the pulse width, duty cycle, or power output within the cycle.

[0046] When the trend analysis results show a sudden abnormal fluctuation in temperature but no sustained deviation, the circuit controller judges it as an external disturbance and refuses to adjust the power.

[0047] Furthermore: the short period is 1 millisecond to 20 milliseconds, and the long period is 50 milliseconds to 500 milliseconds; the circuit controller uses no less than 5 temperature sampling points for trend analysis, and judges the current trend by combining the direction and magnitude of temperature change; the trigger threshold for the small compensation is when the temperature deviates from the target value by more than ±ε and continues for more than n consecutive sampling periods.

[0048] Furthermore: the compensation power adjustment actions are spaced one sampling period apart to prevent over-response; the trend type includes one of the following three:

[0049] a. Normal fluctuation trend: The temperature does not deviate continuously within the preset range;

[0050] b. Continuous deviation trend: The temperature continuously deviates from the target value and changes in the same direction;

[0051] c. Disturbance trend: Single or occasional sharp changes that do not form a sustained trend.

[0052] The circuit controller includes a microprocessor and a non-volatile memory. It integrates a temperature rise prediction model based on first-order linear regression or a lightweight neural network. The model parameters are generated by fitting the model during the trial heating stage using the least squares method or gradient descent method and stored in the memory for subsequent control cycles.

[0053] The sensor structural parameters include the heating element thermal capacity C, the package thermal resistance Rth, the heating element resistance R, the heating element area S, and the ceramic package thickness d, all of which are sensor thermal characteristic parameters known to those skilled in the art.

[0054] The control method of the constant temperature circuit of the sensor of the present invention has the following significant advantages, specifically reflected in multiple aspects such as temperature control accuracy, adaptive capability, anti-disturbance stability, equipment safety, and universal adaptability:

[0055] By dividing the sensor heating process into three stages—exploratory preheating, step-by-step heating, and steady-state maintenance—and combining real-time temperature sampling, dynamic modeling, and control function feedback control, the sensor temperature can be precisely and stably controlled within the target temperature range, such as ±0.3°C. This is significantly better than the traditional single-stage constant pressure control method, ensuring that the sensor operates under optimal thermal conditions and improving detection accuracy and repeatability.

[0056] By analyzing the actual thermal response characteristics of the sensor through trial heating behavior, and dynamically generating a temperature rise control model and step comparison table based on response rate, thermal inertia and batch characteristics, it can automatically adapt to sensor batch differences, structural errors and external environmental interference such as temperature or wind speed, and achieve adaptive control effect every time it starts, supporting different sensors to maintain a high degree of temperature control consistency under different operating conditions.

[0057] During the constant temperature maintenance phase, a trend-based control strategy is adopted, which can distinguish between three types of temperature change trends: normal fluctuations, continuous deviations, and sudden disturbances. This avoids accidental adjustment actions due to short-term disturbances, and can provide timely and small-scale compensation for true continuous deviations. Furthermore, setting adjustment intervals prevents over-adjustment, effectively enhancing the system's anti-disturbance and thermal stability. Attached Figure Description

[0058] Figure 1 This is a flowchart of the sensor constant temperature circuit control based on thermal response modeling of the present invention.

[0059] Figure 2 This is a flowchart of the second cycle temperature rise fine-tuning and reference temperature point determination process of the present invention.

[0060] Figure 3 This is a schematic diagram of the steady-state prediction and regulation logic of the intelligent temperature control system of the present invention.

[0061] Figure 4 This is a flowchart of the tolerance judgment and trend response of the temperature control system of the present invention.

[0062] Figure 5 This is the main flowchart of the adaptive heating intelligent temperature control process in Embodiment 1 of the present invention.

[0063] Figure 6 This is the overall process of stepped heating and constant temperature control in Embodiment 2 of the present invention.

[0064] Figure 7 This is a flowchart of multi-step nonlinear modeling and adaptive control of an industrial heating gas sensor according to Embodiment 3 of the present invention.

[0065] Figure 8 This is a block diagram illustrating the constant temperature control principle of the high-temperature sensor for an industrial furnace in Embodiment 4 of the present invention. Detailed Implementation

[0066] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0067] Combined with appendix Figure 1 As shown, this invention provides a control method for the temperature-controlled circuit of a sensor. This method is led by a circuit controller and based on modeling the thermal response behavior of the sensor to achieve dynamic, precise, and stable temperature control management of the sensor's heating element. Specifically, the circuit controller first acquires initial parameter information related to the current operating environment and the target sensor during the system power-on or startup phase. This includes the current initial ambient temperature T0, preset or automatically identified sensor structural parameters such as the heating element's thermal capacity, encapsulation thermal resistance, and heating element area, as well as heating response curve data generated by the sensor during historical operating conditions.

