Intelligent oven temperature regulation and control system and method based on self-adaptive control method

Abstract

The invention discloses an intelligent oven temperature regulation and control system and method based on a self-adaptive control method, and the method comprises the steps: collecting first temperature data in an oven based on sensors, and obtaining a redundancy check temperature value through a redundancy check method based on multiple temperature sensors; calculating a current temperature drift error value according to the redundancy check temperature value, inputting the temperature drift error value and historical drift trend data into a temperature drift prediction model, and outputting a predicted temperature drift compensation value; fusing the redundancy check temperature value with the predicted temperature drift compensation value to generate a first fused temperature value; calculating a temperature stability index in the oven based on the first fusion temperature value; based on the temperature stability index and a preset temperature fluctuation target range, the heating power of the oven is adjusted through an adaptive frequency heating control algorithm; and second temperature data in the oven are collected again, and cross validation is carried out on the second temperature data and the first fusion temperature value so as to correct the historical drift trend data.

Description

Technical Field

[0001] The present invention relates to the technical field of oven temperature control, and in particular to an intelligent oven temperature control system and method based on an adaptive control method. Background Art

[0002] Ovens are widely used in food processing, industrial heat treatment, laboratory testing and other fields. The accuracy and stability of temperature control directly affect product quality and process reliability. However, existing oven temperature control technology has the following shortcomings: 1. Most ovens use a single or small number of temperature sensors, which are susceptible to sensor drift or damage, resulting in a decrease in long-term measurement accuracy.

[0003] 2. Lack of effective temperature drift prediction methods makes it impossible to accurately compensate for temperature changes in advance, resulting in long-term temperature control deviations.

[0004] 3. Traditional methods use simple linear fusion of measurement data and predicted compensation data, which is difficult to adapt to real-time dynamic changes and has insufficient fusion accuracy.

[0005] 4. The heating control strategy is crude, lacks a real-time adaptive adjustment mechanism, and the temperature fluctuates greatly.

[0006] 5. An effective long-term feedback correction mechanism has not been established, resulting in a gradual decrease in control accuracy as the operating time increases.

[0007] In view of this, the present invention proposes an intelligent oven temperature control system and method based on an adaptive control method. Summary of the Invention

[0008] The purpose of the present invention is to provide an intelligent oven temperature control system and method based on an adaptive control method, which achieves long-term stability and accuracy of temperature control.

[0009] In a first aspect, the present invention provides a temperature control method for an intelligent oven based on an adaptive control method, comprising the following steps: S101: collecting first temperature data inside the oven based on a sensor, and obtaining a redundant verification temperature value using a redundant verification method based on multiple temperature sensors; S102: Calculating a current temperature drift error value based on the redundant check temperature value, inputting the temperature drift error value and historical drift trend data into a temperature drift prediction model, and outputting a predicted temperature drift compensation value; S103: Fusing the redundant check temperature value and the predicted temperature drift compensation value to generate a first fused temperature value; and calculating a temperature stability index inside the oven based on the first fused temperature value. S104: Based on the temperature stability index and the preset temperature fluctuation target range, the heating power of the oven is adjusted by an adaptive frequency heating control algorithm; S105: When the oven enters the next heating cycle, the second temperature data inside the oven is recollected according to the heating power adjusted in S104, and the second temperature data is cross-validated with the first fusion temperature value to correct the historical drift trend data.

[0010] As a preferred technical solution of the first aspect of the present invention, the logic for obtaining the redundant check temperature value is: Arrange multiple temperature sensors in the oven to collect first temperature data, use the spatial coordinates of the sensor measurement points as discrete spatial temperature points, construct a three-dimensional spatial distribution model of the temperature field inside the oven, and extract time-series temperature spatial data; Input the time series temperature spatial data into the pre-built spatial interpolation neural network model to calculate the spatial continuous temperature field data; calculate the deviation between the time series temperature spatial data and the spatial continuous temperature field data in real time, compare and analyze the deviation with the dynamic adaptive threshold, determine the abnormal measurement value and automatically correct it to form the corrected temperature field data; Extract features from the corrected temperature field data to obtain spatial temperature field features, including spatial temperature entropy, spatial temperature gradient mean, and temperature field stability index; combine the spatial temperature field features with the corrected temperature field data through a nonlinear adaptive fusion algorithm to determine the nonlinear fusion weight of each region; The fusion weights are nonlinearly weighted fused with the temperature prediction values ​​of key areas to output redundant verification temperature values.

