Self-adaptive adjustment electric oven temperature control system and method

Abstract

The invention discloses a self-adaptive adjustment electric oven temperature control system and method. The method comprises the following steps: acquiring initial multi-mode oven data; a CNN-Transform oven state prediction model is established, and a CNN-Transform oven state prediction model is established; the initial multi-mode oven data are input into a CNN-Transform oven state prediction model to be recognized, and an oven prediction state is obtained; judging according to the oven prediction state, and if abnormal temperature fluctuation is judged, triggering emergency power off and switching to a PID standby mode; and if it is judged that the food materials are locally overheated, a lateral cold air system is started to directionally cool the interior of the oven. And accurate temperature control can be provided when the cooking requirements of different types and different quantities of food materials are met.

Description

Technical Field

[0001] This invention relates to the field of oven temperature control technology, and in particular to an adaptive electric oven temperature control system and method. Background Technology

[0002] The limitations of traditional oven temperature control methods are becoming increasingly apparent. They cannot perceive changes in the state of food in real time. Foods vary greatly in their temperature and time requirements during cooking due to differences in type, size, and initial moisture content, making it difficult for traditional methods to accurately match these requirements. Furthermore, there is uneven temperature distribution inside the oven; data from a single temperature sensor cannot accurately represent the temperature of the entire oven cavity, easily leading to localized overheating or undercooking of food. In addition, there is a lack of intelligent identification and rapid response capabilities for abnormal situations. When sudden situations such as heating element failure or ventilation malfunction occur, it may result in burnt food, equipment damage, or even safety hazards. Summary of the Invention

[0003] The purpose of this invention is to solve the above-mentioned problems by designing an adaptive temperature control system and method for an electric oven.

[0004] To achieve the above objectives, the technical solution of the present invention is as follows: Further, in the above-mentioned adaptive electric oven temperature control system, the electric oven temperature control system includes a multi-stage reactor coupling module, a gas component monitoring module, a core-shell structure catalytic module, an inverse conversion synthesis module, and a multi-stage waste heat recovery module, wherein:

[0005] The oven data acquisition module is used to acquire temperature data, humidity data, weight data, and image data from the sensor array inside the oven to obtain multimodal oven data; the multimodal oven data is preprocessed to obtain initial multimodal oven data;

[0006] The prediction model building module is used to extract local spatial features based on CNN convolutional neural network, identify color changes in oven food, capture global temporal dependencies using Transformer, integrate multimodal data and predict temperature curves, and build a CNN-Transformer oven state prediction model.

[0007] The oven state prediction module is used to input the initial multimodal oven data into the CNN-Transformer oven state prediction model for identification, and obtain the oven predicted state.

[0008] The oven temperature control module is used to make judgments based on the predicted state of the oven. If an abnormal temperature fluctuation is detected, an emergency power outage is triggered and the system switches to PID standby mode.

[0009] The oven cooling control module is used to activate the side cooling system to cool the interior of the oven in a targeted manner if it is determined that the food is locally overheated.

[0010] Furthermore, in the above-mentioned electric oven temperature control system, the oven data acquisition module includes the following units:

[0011] The temperature data processing unit is used to use a median filtering algorithm to take five consecutive time points of data collected by each temperature sensor to form a sliding window, calculate the median of the data within the window as the effective temperature value at the current moment, and obtain the initial temperature data.

[0012] A humidity data processing unit is used to perform mean filtering on humidity data to obtain filtered humidity data, and to perform normalization processing on the filtered humidity data to obtain initial humidity data.

[0013] The weight data processing unit is used to fit the trend term of the weight data using the least squares method, subtract the trend term from the original data to obtain weight fluctuation data, and standardize the weight fluctuation data to obtain initial weight data.

[0014] The image data processing unit is used to perform grayscale processing on the image, convert the color image into a grayscale image, enhance the grayscale image, and improve the contrast of the image using a histogram equalization method to obtain initial image data.

[0015] Furthermore, in the aforementioned electric oven temperature control system, the CNN convolutional neural network includes:

[0016] The model building layer is used to construct a multi-layer CNN convolutional neural network, where the input layer is used to receive preprocessed image data;

[0017] The first convolutional layer uses 32 3x3 convolutional kernels with a stride of 1 and padding of 1 to perform convolution operations on the input image and extract local features of edges and textures.