[0068] The circuit controller integrates and analyzes the above information to establish a temperature rise prediction model, which can predict temperature rise for different power levels. The output is the rate of temperature rise per unit time, i.e., the rate of temperature change dT / dt, forming a predictive comparison relationship of power-response characteristics. During the heating start-up phase, the circuit controller does not directly apply maximum power heating. Instead, based on the aforementioned predictive model, it selects the target power level from several power levels that offers the optimal rate of temperature rise under current environmental conditions without causing thermal shock, and controls the heating element to operate at the specified power level in a first cycle. The pulse mode is used for on / off control. This pulse mode is an intermittent power supply modulation. By controlling the duty cycle between power on and power off, energy is gradually injected. The resulting temperature rise process is characterized by a gradual linear or nonlinear increase, thus constructing a stepped heating mode with thermal response prediction capability.

[0069] Combined with appendix Figure 2 As shown, after completing the first cycle After the internal stepped temperature increase, the circuit controller then enters the second cycle. The control strategy for this cycle focuses on detailed monitoring and fine-tuning of the response behavior during the temperature rise process. During this phase, the controller continuously acquires the current real-time temperature values ​​from the sensors. It is then matched and calculated with the temperature rise prediction model built in the first cycle to determine the current operating power level. Fine-tuning is performed, which does not change the overall power amplitude, but rather achieves a smooth output transition by optimizing pulse modulation parameters, adjusting the duty cycle, or introducing time perturbations.

[0070] After fine-tuning, the controller calculates the rate of temperature change over a short period of time. ,in This indicates the amount of temperature change within the current period. The length of time during which this change occurs, if the rate of temperature change is monitored to satisfy the following stability condition. If the temperature rise curve shows a slowing down or approaching thermal equilibrium, the controller will immediately record the temperature value collected at that moment. Recorded as the first reference temperature point This temperature point is regarded as the critical value for subsequent thermal balance criteria and isothermal regulation algorithm. It has preliminary stability and can reflect the thermal inertia effect under the current sensor-environment coupling state. The setting of this reference temperature point not only constitutes the target baseline for subsequent closed-loop regulation, but also serves as a key indicator for evaluating the efficiency of the heating process and the matching of the sensor structure, thereby providing a mathematical basis and practical feedback benchmark for further precise control of the heating element.

[0071] Combined with appendix Figure 3 As shown, after obtaining the first reference temperature point Subsequently, the circuit controller enters the steady-state evaluation phase. Its control strategy revolves around the dynamic holding of the reference temperature point. To further improve the adaptability of the temperature control system to environmental disturbances and component differences, the controller, near the reference temperature point (i.e., when temperature fluctuations do not exceed the set range), performs a dynamic holding phase in each control cycle. Continuously perform temperature trend analysis, specifically including analyzing the temperature change trend recorded in the previous period. With response lag prediction time Based on comprehensive analysis and combining previous model parameters or historical behavior data, the circuit controller calculates the short-term predicted temperature rise. Its physical meaning is: if the current power output remains unchanged, the system expects the possible temperature shift after a certain response delay time in the future;

[0072] Based on this, the control logic sets dual threshold judgment conditions; if the predicted temperature rise meets the condition... This indicates that thermal inertia will cause the temperature to continue to rise and there is a risk of overshoot. In this case, the controller will immediately reduce the current power output or directly stop it to release the accumulated heat; conversely, if... This indicates that the current heat input is insufficient to maintain temperature stability, and the system temperature may deviate from the target range due to external cooling or power fluctuations. At this time, the controller will increase the power output, specifically by increasing the pulse duty cycle or briefly increasing the power level.

[0073] Combined with appendix Figure 4 As shown, a tolerance judgment and response strategy for actual temperature fluctuations is further introduced, in which the controller continuously monitors the real-time temperature. Compared with the first reference temperature point absolute deviation between And compare this deviation with the stability tolerance threshold. Compare, if the judgment result satisfies Furthermore, the temperature shift did not show a clear trend over several consecutive periods but exhibited irregular fluctuations, meaning its trajectory was categorized within the noise judgment zone set by the system. If the current temperature control system is in the natural perturbation range of equilibrium, the controller will maintain the current power output to avoid over-response or power oscillation caused by ineffective fine-tuning.

[0074] But if the controller detects And this offset is continuous If the temperature remains in the same direction throughout the sampling period, consistently above or below the reference temperature, a stable offset signal is formed. At this point, the system is considered to have deviated from the thermal equilibrium region, and the controller immediately triggers the temperature compensation mechanism, calculating the compensation power based on the current offset direction. To increase or decrease energy injection, thereby improving real-time temperature. Actively pulled back to the target area The compensation process is gradual and is modified by combining the aforementioned power ladder table and response lag model to ensure that the adjusted temperature rise or fall will not trigger a new round of dynamic shift. This strategy distinguishes the error type by judging the difference between volatility and trend, reflecting the mechanism for separating and responding to random disturbances and trend shifts in the steady-state control stage, and significantly improving the stability and robustness of the system temperature control.

[0075] Example 1: Combined with Appendix Figure 5 As shown, in this embodiment of the invention, the target sensor is set as a heated metal-oxide-semiconductor sensor for gas concentration monitoring, with a normal operating temperature range of 720°C to 760°C and an optimal stable temperature of 740°C. The ambient temperature acquired during initialization by the control system is 25°C. The sensor's structural parameters are given in the factory configuration file: heat capacity is approximately 0.15 J / °C, and thermal resistance is 2.5°C / W. Historical heating response curves show that under a fixed heating power of 0.6W, the temperature rise curve exhibits a near-linear upward trend within 5 seconds, gradually slowing down in the later stages.