[0011] As a preferred technical solution of the first aspect of the present invention, the application logic of the neural network model is: The structure of the neural network model adopts a 3D convolutional neural network or a three-dimensional space Transformer network; the model input is the real-time spatial position coordinates and temperature measurement values ​​of each sensor; the model output is the continuous temperature prediction value of each point in the complete three-dimensional space; Based on a large amount of historical time-series temperature spatial data, a deep neural network model is trained to fit discrete spatial point temperature data into a continuous three-dimensional spatial temperature field; based on the spatial continuous temperature field data output by the neural network in real time; If the deviation between the time series temperature space data and the time series temperature space data exceeds the set threshold, the time series temperature space data is automatically corrected through the time series temperature space data to form corrected temperature field data.

[0012] As a preferred technical solution of the first aspect of the present invention, the training process of the temperature drift prediction model includes: An initial temperature drift data sequence is established based on historical temperature drift error data, and the data sequence is input into a time series prediction model based on a recurrent neural network; The temperature drift error value at future moments is predicted through model iteration, and the recurrent neural network parameters are updated based on the actual error value feedback, and the trained temperature drift prediction model is output.

[0013] The method for constructing the initial temperature drift data sequence includes: Extract the temperature drift error value during multiple consecutive heating cycles of the oven operation; Arrange the temperature drift error values ​​in chronological order to form a continuous time series.

[0014] As a preferred technical solution of the first aspect of the present invention, the logic for obtaining the first fusion temperature value is: Establish the Kalman filter model state equation and observation equation, The process noise covariance is dynamically estimated based on the historical drift compensation value error, and the measurement noise covariance is dynamically adjusted based on the real-time redundant verification of temperature measurement accuracy; The fusion weight is adaptively adjusted by Kalman gain to achieve real-time nonlinear fusion of redundant check temperature values ​​and predicted temperature drift compensation values.

[0015] As a preferred technical solution of the first aspect of the present invention, the temperature stability index calculation formula is: ; in: is the i-th measurement value in the real-time temperature measurement sequence; is the average temperature value of the measurement sequence; N is the number of measurements in the sampling window.

[0016] As a preferred technical solution of the first aspect of the present invention, adjusting the heating power of the oven by an adaptive frequency heating control algorithm based on the temperature stability index and the preset temperature fluctuation target range includes: Calculate the deviation between the current temperature stability index and the target range, define a control factor to quantify the temperature deviation, and define a nonlinear control factor based on the deviation; The adjustment range of the heating frequency and power is determined by a nonlinear mapping rule. According to the control factor value, the adjustment range of the heating cycle frequency and power is determined by a predefined nonlinear mapping rule.

[0017] As a preferred technical solution of the first aspect of the present invention, cross-validating the second temperature data with the first fused temperature value to correct the historical drift trend data includes: Calculate the second temperature data and compare it with the first fusion temperature value point by point or as a whole to obtain a temperature difference value; Based on the temperature difference value, determine the average deviation between the current temperature control system prediction and actual operating conditions: The drift trend in the historical data is corrected by the average deviation size to predict the temperature prediction and temperature compensation for the next period.

[0018] In a second aspect, the present invention provides an intelligent oven temperature control system based on an adaptive control method, which, based on the implementation of the first aspect, includes a temperature acquisition and redundancy check module, a temperature drift prediction and compensation module, a temperature fusion and stability assessment module, an adaptive heating control module, and a drift trend feedback correction module: The temperature acquisition and redundancy check module collects the first temperature data inside the oven based on the sensor and obtains the redundant check temperature value using a redundant check method based on multiple temperature sensors; The temperature drift prediction and compensation module calculates the current temperature drift error value based on the redundant check temperature value, inputs the temperature drift error value and historical drift trend data into the temperature drift prediction model, and outputs the predicted temperature drift compensation value; The temperature fusion and stability evaluation module fuses the redundant check temperature value with the predicted temperature drift compensation value to generate a first fusion temperature value; and calculates the temperature stability index inside the oven based on the first fusion temperature value; Adaptive heating control module, which adjusts the oven's heating power through an adaptive frequency heating control algorithm based on the temperature stability index and a preset temperature fluctuation target range; The drift trend feedback correction module re-collects the second temperature data inside the oven according to the heating power adjusted in S104 when the oven enters the next heating cycle, and cross-validates the second temperature data with the first fusion temperature value to correct the historical drift trend data.

[0019] In the above technical solution, the technical effects and advantages provided by the present invention are: The intelligent oven temperature control method proposed in the present invention achieves long-term stable and accurate temperature measurement through multi-sensor redundant verification and spatial temperature field modeling; uses a recurrent neural network to predict temperature drift trends and implements compensation accurately in advance; adopts a nonlinear fusion algorithm to improve fusion accuracy; and adaptively adjusts the heating power in real time based on the temperature stability index to significantly reduce temperature fluctuations; and continuously optimizes the drift trend model through a feedback mechanism to ensure long-term control accuracy and stability to meet the needs of high-end applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0021] Figure 1 This is a framework diagram of the intelligent oven temperature control system based on the adaptive control method of the present invention; Figure 2 This is a flow chart of the intelligent oven temperature control method based on the adaptive control method of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solutions and advantages of the implementation of this application clearer, the technical solutions in the implementation of this application will be described in more detail below in conjunction with the drawings in the implementation of this application.