[0018] The first pooling layer is used for the max pooling method. The pooling kernel size is 2x2 and the stride is 2. It downsamples the output of convolutional layer 1 to reduce the amount of data and retain the main features.

[0019] The second convolutional layer uses 64 3x3 convolutional kernels with a stride of 1 and padding of 1 to convolve the output of pooling layer 1 again.

[0020] The color recognition layer is used to establish a color recognition layer by connecting a fully connected layer after the convolutional layer, corresponding to the three channels of the RGB color space.

[0021] Furthermore, in the above-mentioned electric oven temperature control system, the prediction model establishment module includes:

[0022] The data integration unit is used to integrate temperature data, humidity data, weight data, and image features extracted by CNN according to time series to obtain multimodal time series data;

[0023] A data input unit is used to input the multimodal time series data into a Transformer model, wherein the Transformer model includes at least an encoder and a decoder;

[0024] The data computation unit is used to calculate the attention scores between multimodal data at different time points through a self-attention mechanism.

[0025] Furthermore, in the above-mentioned electric oven temperature control system, the oven temperature control module includes:

[0026] The status judgment unit is used to determine if the oven prediction status is abnormal, and then calculate the root mean square error (RMSE) between the predicted temperature curve and the set normal temperature curve.

[0027] The threshold calculation unit is used to determine abnormal temperature fluctuations if the root mean square error of RMSE exceeds a preset threshold.

[0028] Furthermore, in the above-mentioned electric oven temperature control system, the oven temperature control module includes:

[0029] The mode switching unit is used to immediately trigger an emergency power-off mechanism when abnormal temperature fluctuations are detected, cutting off the heating power supply to the electric oven and switching to PID standby mode.

[0030] The temperature regulation unit is used to calculate the temperature control quantity based on the deviation between the current temperature feedback value and the set value using a PID controller, generate an adjustment command based on the temperature control quantity, and adjust the power of the heating element and the fan speed based on the adjustment command.

[0031] Furthermore, in the above-mentioned electric oven temperature control system, the oven cooling control module includes:

[0032] The directional cooling unit is used to activate the side cooling system to cool the inside of the oven in a directional manner when it is determined that the food is locally overheated. The side cooling system consists of adjustable-angle fans and cold air ducts installed on both sides of the oven.

[0033] The cooling adjustment unit is used to control the system to generate cooling adjustment commands based on the local overheating location determined by image recognition. It automatically adjusts the angle and speed of the fan to direct the cold air towards the overheated area and quickly reduce the temperature of that area.

[0034] In an adaptive temperature control method for an electric oven, the method includes the following steps:

[0035] Temperature data, humidity data, weight data, and image data from the sensor array inside the oven are acquired to obtain multimodal oven data; the multimodal oven data is preprocessed to obtain initial multimodal oven data;

[0036] Based on CNN convolutional neural network to extract local spatial features and identify color changes in oven food, Transformer is used to capture global temporal dependencies, and multimodal data is integrated to predict temperature curves, thus establishing a CNN-Transformer oven state prediction model.

[0037] The initial multimodal oven data is input into the CNN-Transformer oven state prediction model for identification, and the predicted oven state is obtained.

[0038] Based on the oven's predicted status, if an abnormal temperature fluctuation is detected, an emergency power outage is triggered and the system switches to PID standby mode.

[0039] If it is determined that the food is overheated in a localized area, the side cooling system will be activated to cool the interior of the oven in a targeted manner.

[0040] Furthermore, in the above-mentioned electric oven temperature control method, the step of judging based on the oven's predicted state, and if it is determined to be an abnormal temperature fluctuation, triggering an emergency power outage and switching to PID standby mode, includes:

[0041] If an abnormal temperature fluctuation is detected, the system immediately triggers an emergency power-off mechanism, cutting off the heating power to the electric oven and switching to PID standby mode.

[0042] The PID controller calculates the temperature control quantity based on the deviation between the current temperature feedback value and the set value, generates an adjustment command based on the temperature control quantity, and adjusts the power of the heating element and the fan speed based on the adjustment command.