[0076] The first cycle after system startup Internally, the circuit controller uses a pulse method to tentatively drive the heating element. The initial pulse width is set to 5ms, the duty cycle to 40%, and the period to 50ms. A total of 10 pulses are applied continuously, and the temperature change is measured in real time. The average temperature rise rate during this stage is 0.041°C / s, which is close to but has not yet reached the threshold of the temperature rise prediction model. Therefore, the system adaptively increases the pulse width from 5ms to 8ms in subsequent pulses and adjusts the duty cycle to 55% in the next cycle, with a step accuracy of 0.1ms for finer control of power injection. The temperature rise rate then increases to 0.053°C / s, exceeding... The system enters the second cycle. The cycle duration was set to 200ms, and a preliminary temperature rise prediction model was established using first-order linear regression: the temperature change data in the first 5 seconds was used as the training set to fit the temperature rise trend, and the slope of the prediction model was obtained, that is, the temperature rise rate was about 0.051°C / s.

[0077] exist During the cycle, the controller predicts the temperature rise within a future delay of 0.4 seconds based on the current temperature trend and power output. Using regression models or nonlinear neural networks, it can be inferred that when the system's estimated temperature fluctuates around 740.5°C at a certain moment, that is... Greater than the set value If this happens, the controller will immediately reduce the duty cycle from 55% to 50% in the next cycle and insert a 10ms power pause.

[0078] After several cycles, a second test showed that the temperature rise had stabilized, and At this point, the system records the current temperature as 739.8°C and sets it as the first reference temperature point. Next, the controller will use each Temperature stability is monitored in milliseconds (ms). If in five consecutive samplings... And the trend of change is within the noise judgment area. within, that is If there is no directional drift, the controller will maintain the current power.

[0079] Suppose that the system detects a continuous temperature increase trend in the direction of the continuous offset during a certain period, and in the continuous Each sampling resulted in a temperature increase of approximately 0.06°C, leading to a total offset of approximately 0.3°C. C = 0.15 J / °C, exceeding [the specified value]. The controller identifies a valid trend deviation and immediately calculates the power required for compensation. =0.8×(0.3 / 0.3)×0.15=0.12W, corresponding to a pulse width adjustment Δtw=-1.0ms. The pulse width is adjusted in 0.1ms increments based on the PWM modulation accuracy of the power supply module, which is the standard modulation accuracy for industrial-grade constant temperature circuits. By reducing the pulse width from 8.0ms to 7.0ms and adjusting the duty cycle to 47%, the average power is reduced by approximately 8%, and the temperature remains within two... The temperature fell back to 740.1°C during the cycle and then stabilized. [ Within the ]°C range.

[0080] During this process, the control system dynamically switches the temperature rise prediction model. Initially, linear regression is used to reduce the computational burden. Later, after detecting enhanced nonlinearity in the sensor's thermal response and non-uniform temperature rise rate, a neural network model based on a three-layer perceptron is automatically activated to refit the prediction path, enhancing the control accuracy during the high-temperature stage. The error rate was reduced from ±0.07°C to ±0.02°C, further improving the isothermal stability.

[0081] Example 2: Combined with Appendix Figure 6 As shown, in the implementation of the present invention, the target operating temperature of a heating sensor is set to 750°C, the initial ambient temperature of the sensor is 28°C, and the system controller executes a stepped heating strategy after power-on, specifically according to five steps (S1) to (S5).

[0082] First, in step (S1), the circuit controller automatically generates a step table corresponding to the power-temperature rise increment based on the initial conditions such as the current ambient temperature, sensor packaging structure, and material heat capacity, combined with the historical sensor heating response model. This table divides the heating process into several temperature segments, each with a target temperature rise value, such as 5°C, and provides the appropriate power level for that segment, such as a duty cycle of 50%, 60%, or 70%, along with the predicted heating time. Based on the actual measured sensor response characteristics, the current version of this table contains 12 temperature steps, each with a temperature span of 5°C, an initial power level set to a 50% duty cycle, a pulse period of 100ms, and an initial pulse on-time of 5ms. Then, in step (S2), the controller reads the step table... The parameters for the first temperature rise stage, i.e., rising from 28°C to 33°C, correspond to a power level of 50%. Pulse on / off control is then executed, cycling in 100ms increments. After each pulse on / off cycle, the temperature sensor value is immediately read and the heating rate is recorded. After one complete cycle, the pulse width for the next cycle is updated. If the average heating rate of the first three cycles is 0.047°C / s, slightly lower than the current step prediction range of 0.05°C / s to 0.06°C / s, then in step (S3), the controller automatically determines that the heating is insufficient. In the next pulse cycle, the on-time is increased from 5ms to 6.2ms, and the duty cycle is increased to approximately 62%. If the heating rate is measured again to be 0.052°C / s, entering the prediction range, then the controller maintains a stable output of this parameter.