[0023] In the accompanying drawings, the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The described embodiments are part of the embodiments of the present application, rather than all of the embodiments. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be understood as limitations on the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application. The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0024] Example 1 See also Figure 1 As shown, this embodiment provides an intelligent oven temperature control system based on an adaptive control method, including a temperature acquisition and redundancy check module, a temperature drift prediction and compensation module, a temperature fusion and stability assessment module, an adaptive heating control module, and a drift trend feedback correction module. Each module is connected by wired and / or wireless connections; The temperature acquisition and redundancy verification module collects temperature data from multiple sensors inside the oven in real time, and uses a multi-sensor redundancy verification method to determine the precise redundant verification temperature value through three-dimensional space interpolation, neural network and fuzzy fusion algorithm.

[0025] Specifically, multiple high-precision temperature sensors are installed, their spatial positions are recorded, a 3D-CNN or Transformer model is built, and historical data is used for model training. Dynamic adaptive thresholds are used to determine sensor anomalies in real time and automatically correct them. Spatial entropy, temperature gradient, and stability index features are extracted, the fusion weight is determined, and the fusion temperature is calculated.

[0026] The temperature drift prediction and compensation module establishes a drift trend prediction model based on historical temperature data; calculates the current drift error in real time, predicts future temperature drift trends; and provides predicted drift compensation values.

[0027] Specifically, a time series is established using temperature drift error data collected over a long period of time, and a prediction model is built based on the LSTM recurrent neural network. Real-time measurement and historical trend data are input into the model to obtain compensation values, and the model parameters are updated based on actual measurement value feedback to maintain long-term accuracy.

[0028] The temperature fusion and stability assessment module combines the redundant calibration temperature with the predicted drift compensation value, and calculates the real-time temperature stability index to evaluate the temperature stability inside the oven.

[0029] Specifically, the Kalman filter algorithm is used to fuse redundant temperatures and drift compensation values ​​in real time; the temperature stability index is calculated based on the fused temperature sequence to reflect the real-time temperature fluctuation.

[0030] The adaptive heating control module adaptively adjusts the oven's heating power and frequency based on the real-time temperature stability index and the set temperature fluctuation range, dynamically and accurately controlling the heating process to achieve temperature stability.

[0031] Specifically, the target range of temperature fluctuation is set and the stability index deviation is calculated in real time. The nonlinear control factor is calculated and the adjustment range is determined through predefined nonlinear mapping rules. The heating frequency and power are adjusted in real time by controlling the oven heating device to achieve precise closed-loop control.

[0032] The drift trend feedback correction module measures temperature data in real time when entering a new heating cycle and cross-validates it with the fused temperature value of the previous cycle; it corrects historical drift trend data based on actual differences to further optimize the drift prediction model.

[0033] Specifically, the measured temperature of the new cycle is compared with the fused temperature of the previous cycle to calculate the deviation; the historical drift trend model is adjusted using algorithms such as the least squares method, and trend data is updated in real time to ensure that the prediction and compensation of the next cycle have higher accuracy and stability.

[0034] After the system is started, the sensor array collects temperature in real time, constructs a spatially continuous temperature field, and calculates redundant check temperature. Based on the historical trend model, it predicts future temperature drift error values ​​in real time and determines the drift compensation value. Kalman filtering is used to fuse the predicted temperature compensation and redundant temperature, and the temperature stability index is calculated in real time. The adaptive algorithm dynamically adjusts the frequency and power of the heating device according to the stability index to stabilize the temperature. After entering a new cycle, the system re-collects the actual temperature and cross-validates it, using difference feedback to update the historical drift trend data. The above process is repeated continuously to continuously optimize and stabilize the oven temperature control.

[0035] Example 2 Based on the embodiment, this embodiment provides a temperature control method for an intelligent oven based on an adaptive control method, including the following steps: S101: collecting first temperature data inside the oven based on a sensor, and obtaining a redundant verification temperature value using a redundant verification method based on multiple temperature sensors; Specifically, the original temperature measurement data collected in real time by multiple temperature sensors in the oven is marked as the first temperature data, providing the basic input for temperature control and stability evaluation, and used for subsequent verification and compensation processing. However, the data collected by multiple sensors have certain data differences, and abnormal measured temperature data is identified for subsequent accurate correction.