[0043] Furthermore, in the above-mentioned electric oven temperature control method, the step of activating the lateral cooling system to directionally cool the interior of the oven if it is determined that the food is locally overheated includes:

[0044] If it is determined that the food is locally overheated, the side cooling system will be activated to cool the inside of the oven in a directional manner. The side cooling system consists of adjustable fans and cooling ducts installed on both sides of the oven.

[0045] Based on the location of localized overheating determined by image recognition, the control system generates a cooling adjustment command, automatically adjusts the angle and speed of the fan, and directs the cool air toward the overheated area to quickly reduce the temperature of that area.

[0046] Its beneficial effects lie in obtaining multimodal oven data by acquiring temperature, humidity, weight, and image data from the internal sensor array of the oven; preprocessing the multimodal oven data to obtain initial multimodal oven data; extracting local spatial features based on a CNN convolutional neural network and identifying color changes in the oven food, using a Transformer to capture global temporal dependencies, and simultaneously integrating multimodal data and predicting temperature curves to establish a CNN-Transformer oven state prediction model; inputting the initial multimodal oven data into the CNN-Transformer oven state prediction model for identification to obtain the oven predicted state; judging based on the oven predicted state, if it is determined to be an abnormal temperature fluctuation, an emergency power cut is triggered and the system switches to PID standby mode; if it is determined to be local overheating of the food, the side cooling system is activated to directionally cool the oven interior. 1. Compared with the traditional single temperature control method, the temperature curve can be dynamically adjusted according to the characteristics of the food and the temperature and humidity distribution inside the oven, significantly improving the accuracy and consistency of cooking and meeting the diverse needs of users for different foods and different cooking effects. 2. Based on the oven's status prediction index, the system can promptly identify abnormal temperature fluctuations and localized overheating of food, and quickly take corresponding measures, shifting from passive control to proactive intelligent decision-making, effectively avoiding cooking failures caused by improper temperature control. 3. When abnormal temperature fluctuations are detected, the heating power is immediately cut off and PID control is activated to prevent temperature runaway and potential fires or other safety accidents. For locally overheated food, a lateral cooling system provides targeted cooling to prevent over-burning and the generation of harmful substances, while simultaneously protecting the oven's internal structure from high-temperature damage, ensuring comprehensive user safety. 4. It can adapt to the cooking needs of different types and quantities of food, providing precise temperature control whether baking bread, meat, or pastries. Furthermore, the intelligent control and rapid response functions reduce the tedious manual adjustments required by users. Users only need to set basic cooking parameters, and the oven will automatically optimize the temperature control process, providing a convenient, efficient, and intelligent cooking experience. Attached Figure Description

[0047] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0048] Figure 1 This is a schematic diagram of the first embodiment of an adaptive temperature control system for an electric oven according to the present invention;

[0049] Figure 2This is a schematic diagram of the first embodiment of an adaptive adjustment electric oven temperature control method according to the present invention;

[0050] Figure 3 This is a schematic diagram of the structure of an electric oven according to an embodiment of the present invention, which is an adaptive temperature control system for an electric oven. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0052] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this specification means the presence of features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0053] The present invention will now be described in detail with reference to the accompanying drawings, such as... Figure 1 As shown, an adaptive temperature control system for an electric oven includes the following modules:

[0054] The oven data acquisition module is used to acquire temperature data, humidity data, weight data, and image data from the sensor array inside the oven to obtain multimodal oven data; the multimodal oven data is preprocessed to obtain initial multimodal oven data;

[0055] Specifically, this embodiment also includes a temperature data processing unit, which uses a median filtering algorithm to take data from five consecutive time points of each temperature sensor to form a sliding window, calculates the median of the data within the window as the effective temperature value at the current moment, and obtains the initial temperature data.

[0056] The humidity data processing unit is used to perform mean filtering on humidity data to obtain filtered humidity data, and to perform normalization processing on the filtered humidity data to obtain initial humidity data.

[0057] The weight data processing unit is used to fit the trend term of the weight data using the least squares method, subtract the trend term from the original data to obtain the weight fluctuation data, and standardize the weight fluctuation data to obtain the initial weight data.