[0083] If the detected heating rate in a certain cycle is 0.065°C / s, exceeding the predicted upper limit, the power-on time is automatically shortened to 4.8ms to eliminate the temperature deviation caused by excessively rapid heating. After completing the current step, i.e., when the temperature reaches 33°C, in step (S4), the controller immediately switches to the next step in the step lookup table, corresponding to a target temperature of 38°C. The power level is increased to 60%, and the next round of pulse heating begins, repeating the aforementioned pulse adjustment and rate analysis process. This cycle is executed in a stepwise manner, with the control system advancing the temperature rise curve segment by segment in a step-like manner. After each step temperature is reached, the system automatically and smoothly transitions to the next segment. The entire heating process completes the temperature rise from 28°C to 745°C within 8.5 minutes.

[0084] Next, proceed to step (S5), where after the system temperature reaches the target constant temperature range of 745°C to 755°C, the controller automatically extends the pulse period to 300ms and reduces the adjustment frequency, sampling temperature change data only once per cycle and analyzing the trend. If the temperature remains stable within ±0.1°C for two consecutive cycles and the rate of change is lower than [the specified value], [the system will be considered stable]. If the current power remains unchanged, then the overall power will remain unchanged; if a slow decrease in temperature is detected, and the overall deviation exceeds [a certain threshold], then [the power will remain unchanged]. In the next cycle, the controller fine-tunes the power output, such as increasing the power-on time by 0.6ms. If the temperature returns to within ±0.05°C of the reference temperature, the configuration is maintained. This control mechanism forms dynamic constant temperature control through a step table—real-time adjustment—error feedback closed loop. Specifically, in this example, the controller regenerates the step lookup table each time the system is started, ensuring that even if the sensor batch is replaced with a resistance deviation of ±5% and a thermal capacity error of ±8%, or if the operating environment changes, such as the ambient temperature changing from 28°C to 35°C, the system can still automatically adapt and maintain the consistency of the heating process and control accuracy.

[0085] In addition, the temperature rise error judgment criterion in step (S3) is not based on the static temperature offset, but on the temperature rise rate range. The upper and lower limits are preset within the current step, such as 0.05°C / s to 0.06°C / s. The dynamic thermal response is the core judgment criterion to improve the real-time control.

[0086] Example 3: Combined with Appendix Figure 7 As shown, in a certain type of industrial flue gas monitoring equipment, the core sensor is a heated metal oxide gas sensor with a target operating temperature range of 730°C to 750°C. After the control system is powered on, it enters the initialization phase, and the circuit controller immediately executes the step lookup table generation method proposed in this invention. First, in the trial heating and data acquisition phase of step 1, the controller controls the heating element to preheat for a short time with extremely low power, and the pulse period is set to... The system operates for 3 milliseconds (ms), with a power-on width of 3 ms and a duty cycle of 3.75%, executing for a total of 30 cycles. The initial ambient temperature recorded by the system is 27.5°C. During this period, temperature rise data is sampled at 10 ms intervals to obtain a temperature change sequence. Analysis shows that the average temperature rise rate in the first 10 seconds is 0.038°C / s, the initial thermal response delay is 0.7 seconds, and the thermal inertial response is significantly nonlinear. Step 2, control function modeling and step table generation, is then performed. Based on this thermal response data, the controller initiates the embedded function modeling algorithm, selects the nonlinear fitting path, and substitutes the function:

[0087]

[0088] The range of coefficient values ​​is dynamically adjusted during the fitting process, ultimately yielding the following effective range: These coefficients ensure physical consistency in actual power control across different temperature zones; for example, in the first temperature control step, from 27.5°C to 32.5°C, the final fitted parameters are... The controller then automatically generates a step reference table with a total of 15 temperature control steps. The temperature rise increment for each step is set to 5°C, and the corresponding power level and pulse parameters are preset according to the fitted model.

[0089] After entering step 3, the gradual heating and real-time correction mechanism, the controller controls the heating element to heat up step by step according to the lookup table. For example, the target temperature range for the second step is 32.5°C to 37.5°C, the initial pulse width is 6.8ms, the duty cycle is 8.5%, and the sampling data shows that the current temperature rise rate is 0.061°C / s, while the model prediction range is 0.045°C / s to 0.055°C / s. Therefore, the temperature rise is judged to be too high, and the controller immediately shortens the pulse width to 6.1ms, reduces the duty cycle to 7.6%, and inserts an 8ms delay gap to suppress excessively rapid heating. Subsequently, if external airflow disturbance is detected causing the heating rate to suddenly drop to 0.031°C / s, the system automatically inserts a temporary step, adding a compensation power step based on the current step. The duty cycle is increased to 10%, and after 3 cycles, the temperature rise trend returns to normal. The model then updates the response weights for this step.

[0090] After the temperature reaches 735°C, the system enters step 4, the steady-state phase, and the isothermal maintenance phase. At this point, the controller activates the dominant role of the exponential term in the control function, i.e., through... The decreasing trend of buffer energy injection gradually shortens the pulse width from 6.0ms to 4.0ms, reduces the duty cycle to 5.3%, maintains a period of 250ms, and enters the steady-state isothermal judgment stage; if the sampling shows that the temperature fluctuates around 742.2°C and If the system determines that the steady-state condition is met, it will no longer aggressively adjust the power, but only maintain fine-tuning; if the continuous offset trend within a certain period is greater than 0.2°C, the function fitting module will be reactivated to reconstruct the local temperature step. value.