[0036] In this way, redundant and verified temperature values ​​are obtained after redundancy and verification calculations are performed on multiple sensor data. These values ​​are closer to the actual temperature than the values ​​measured by a single sensor. Therefore, the temperature measurement is no longer regarded as a single value, but as a distributed temperature field inside the oven. By constructing a virtual space field model, spatial correlation nonlinear fusion is achieved to obtain stable redundant and verified temperature values, which can be used for long-term stable temperature measurement to suppress the errors caused by long-term drift of the sensor, avoid the degradation of the overall system performance due to aging or drift of a single sensor, and ensure the accuracy and effectiveness of the compensation measures.

[0037] Specifically, the logic for obtaining the redundant verification temperature value is as follows: Multiple temperature sensors are placed inside the oven. The temperature data at the sensor measurement points are used as discrete spatial temperature points to construct a three-dimensional spatial distribution model of the oven's internal temperature field and extract time-series temperature spatial data. Specifically, multiple high-precision temperature sensors are precisely installed at different locations inside the oven, and the precise spatial coordinates (X, Y, Z) of each sensor are recorded to form a sensor spatial position coordinate matrix. The temperature data measured by each sensor is collected in real time, and the measured data and spatial coordinates are matched to form time-series temperature spatial data.

[0038] Input the time series temperature spatial data into the pre-built spatial interpolation neural network model to calculate the spatial continuous temperature field data; then use the neural network to adaptively correct the possible drift and abnormal points of the sensor to form the corrected temperature field data; Specifically, based on a large amount of historical time-series temperature spatial data, a deep neural network model is trained and constructed that can fit discrete spatial point temperature data into a continuous three-dimensional spatial temperature field; based on the spatial continuous temperature field data output in real time by the neural network, if the deviation between a sensor measurement value and the spatial model prediction value exceeds the set threshold, the threshold here is determined using a dynamic adaptive threshold algorithm, which specifically uses the statistical distribution of the deviation between the sensor measurement value and the model prediction value within a certain historical period (such as the past 100 moments); calculates the standard deviation and average value of the historical deviation sequence; then calculates the dynamic threshold, and makes outlier judgments in real time based on the dynamic threshold, automatically replacing or correcting the abnormal measurement value with the model prediction value to form corrected temperature field data to maintain the continuity of the temperature field.

[0039] The neural network model uses a 3D convolutional neural network (3D-CNN) or a three-dimensional Transformer network. The model input is the real-time spatial position coordinates and temperature measurement values ​​of each sensor. The model output is the continuous temperature prediction value of each point in the complete three-dimensional space (inside the oven). Training method: Collect data from ovens that have been running stably for a long time to construct a training set. The data is divided into a training set and a validation set, and model parameters are iteratively optimized. The Adam optimization algorithm is used, with the minimum prediction error (e.g., mean square error (MSE)) as the loss function optimization target.

[0040] Feature extraction is performed on the corrected temperature field data to obtain spatial temperature field features, which include spatial temperature entropy, spatial temperature gradient mean and temperature field stability index; the spatial temperature field features and the corrected temperature field data are fused through a nonlinear adaptive algorithm to output redundant verification temperature values.

[0041] The temperature space entropy reflects the uniformity of the temperature field; the formula is: ; ; in, is the temperature value of the i-th spatial region, and N is the total number of regions.

[0042] The spatial temperature gradient distribution reflects the trend of spatial temperature change; the formula is: ; in, is the temperature value of the adjacent area of ​​the ith spatial position, and M is the total number of temperature gradient calculation points.

[0043] The field stability index indicates the overall temperature field fluctuation state; the formula is: ; in, is the average temperature of the overall temperature field at time t, and K is the length of the calculation time window.