[0058] The image data processing unit is used to perform grayscale processing on the image, convert the color image to a grayscale image, enhance the grayscale image, improve the image contrast using histogram equalization, and obtain the initial image data.

[0059] Specifically,

[0060] (I) Multi-dimensional data acquisition

[0061] A sensor array is strategically deployed inside the electric oven, including temperature sensors, humidity sensors, weight sensors, and an image acquisition device. High-precision thermocouple sensors are evenly distributed across the upper, middle, and lower layers of the oven cavity, with three sensors on each layer (left, center, and right), totaling nine sensors, to comprehensively acquire temperature data from different areas within the oven. Capacitive humidity sensors are used, with two sensors installed at the top and two at the bottom of the oven cavity to collect real-time humidity information. The weight sensors, integrated under the oven tray, are strain gauge load cells capable of accurately measuring weight changes in the food. A high-resolution camera is installed above the inside of the oven door to clearly capture real-time images of the food inside the oven.

[0062] The aforementioned sensor array continuously collects temperature, humidity, weight, and image data from inside the oven, forming multi-dimensional oven data. Temperature data is collected 10 times per second, humidity and weight data are collected 5 times per second, and image data is collected 2 frames per second.

[0063] (ii) Data preprocessing

[0064] The acquired multimodal oven data is preprocessed to obtain initial multimodal oven data. The specific preprocessing steps are as follows:

[0065] Temperature data preprocessing: Median filtering algorithm is used to denoise the temperature data, removing outliers caused by sensor noise or external interference. For each temperature sensor, data from five consecutive time points are used to form a sliding window, and the median of the data within the window is calculated as the valid temperature value at the current moment.

[0066] Humidity data preprocessing: First, mean filtering is applied to the humidity data to eliminate the influence of random noise. Then, normalization is performed to map the humidity values ​​to the [0,1] interval for subsequent model processing.

[0067] Weight data preprocessing: Since weight data mainly reflects changes such as moisture evaporation in the food during oven operation and is relatively stable, a detrending process is first performed. The least squares method is used to fit the trend term of the weight data, and then the trend term is subtracted from the original data to obtain the weight fluctuation data after removing the long-term trend. Then, the data is standardized to ensure a mean of 0 and a standard deviation of 1.

[0068] Image data preprocessing: First, the image is converted to grayscale to reduce data volume and highlight the brightness variations of the ingredients. Then, image enhancement is performed using histogram equalization to improve contrast and make the color variations of the ingredients more pronounced. Finally, the image is resized to a uniform size of 224x224 pixels for input into the convolutional neural network.

[0069] The prediction model building module is used to extract local spatial features based on CNN convolutional neural network, identify color changes in oven food, capture global temporal dependencies using Transformer, integrate multimodal data and predict temperature curves, and build a CNN-Transformer oven state prediction model.

[0070] Specifically, this embodiment also includes a model building layer for constructing a multi-layer CNN convolutional neural network, wherein the input layer is used to receive preprocessed image data;

[0071] The first convolutional layer uses 32 3x3 convolutional kernels with a stride of 1 and padding of 1 to perform convolution operations on the input image and extract local features of edges and textures.

[0072] The first pooling layer is used for the max pooling method. The pooling kernel size is 2x2 and the stride is 2. It downsamples the output of convolutional layer 1 to reduce the amount of data and retain the main features.

[0073] The second convolutional layer uses 64 3x3 convolutional kernels with a stride of 1 and padding of 1 to convolve the output of pooling layer 1 again.

[0074] The color recognition layer is used to establish a color recognition layer by connecting a fully connected layer after the convolutional layer, corresponding to the three channels of the RGB color space.

[0075] The data integration unit is used to integrate temperature data, humidity data, weight data, and image features extracted by CNN according to time series to obtain multimodal time series data;

[0076] The data input unit is used to input multimodal time series data into the Transformer model, which includes at least an encoder and a decoder.

[0077] The data computation unit is used to calculate the attention scores between multimodal data at different time points through a self-attention mechanism.