[0091] When the device restarts due to fluctuations in the urban power grid or is powered on for the first time, the controller, to adapt to the current climate environment (e.g., ambient temperature 27.3°C, relative humidity 62%, wind speed approximately 1.2 m / s), will immediately enter the experimental preheating phase designed in this invention. To avoid thermal shock to the ceramic encapsulation and sensitive film layer of the sensor, the controller's start-up duty cycle is set to only 4.5%, lower than the specified upper limit of 5%, with a pulse period of 90 ms, a power-on width of 4.05 ms, and a total duration of 1.8 seconds. During this period, the controller collects temperature feedback data at 10 ms intervals, records the first 180 samples, and calculates that the temperature rise trend is non-linear and gradually increasing, with an initial response delay of approximately 0.4 seconds and a stable temperature rise rate of approximately 0.039°C / s, slightly lower than the typical value of 0.05°C / s in the standard model, indicating that the heat capacity of this batch of sensors is slightly higher than the typical parameters. The controller calls the local fitting module to dynamically generate a temperature rise control model based on the experimental heating data, using a non-linear function. Fitting is performed to form a thermal response control function, and the core coefficients of the fitted function are as follows: the parameters of the first temperature control step are set as follows: A real-time temperature rise reference table containing 15 temperature control steps was constructed based on this. Each step corresponds to a fixed temperature rise increment of 4°C, a basic power output level (e.g., starting with a duty cycle of 6.0%), and a pulse control rhythm (e.g., an initial cycle of 120ms and a power-on time of 6.8ms). Each step control block in this step table structure contains three sets of control parameters: the desired temperature rise (e.g., 30°C to 34°C), the pulse width and cycle frequency used, and the target duration window (e.g., completed within 6 to 9 seconds). The table content is reconstructed based on the current preheating sampling data each time power is applied, ensuring adaptation to individual sensor differences and external disturbances.

[0092] During the gradual temperature rise control process, the controller executes the temperature rise operations of each step according to the table. For example, in the fourth step, from 39°C to 43°C, the pulse width used is 9.2ms, the duty cycle is 7.7%, and the period is maintained at 120ms. The actual temperature rise rate is sampled as 0.062°C / s, slightly exceeding the prediction range of 0.05 to 0.06°C / s. Therefore, the controller adjusts the power-on time to 8.6ms and lengthens the period interval to 135ms in subsequent cycles to bring the heat input back to the model requirements. During the temperature rise process, if a sudden increase in wind speed or enhanced heat dissipation from the casing causes the temperature rise trend to drop to 0.031°C / s, the controller will immediately perform a local lookup table adjustment, dynamically inserting a fine-tuning sub-step in the current step. For example, the target temperature rise of 4°C is split into two 2°C micro-segments, and new power levels are matched sequentially. For example, the duty cycle of 10.8% is divided into two stages: 9.2% and 11.4%, achieving segment-by-segment compensation.

[0093] As the temperature gradually rises above 720°C, the controller gradually enters the tail-stage step control range. Exponential suppression term in The system automatically reduces the duty cycle and extends the cycle frequency to lower the temperature rise rate to below 0.028°C / s, preventing temperature overshoot due to excessive thermal inertia. Within the final isothermal range of approximately 749.0°C to 750.5°C, the controller increases the cycle to 350ms and controls the pulse width between 6.5ms and 7.2ms, based on the fluctuation trend. Determine whether to enable micro-compensation power or postpone the adjustment.

[0094] Experiments show that even when the ambient temperature fluctuates by more than 8°C, the relative humidity difference reaches 20%, and the sensor thermal capacity deviation reaches 10% before startup, the proposed exploratory preheating-real-time table building-multi-parameter step control can still stably raise the system temperature to the set target constant temperature zone within 12 minutes, ultimately maintaining the temperature fluctuation range within ±0.3°C and the dynamic response delay less than 0.6 seconds. This verifies the adaptive heating modeling capability and constant temperature control stability of the method under multiple operating conditions, and it is particularly suitable for the batch deployment of industrial embedded sensing platforms and high environmental noise scenarios.

[0095] Upon initial power-on, the system enters an exploratory heating phase and collects the initial thermal response behavior of the sensor, including three key indicators: First, the rate of temperature change, i.e., the temperature rise gradient per unit time, which was measured to be 0.039°C / s in this startup; second, the temperature rise delay time, i.e. the time when the sensor first shows temperature rise feedback after the controller applies power, which was measured to be 0.45 seconds; and third, the result of the heat transfer inertia judgment, which was calculated by the system based on the slope change trend of the temperature rise curve in the first three seconds and judged to be a slow-change response, belonging to the subcategory of this model with relatively high heat capacity.