[0044] Specifically, a nonlinear mapping relationship between the spatial temperature field characteristics and the fusion weight is established, taking the establishment of rules of the fuzzy logic system (Sugeno model) as an example: Input conditions Fusion weight setting example If the spatial entropy is low and the stability index is high The center temperature has a high weight, while the corner temperature has a low weight. If the spatial entropy is high and the stability index is low All regions have their temperature weights evenly distributed If the spatial entropy is medium and the stability index is medium Hotspot areas have moderate weights, while stable areas have higher weights. The following is an example of a complete Sugeno fuzzy rule library: Space temperature entropy Temperature field stability index Fusion weight distribution principle Low high Center area weight = 0.7, edge area weight = 0.3 Low middle Center area weight = 0.6, edge area weight = 0.4 Low Low Center area weight = 0.5, edge area weight = 0.5 middle high Center weight = 0.5, hotspot = 0.3, edge = 0.2 middle middle Hotspot weight = 0.4, stable area = 0.4, edge = 0.2 middle Low Even weight distribution, all regions weight = average high high Hotspot weight = 0.4, edge area weight = 0.6 high middle Hotspot area weight = 0.3, edge area weight = 0.7 high Low Even weight distribution, all regions weight = average The membership degree is calculated for the real-time measured values ​​of each feature in the spatial temperature field feature. The membership degree calculation formula is: ; in: Indicates the degree to which the spatial temperature field feature x belongs to the “low” membership. Indicates the degree to which the spatial temperature field feature x belongs to the "high" membership; the membership value range is between [0,1]. The closer the value is to 1, the higher the degree to which x belongs to the fuzzy set, and the closer it is to 0, the lower the degree. is the spatial temperature field characteristic, 、 is a parameter obtained from model training. a represents the center position of the membership function. When the spatial temperature field feature x is close to a, the membership function value is the highest (i.e., closest to 1), indicating that the feature value at this time conforms to the optimal state of the fuzzy set (such as "low entropy" and "high stability"). b represents the width of the membership function, that is, the speed at which the membership function changes as x deviates from the center value a. The larger the b value, the wider the membership function graph, indicating a higher tolerance for the fuzzy set; the smaller the b value, the narrower the membership function graph, indicating a lower tolerance for the fuzzy set and greater sensitivity to deviations near the center value.

[0045] The fuzzy inference engine is used to perform weighted summation of the membership of the input spatial temperature field characteristics (spatial entropy, temperature gradient mean, stability index, etc.) to determine the fusion weight of the temperature in each region.

[0046] ; in: is the fusion weight of the temperature of the ith region, For input conditions The membership degree of the corresponding j-th fuzzy rule; the division of each area is obtained through statistical analysis of long-term historical oven operation data. Specifically, a cluster analysis algorithm (such as K-means or DBSCAN) is used to accurately determine the sensor installation location and area division boundaries based on clear quantitative standards such as spatial temperature gradient significance and temperature drift sensitivity.

[0047] Calibrate and train historical oven spatial field characteristics (such as spatial entropy, mean temperature gradient, and stability index) against regional weight assignments to establish a historical feature-weight configuration library, clearly recording the correspondence between features and weights. This library data is optimized using a genetic algorithm or particle swarm optimization algorithm to determine the optimal weight distribution. During real-time operation, the historical feature-weight configuration library is dynamically updated based on feedback from online measured data, continuously optimizing the weight assignment.

[0048] Perform nonlinear weighted fusion on the fusion weights and the model-predicted temperature values ​​of key areas to ultimately determine the redundant check temperature values: ; in: is the final redundant verification temperature value; is the fusion weight obtained by the nonlinear fusion algorithm; Predict temperature values ​​for the calibrated model.

[0049] S102: Calculating a current temperature drift error value based on the redundant check temperature value, inputting the temperature drift error value and historical drift trend data into a temperature drift prediction model, and outputting a predicted temperature drift compensation value; First of all, it should be noted that during multiple consecutive heating cycles of the oven, redundant calibration temperature values ​​and expected set temperature values ​​are collected in real time, and the temperature drift error value is defined as the difference between the actual measured temperature value and the expected set temperature value; The temperature drift error values ​​obtained from multiple consecutive heating cycles are sorted according to timestamps to form a drift error value sequence; the drift error value sequence is input into the temperature drift prediction model to predict the predicted drift error value at a future moment.

[0050] The training process of the temperature drift prediction model includes: An initial temperature drift data sequence is established based on historical temperature drift error data, and the data sequence is input into a time series prediction model based on a recurrent neural network; The temperature drift error value at future moments is predicted through model iteration, and the recurrent neural network parameters are updated based on the actual error value feedback, and the trained temperature drift prediction model is output.

[0051] The method for constructing the initial temperature drift data sequence includes: Extract the temperature drift error value during multiple consecutive heating cycles of the oven operation; Arrange the temperature drift error values ​​in chronological order to form a continuous time series.

[0052] For example, whenever oven temperature control is required, the temperature drift error sequence of the current moment and several previous moments (for example, the past 30 minutes to several hours) is input into the trained LSTM model, and the predicted temperature drift error values ​​of several future moments are output; based on the predicted temperature drift error value, the predicted temperature drift compensation value is determined by reverse compensation. After actually controlling the oven, the new temperature drift error value is measured, and the new error value is compared with the predicted error value. The model parameters are updated based on the difference feedback, and the optimization is continued. The weights and bias parameters of the recurrent neural network are continuously optimized through online learning to ensure long-term effectiveness.

[0053] S103: Fusing the redundant check temperature value and the predicted temperature drift compensation value to generate a first fused temperature value; and calculating a temperature stability index inside the oven based on the first fused temperature value. Specifically, the logic for obtaining the first fusion temperature value is as follows: Establish the Kalman filter model state equation and observation equation, Equation of state:

[0054] in: is the true temperature, To predict the drift compensation value, is the process noise.