[0096] Because this device supports scanning the batch identification information pre-written in the EEPROM of the sensor chip, the controller automatically identifies that the sensor belongs to batch B07. Combining this with the historical heating response database for batch B07 stored locally, the system automatically imports the established temperature rise data model for this batch, which contains 12 sets of response curves and empirical parameters for each step. The controller performs similarity fitting between the current measured data and historical characteristics (with the average deviation controlled within ±6%), and corrects the fitting function based on the matching results. The initial parameters in the original preheating model will be used to... Fine-tuning This optimizes the power output of the first segment of the lookup table, ensuring that the temperature control more closely matches the batch characteristics. The generated step lookup table still maintains 15 steps, with a temperature rise increment of 4°C for each step. The corresponding power output parameters are derived according to the corrected model. In the first four steps, the duty cycles are 5.2%, 6.1%, 7.4%, and 8.3%, respectively, and the period frequency remains unchanged at 120ms.

[0097] After entering the temperature rise control process, the controller collects the current temperature and temperature rise trend in real time. During the operation of the third stage (35°C to 39°C), the controller continuously detected that the temperature rise rate decreased from 0.051°C / s to 0.032°C / s, and there were three instances of trend consistency deviation between adjacent data points. The system judged this to be triggered by an external disturbance event, possibly due to air convection interference caused by the opening of doors and windows. This disturbance lasted for more than 280ms, with a cumulative disturbance amplitude of 0.26°C. The controller then entered the adaptive disturbance identification and response stage. Without reconstructing the entire step structure, the controller dynamically adjusted the power output parameters of the current stage according to the method of this invention, increasing the pulse width of the current stage from 9.4ms to 10.6ms, adjusting the duty cycle from 7.4% to 8.8%, and reducing the cycle interval of this stage by 10ms to increase the short-term power density to counteract the heat loss caused by the disturbance. If the disturbance is not alleviated within the next two cycles, the controller further judges the trend of the disturbance amplitude stabilization and allows early entry into the fourth stage, where it is activated. This strategy aims to improve overall heating efficiency and ultimately maintain the system's temperature rise rhythm without disruption, successfully completing the target temperature rise within 6.2 seconds. It avoids oscillations caused by repeated pullbacks due to minor disturbances and also prevents heating from becoming discontinuous.

[0098] Example 4: Combined with Appendix Figure 8 As shown, in a high-temperature semiconductor sensor device used for monitoring gas in an industrial furnace, the sensor's normal operating range is 735°C to 750°C. To ensure detection accuracy, the control system needs to quickly and stably enter the constant temperature maintenance phase after the heating process ends. When the device reaches 749.2°C, the circuit controller automatically detects that the current temperature is close to the target value of 750°C, and the temperature rise rate has dropped to 0.031°C / s, which is below the system's set stable entry threshold. The system then activates the constant temperature maintenance control method proposed in this invention. First, it switches from the short-cycle pulse mode (120ms pulse period, approximately 9.6% duty cycle) used during the heating phase to a long-cycle pulse mode, increasing the pulse period to 300ms, setting the initial pulse width to 6.5ms, and adjusting the duty cycle to 2.17%. The controller collects temperature data every 300ms at fixed time intervals, continuously acquiring 12 sets of data to form a temperature trend sampling sequence. This sequence is then subjected to trend classification analysis: if the calculated average fluctuation amplitude is <±0.2°C, the maximum deviation does not exceed ±0.3°C, and the temperature change exhibits alternating small fluctuations, the system identifies it as a stable fluctuation trend. At this point, the controller determines that the heating state has reached thermal equilibrium, performs no adjustment operations, continues to maintain the current power output structure, and continues into the next sampling cycle.

[0099] Upon entering the next sampling cycle, the controller detected a slow downward trend in five consecutive temperature samples, with an average decrease rate of 0.027°C / s and a total offset of 0.15°C. Although still within the set safe temperature range, the system's offset confirmation threshold was set at five consecutive samples with consistent direction. Based on the continuous offset judgment rule set according to the invention, the controller decided to trigger a small power compensation operation, increasing the pulse width from 6.5ms to 7.2ms and the duty cycle to 2.4%, while maintaining the period unchanged. After adjustment, the temperature curve stabilized and rebounded after two cycles, maintaining a range of 749.6–750.1°C. The controller confirmed that the target had been recovered and automatically returned to observation mode. After approximately two minutes of operation during the 400th cycle, the system suddenly detected a temperature fluctuation of +0.75°C, far exceeding the average fluctuation limit. However, the subsequent four data points returned to the original level, and the trend no longer continued to deviate. Based on the sudden disturbance identification mechanism of this invention, the system determined it to be an external thermal disturbance, such as instantaneous air convection. Even though the deviation was significant, because it did not form a sustained trend, the controller refused to adjust the power, avoiding system oscillations caused by misadjustment. Experimental statistics show that, under the control of the method of this invention, the average temperature deviation from the target value during the isothermal phase of the sensor was ±0.18°C, and the peak fluctuation was less than 0.35°C, significantly better than the ±0.6°C fluctuation performance under the traditional control strategy. Furthermore, the adaptive adjustment frequency was reduced by 32%, effectively extending the sensor's lifespan and stabilizing the measurement output. This verifies that the isothermal stabilization control method of this invention has accuracy, anti-interference capabilities, and intelligent adaptability in complex environments.