[0055] Observation equation: ; in, To check the temperature value for redundancy, To measure noise.

[0056] The process noise covariance is dynamically estimated based on the historical drift compensation value error, and the measurement noise covariance is dynamically adjusted based on the real-time redundant verification of temperature measurement accuracy; The fusion weight is adaptively adjusted by the Kalman gain to achieve real-time nonlinear fusion of redundant verification temperature values ​​and predicted temperature drift compensation values: .

[0057] Further explanation, the temperature stability index calculation formula is: ; in: is the i-th measurement value in the real-time temperature measurement sequence; is the average value of the temperature values ​​in the measurement sequence; N is the number of measurement values ​​in the sampling window (generally taking the most recent consecutive measurement points, such as 30 points or 30 seconds of real-time data); the temperature stability index reflects the degree of fluctuation of the temperature measurement sequence. The lower the value, the more stable the temperature.

[0058] S104: Based on the temperature stability index and the preset temperature fluctuation target range, the heating power of the oven is adjusted by an adaptive frequency heating control algorithm; It should be noted that: to achieve fine and stable control of the oven temperature, the temperature stability index monitored in real time is compared with the preset temperature fluctuation target range. According to the deviation between the two, an adaptive frequency heating control algorithm is used to dynamically adjust the heating frequency and power of the oven to ensure long-term stability and small fluctuations in the temperature inside the oven.

[0059] Specifically, the method of adjusting the heating power of the oven by an adaptive frequency heating control algorithm based on the temperature stability index and the preset temperature fluctuation target range includes: Calculate the deviation between the current temperature stability index and the target range. The deviation formula is: ; Has positive and negative connotations; when , then the temperature stability index exceeds the target range, indicating large fluctuations; when , the temperature stability index is better than the target range, indicating small fluctuations (possibly too stable or conservatively controlled).

[0060] Temperature fluctuation target range setting, set the oven temperature fluctuation target range in advance (for example, ±0.5℃), and record it as the target stability index ;The selection of target stability index is determined by the baking process requirements and is determined when the equipment leaves the factory or is set by the user.

[0061] Define the control factor to quantify the temperature deviation, and define the nonlinear control factor (CF) according to the deviation. The control factor (CF) is calculated as follows: ; Where: k is the sensitivity coefficient, which controls the sensitivity of the deviation to the adjustment range (the empirical value range is generally 3~10); CF ranges from [-1,1]: when CF is close to 1, it means that the temperature fluctuation is large and significant adjustment is required; when CF is close to -1, it means that the temperature fluctuation is very small and the power can be slightly reduced; when CF is close to 0, no significant adjustment is required.

[0062] The adjustment range of heating frequency and power is determined by nonlinear mapping rules. According to the control factor CF value, the adjustment range of heating cycle frequency (f) and power (P) is determined by predefined nonlinear mapping rules: Non-linear mapping rules (example): Control factor CF Heating frequency adjustment (Δf) Power adjustment (ΔP) 0.7 ~ 1.0 +30% (increase frequency) +20% (increase power) 0.3 ~ 0.7 +20% (increase frequency) +10% (increase power) 0.0 ~ 0.3 +10% (fine-tuning frequency) +5% (fine-tuning power) -0.3 ~ 0.0 0% (maintain) 0% (maintain) -0.7 ~ -0.3 -10% (slightly reduced frequency) -5% (slightly reduced power) -1.0 ~ -0.7 -20% (reduce frequency) -10% (reduced power) Real-time dynamic adjustment of heating power and frequency: Combined with the above mapping rules, the oven heating control system is adjusted in real time: Heating cycle frequency: ; Heating power:

[0063] in: and Respectively represent the frequency and power before adjustment; and is the adjustment ratio determined in the above table.

[0064] In this embodiment, the deviation between the real-time volatility of the temperature stability index and the target range is used to define a nonlinear control factor to achieve dynamic and precise control. The nonlinear sigmoid function mapping is used to accurately capture the changes in the deviation amount in different ranges, avoiding overshoot or slow response caused by simple linear control. The concept of adaptive control is integrated to enable the control algorithm to adapt to different oven operating states and environmental changes. The control parameters are dynamically adjusted according to real-time operation feedback to achieve intelligent closed-loop control, which is accurate and robust.

[0065] S105: When the oven enters the next heating cycle, the second temperature data inside the oven is recollected according to the heating power adjusted in S104, and the second temperature data is cross-validated with the first fusion temperature value to correct the historical drift trend data.