[0100] After the temperature rises to 749.3°C and enters the target constant temperature zone of 750±0.5°C, the controller automatically starts the constant temperature maintenance control mechanism. According to the design requirements of this invention, the heating control is first switched from short cycle mode to long cycle mode. Specifically, the pulse period of 10ms used in the heating stage is switched to long cycle control in the constant temperature stage. At this time, the cycle is set to 300ms, the pulse width is 7ms, and the duty cycle is about 2.3%. This long cycle setting is within the range of 50ms to 500ms specified in this invention, which can effectively reduce power supply noise and thermal stress caused by frequent heating, while providing a sufficiently slow thermal response window to facilitate trend identification and analysis.

[0101] During the constant temperature maintenance period, the controller samples temperature data once per cycle and continuously acquires at least 5 data points to form a trend sequence. It then executes trend analysis logic, which includes two core dimensions for judgment: first, the direction of temperature change, i.e., whether it is a continuous increase or decrease; and second, the magnitude of change, i.e., whether each deviation exceeds a set threshold. The system comprehensively judges the current trend type. During the initial trend analysis, the five data points were 749.3°C, 749.4°C, 749.2°C, 749.5°C, and 749.1°C, with a maximum deviation of ±0.4°C but no continuous deviations and inconsistent directions. The controller judged this as a normal fluctuation trend, and the system maintained the current power output. In the second round of analysis, the sampled data were 749.4°C, 749.5°C, 749.6°C, 749.7°C, and 749.9°C, showing five consecutive increases. The average single deviation was approximately 0.12°C, with a maximum deviation of 0.6°C, exceeding the deviation threshold. And continuously exceed the set value In each cycle, the system identifies a continuous offset trend (b) and triggers a small power compensation. The pulse width increases from 7ms to 7.8ms, the duty cycle increases to 2.6%, and the cycle remains unchanged. After this compensation action is completed, the controller, according to the requirements of the invention, forces a one-cycle interval without a second adjustment, i.e., the compensation interval window is 300ms, to avoid over-adjustment caused by sampling errors or hysteresis response. The system then continues sampling and finds that the compensated temperature data drops to 750.0°C, 749.8°C, 749.7°C, 749.6°C, and 749.7°C, with the direction of change tending to balance. The system returns to a normal fluctuation state and maintains the current output. At a certain moment during operation, the controller detected a sudden increase in temperature from 749.3°C to 750.8°C during cycle 213. However, the subsequent four samples were 750.2°C, 749.9°C, 750.0°C, and 749.8°C, respectively, with the temperature quickly returning to near the target value and not showing a consistent trend. The controller determined this anomaly to be a c. interference trend, inferring that it might be due to flue disturbance or a momentary voltage spike in the heating element. Based on this, the system rejected any power adjustment operation to ensure the overall temperature control's anti-interference stability.

[0102] Experimental data show that, after adopting the trend recognition classification and hierarchical response mechanism proposed in this invention, during the 30-minute constant temperature phase, the system triggered compensation actions 3 times, rejected interference responses 5 times, and made normal fluctuation judgments 28 times. The overall temperature fluctuation was controlled within ±0.23°C. Compared with the traditional dead-zone control structure, this reduced the number of over-adjustments by 38% and the average adjustment power change by 27%, significantly improving the system's temperature control accuracy and stability. It also effectively suppressed unnecessary control actions caused by minor disturbances, verifying that the long-cycle stability control method based on trend judgment in this invention has engineering effectiveness and wide adaptability in high-stability demand scenarios.

[0103] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A control method for a temperature-controlled circuit of a sensor, applied to a temperature-controlled circuit including a heating element, a temperature acquisition unit, and a power supply module, wherein the temperature-controlled circuit is driven by a circuit controller, characterized in that: Includes the following steps, The circuit controller acquires information including the initial ambient temperature. Based on the sensor's structural parameters and historical heating response curves, a temperature rise prediction model is established, and the output is displayed at multiple power levels. The rate of temperature rise at the following conditions ; During the heating start-up phase, the circuit controller selects the target power level according to the prediction model and controls the heating element to operate in a first cycle. The pulse mode alternates on and off to achieve stepped heating; The circuit controller in the second cycle Internally based on real-time temperature Fine-tuning the power level with the temperature rise prediction model Calculate the rate of temperature change ΔT / Δt; when |ΔT / Δt|≤ Record the current temperature. As the first reference temperature point ; Near the first reference temperature point, the circuit controller in each control cycle Internal analysis of the temperature change trend in the previous cycle And the predicted temperature rise Δt is obtained by combining the prediction delay δt. If Δ > Then reduce or pause power output, if <– This increases power output; When | – |< And when the fluctuation trend falls within the noise judgment region N, maintain the current power output; when | – |≥ When sampling is continuously offset n times in the same direction, a compensation power ΔP is applied to adjust the real-time temperature. Return to the first reference temperature point ± The noise judgment area N refers to the real-time temperature. Compared with the first reference temperature point The absolute value of the deviation is less than Furthermore, no single-direction trend of change was formed within multiple consecutive sampling periods.

2. The control method for the constant temperature circuit of the sensor according to claim 1, characterized in that: First cycle The pulse width is 1ms to 20ms, the second cycle The duration is 50ms to 500ms; where... The pulse width is set to 0.05°C / s, 0.1°C, and 0.2°C respectively. Within each control cycle, the pulse width is adjusted in steps of 0.1ms based on the deviation between the current temperature and the target temperature until the temperature change rate falls within the preset threshold range. The temperature rise prediction model adopts first-order linear regression or nonlinear fitting based on neural networks.