[0066] It should be noted that after adjusting the heating power according to the adaptive frequency heating control in step S104, the next heating cycle is entered, and multiple temperature sensors inside the oven are used to measure and record new temperature measurement data in real time, and marked as second temperature data. The second temperature data is a real-time measurement value of a time series.

[0067] Cross-validating the second temperature data with the first fused temperature value to correct historical drift trend data includes: Calculate the second temperature data and compare it with the first fusion temperature value point by point or as a whole to obtain the temperature difference value, and understand the actual error or deviation between the actual measured temperature in the current cycle and the previously predicted temperature; Based on the temperature difference values, the average deviation between the current temperature control system prediction and actual operating conditions is determined. This deviation reflects the degree of match between the system's predicted performance and the actual operation of the oven.

[0068] By correcting drift trends in historical data by averaging the deviation, the historical trend data more accurately reflects the oven's true temperature variations. This provides a more accurate historical reference for temperature prediction and compensation in the next cycle. This continuous feedback correction mechanism based on actual measurement data can continuously reduce temperature prediction errors in future cycles, further enhancing the system's long-term control accuracy and stability of the oven temperature.

[0069] It should be noted that when the oven enters the next heating cycle, the system recollects internal oven temperature data based on the heating power adjusted in the previous step (S104). This data is referred to as "second temperature data." To ensure the long-term accuracy of the oven temperature control system, the system compares and verifies the newly collected second temperature data for the current cycle with the first fusion temperature value obtained in the previous cycle. This process is called "cross-validation." During cross-validation, the difference between the second temperature data and the first fusion temperature value from the previous cycle is first calculated. This difference reflects the degree of change between the current actual temperature and the system's previous predicted temperature.

[0070] Next, the system uses this difference value and a mathematical method called "least squares" to readjust and correct the historical temperature drift trend data. This aims to make the historical temperature drift trend data more accurately reflect the actual operation of the oven and ensure that the next temperature prediction and compensation will be more precise and reduce errors. Through this process, the system achieves continuous optimization and precise adjustment of the oven's temperature control, ensuring the long-term stable and reliable operation of the oven.

[0071] This embodiment utilizes a multi-sensor redundancy verification strategy, employing abnormal sensor rejection and covariance matrix analysis to avoid error accumulation caused by long-term drift in a single sensor. A recurrent neural network model is incorporated into the temperature drift prediction model, which uses historical data to predict long-term drift trends and provides dynamic compensation. Real-time dynamic temperature feedback allows for nonlinear adjustment of the heating frequency to ensure temperature stability, enabling closed-loop feedback to dynamically correct temperature drift trend data and ensure long-term accuracy. This multi-path cross-feedback and verification creates a nonlinear closed-loop control structure, offering substantial advancements over existing technologies.

[0072] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A temperature control method for an intelligent oven based on an adaptive control method, characterized in that: The following steps are involved: S101: collecting first temperature data inside the oven based on a sensor, and obtaining a redundant verification temperature value using a redundant verification method based on multiple temperature sensors; S102: Calculating a current temperature drift error value based on the redundant check temperature value, inputting the temperature drift error value and historical drift trend data into a temperature drift prediction model, and outputting a predicted temperature drift compensation value; S103: Fusing the redundant check temperature value and the predicted temperature drift compensation value to generate a first fused temperature value; calculating a temperature stability index inside the oven based on the first fusion temperature value; S104: Based on the temperature stability index and the preset temperature fluctuation target range, the heating power of the oven is adjusted by an adaptive frequency heating control algorithm; S105: When the oven enters the next heating cycle, the second temperature data inside the oven is recollected according to the heating power adjusted in S104, and the second temperature data is cross-validated with the first fusion temperature value to correct the historical drift trend data.

2. The temperature control method of an intelligent oven based on an adaptive control method according to claim 1, characterized in that: The acquisition logic of the redundant verification temperature value: Arrange multiple temperature sensors in the oven to collect first temperature data, use the spatial coordinates of the sensor measurement points as discrete spatial temperature points, construct a three-dimensional spatial distribution model of the temperature field inside the oven, and extract time-series temperature spatial data; Input the time series temperature spatial data into the pre-built spatial interpolation neural network model to calculate the spatial continuous temperature field data; calculate the deviation between the time series temperature spatial data and the spatial continuous temperature field data in real time, compare and analyze the deviation with the dynamic adaptive threshold, determine the abnormal measurement value and automatically correct it to form the corrected temperature field data; Extracting features from the corrected temperature field data to obtain spatial temperature field features, wherein the spatial temperature field features include spatial temperature entropy, spatial temperature gradient mean, and temperature field stability index; The spatial temperature field characteristics and the corrected temperature field data are fused through a nonlinear adaptive algorithm to determine the nonlinear fusion weight of each region; The fusion weights are nonlinearly weighted fused with the temperature prediction values ​​of key areas to output redundant verification temperature values.