3. The control method for the constant temperature circuit of the sensor according to claim 1, characterized in that: The method for achieving stepped heating includes the following steps. (S1) The circuit controller generates a step comparison table corresponding to different power levels and target temperature rise increments; (S2) Select the first power level according to the step reference table, control the heating element to alternately turn on and off in the pulse mode of the first cycle, collect the current temperature after each pulse and update the next pulse width or duty cycle, so that the temperature rise curve rises in a step-like manner. (S3) If the real-time temperature rise rate of this step is detected to be lower than the predicted range, extend the power-on time of the next pulse; if it is higher than the predicted range, shorten the power-on time to eliminate the temperature rise error. (S4) When the target temperature rise increment corresponding to this step reaches the preset value, the circuit controller immediately switches to the next power level in the step lookup table and repeats steps (S2) to (S3) until the temperature approaches the target constant temperature zone. (S5) After entering the target constant temperature zone, the circuit controller will switch the pulse period between long and short, and determine whether to perform small-amplitude power compensation based on the temperature trend obtained by continuous sampling in order to maintain temperature stability.

4. The control method for the constant temperature circuit of the sensor according to claim 3, characterized in that: The step reference table is regenerated based on real-time initial conditions each time it is started, in order to adapt to differences in sensor batches and changes in the external environment; the temperature rise error determination in step (S3) is based on the temperature change rate range.

5. The control method for the constant temperature circuit of the sensor according to claim 4, characterized in that: The method for generating the step lookup table includes: The circuit controller controls the heating element to perform short-term exploratory heating at low power and continuously collects temperature change data to analyze the current thermal response behavior of the sensor. The circuit controller dynamically generates a step lookup table for temperature control based on the temperature rise rate, response delay, and thermal inertia during preheating. The lookup table includes temperature increment steps, with each step corresponding to a single heating power level and pulse control parameters. During the heating process, the circuit controller adjusts the power output and pulse rhythm of the heating element step by step according to the step reference table to ensure a stable temperature rise. If a sudden change in environment or a temperature rise deviation is detected during the heating process, the circuit controller will revise the step lookup table or adjust the current step parameters to adapt to external disturbances or device differences.

6. The control method for the constant temperature circuit of the sensor according to claim 5, characterized in that: The power duty cycle of the exploratory preheating is 1% to 5%, with the lower limit of 1% being the minimum duty cycle to ensure the collection of effective temperature change data, and the duration is 1 to 2 seconds; the step reference table is generated in real time after each power-on start-up of the equipment; the power output corresponding to each step in the reference table includes the heating pulse width, cycle frequency, and expected temperature rise increment.

7. The control method for the constant temperature circuit of the sensor according to claim 6, characterized in that: The thermal response behavior includes one of the following indicators: temperature change rate, heating delay time, and heat transfer inertia judgment result; when the batch identifier of the sensor is available, the circuit controller combines the historical heating response data of the batch to perform model fitting and comparison table optimization; the adaptive adjustment process includes identifying the disturbance trend in the heating process and adjusting the current step power or switching to the next step in advance according to the disturbance amplitude and duration.

8. The control method for the constant temperature circuit of the sensor according to claim 3, characterized in that: The method for maintaining temperature stability includes... When the temperature approaches the target constant temperature value, the circuit controller switches the heating control mode from short-cycle pulse mode to long-cycle pulse mode. During the constant temperature maintenance phase, the circuit controller continuously collects multiple sets of temperature data at fixed time intervals and analyzes the temperature change trend type based on the sampling results, including stable fluctuations, continuous deviations, or sudden changes. When the trend analysis result shows that the temperature is near the target value and the fluctuation is within the set threshold range, the current power output is maintained without adjustment. When the trend analysis result shows that the temperature continues to shift upward or downward and continues to exceed a preset number of times, the circuit controller triggers a small power compensation, which includes appropriately increasing or decreasing the pulse width, duty cycle, or power output within the cycle. When the trend analysis results show a sudden abnormal fluctuation in temperature but no sustained deviation, the circuit controller judges it as an external disturbance and refuses to adjust the power.

9. The control method for the constant temperature circuit of the sensor according to claim 8, characterized in that: The short period is 1 to 20 milliseconds, and the long period is 50 to 500 milliseconds; the circuit controller uses no less than 5 temperature sampling points for trend analysis, and judges the current trend by combining the direction and magnitude of temperature changes; the trigger threshold for the small compensation is when the temperature deviates from the target value by more than ±ε and continues for more than n consecutive sampling periods.

10. The control method for the constant temperature circuit of the sensor according to claim 9, characterized in that: The compensation power adjustment actions are spaced one sampling period apart to prevent over-response; the trend type includes one of the following three: a. Normal fluctuation trend: The temperature does not deviate continuously within the preset range; b. Continuous deviation trend: The temperature continuously deviates from the target value and changes in the same direction; c. Disturbance trend: Single or occasional sharp changes that do not form a sustained trend.

Images