3. The temperature control method of an intelligent oven based on an adaptive control method according to claim 2, characterized in that: The application logic of the neural network model: The structure of the neural network model adopts a 3D convolutional neural network or a three-dimensional space Transformer network; the model input is the real-time spatial position coordinates and temperature measurement values ​​of each sensor; the model output is the continuous temperature prediction value of each point in the complete three-dimensional space; Based on a large amount of historical time-series temperature spatial data, a deep neural network model is trained to fit discrete spatial point temperature data into a continuous three-dimensional spatial temperature field; based on the spatial continuous temperature field data output by the neural network in real time; If the deviation between the time series temperature space data and the time series temperature space data exceeds the set threshold, the time series temperature space data is automatically corrected through the time series temperature space data to form corrected temperature field data.

4. The temperature control method of an intelligent oven based on an adaptive control method according to claim 3, characterized in that: The training process of the temperature drift prediction model includes: An initial temperature drift data sequence is established based on historical temperature drift error data, and the data sequence is input into a time series prediction model based on a recurrent neural network; The temperature drift error value at future moments is predicted through model iteration, and the recurrent neural network parameters are updated based on the actual error value feedback, and the trained temperature drift prediction model is output; The method for constructing the initial temperature drift data sequence includes: Extract the temperature drift error value during multiple consecutive heating cycles of the oven operation; Arrange the temperature drift error values ​​in chronological order to form a continuous time series.

5. The temperature control method of an intelligent oven based on an adaptive control method according to claim 4, characterized in that: The logic for obtaining the first fusion temperature value is as follows: Establish the Kalman filter model state equation and observation equation, The process noise covariance is dynamically estimated based on the historical drift compensation value error, and the measurement noise covariance is dynamically adjusted based on the real-time redundant verification of temperature measurement accuracy; The fusion weight is adaptively adjusted by Kalman gain to achieve real-time nonlinear fusion of redundant check temperature values ​​and predicted temperature drift compensation values.

6. The temperature control method of an intelligent oven based on an adaptive control method according to claim 5, characterized in that: The temperature stability index calculation formula is: ； in: is the i-th measurement value in the real-time temperature measurement sequence; is the average temperature value of the measurement sequence; N is the number of measurements in the sampling window.

7. The temperature control method of an intelligent oven based on an adaptive control method according to claim 6, characterized in that: The method of adjusting the heating power of the oven by an adaptive frequency heating control algorithm based on the temperature stability index and the preset temperature fluctuation target range includes: Calculate the deviation between the current temperature stability index and the target range, define a control factor to quantify the temperature deviation, and define a nonlinear control factor based on the deviation; The adjustment range of the heating frequency and power is determined by a nonlinear mapping rule. According to the control factor value, the adjustment range of the heating cycle frequency and power is determined by a predefined nonlinear mapping rule.

8. The temperature control method of an intelligent oven based on an adaptive control method according to claim 7, characterized in that: Cross-validating the second temperature data with the first fused temperature value to correct historical drift trend data includes: Calculate the second temperature data and compare it with the first fusion temperature value point by point or as a whole to obtain a temperature difference value; Based on the temperature difference value, determine the average deviation between the current temperature control system prediction and actual operating conditions: The drift trend in the historical data is corrected by the average deviation size to predict the temperature prediction and temperature compensation for the next period.

9. An intelligent oven temperature control system based on an adaptive control method, based on the implementation of the intelligent oven temperature control method based on an adaptive control method according to any one of claims 1 to 8, characterized in that: It includes temperature acquisition and redundancy verification module, temperature drift prediction and compensation module, temperature fusion and stability assessment module, adaptive heating control module and drift trend feedback correction module: The temperature acquisition and redundancy check module collects the first temperature data inside the oven based on the sensor and obtains the redundant check temperature value using a redundant check method based on multiple temperature sensors; The temperature drift prediction and compensation module calculates the current temperature drift error value based on the redundant check temperature value, inputs the temperature drift error value and historical drift trend data into the temperature drift prediction model, and outputs the predicted temperature drift compensation value; The temperature fusion and stability evaluation module fuses the redundant check temperature value with the predicted temperature drift compensation value to generate a first fused temperature value; calculating a temperature stability index inside the oven based on the first fusion temperature value; Adaptive heating control module, which adjusts the oven's heating power through an adaptive frequency heating control algorithm based on the temperature stability index and a preset temperature fluctuation target range; The drift trend feedback correction module re-collects the second temperature data inside the oven according to the heating power adjusted in S104 when the oven enters the next heating cycle, and cross-validates the second temperature data with the first fusion temperature value to correct the historical drift trend data.

Images