[0078] Specifically,

[0079] (I) CNN Extraction of Local Spatial Features and Color Change Recognition

[0080] A multi-layer convolutional neural network (CNN) is constructed to extract local spatial features from oven-baked food and identify color changes. The CNN network structure is as follows:

[0081] Input layer: Receives preprocessed image data, with an input dimension of 224x224x1 (grayscale image).

[0082] Convolutional Layer 1: Using 32 3x3 convolutional kernels with a stride of 1 and padding of 1, the input image is convolved to extract primary local features such as edges and textures. The ReLU activation function is used to introduce non-linearity.

[0083] Pooling layer 1: Max pooling is used with a 2x2 kernel size and a stride of 2. The output of convolutional layer 1 is downsampled to reduce the amount of data and retain the main features.

[0084] Convolutional layer 2: Using 64 3x3 convolutional kernels with a stride of 1 and padding of 1, the output of pooling layer 1 is convolved again to extract more complex local spatial features.

[0085] Pooling layer 2: Also uses 2x2 max pooling with a step size of 2.

[0086] Color recognition layer: A fully connected layer follows the convolutional layer. The input dimension is 64x56x56 (assuming the size after pooling is 56x56), and the output dimension is 3, corresponding to the three channels of the RGB color space. By training this fully connected layer, the model can recognize the color changes of food, such as the color gradation from raw to cooked.

[0087] (ii) Transformer captures global temporal dependencies

[0088] Temperature, humidity, and weight data, along with image features extracted by a CNN (after flattening), are integrated into a time series dataset to form multimodal temporal data. This time series data is then input into a Transformer model to capture global temporal dependencies. The Transformer architecture includes an encoder and a decoder; here, the encoder is primarily used to process the multimodal temporal data.

[0089] The encoder consists of multiple identical stacked layers, each containing two sublayers: a self-attention mechanism sublayer and a feedforward neural network sublayer. In the self-attention mechanism, attention scores are calculated between multimodal data points at different time points, enabling the model to focus on historical information important for the current prediction. For example, when predicting the temperature at a future time, the model automatically assigns higher attention weights to historical temperature, humidity, food weight, and color changes that are related to temperature variations.

[0090] (III) Multimodal data integration and temperature curve prediction

[0091] The local spatial features and color change information extracted by the CNN are integrated with the global temporal dependency information captured by the Transformer. Specifically, the feature vector output by the CNN is concatenated with the output of the Transformer encoder, and then input into a fully connected neural network. The output dimension of this network is the temperature prediction value for a future time period (e.g., the next 30 minutes), with each time point spaced 1 minute apart, thus forming a temperature prediction curve.

[0092] During model training, mean squared error (MSE) was used as the loss function, and the optimization objective was to minimize the error between the predicted temperature curve and the actual temperature curve. The training data came from the historical operating data of the electric oven under different ingredients and cooking modes. Through extensive training, the model was able to accurately predict the oven's condition.

[0093] The oven state prediction module is used to input the initial multimodal oven data into the CNN-Transformer oven state prediction model for recognition and to obtain the predicted oven state.

[0094] Specifically, in this embodiment, the preprocessed initial multimodal oven data (including processed temperature, humidity, weight data, and image data) is input into the trained CNN-Transformer oven state prediction model according to the input format required by the model. The model first extracts local spatial features and identifies color changes from the image data using a CNN, then inputs the extracted features along with temporal data from other modalities into the Transformer to capture global temporal dependencies, and finally outputs the oven state prediction index through a fully connected layer.

[0095] The Oven Condition Prediction Index is a numerical value that comprehensively reflects the current condition of the oven, ranging from 0 to 100. The higher the index, the closer the oven is to normal cooking conditions; the lower the index, the more abnormal the oven condition, and there may be problems such as temperature fluctuations or localized overheating.

[0096] The oven temperature control module is used to make judgments based on the oven's predicted status. If an abnormal temperature fluctuation is detected, it will trigger an emergency power outage and switch to PID standby mode.

[0097] Specifically, this embodiment also includes a status judgment unit, which is used to determine if the oven prediction status is abnormal, and then calculate the root mean square error of RMSE between the predicted temperature curve and the set normal temperature curve.

[0098] The threshold calculation unit is used to determine abnormal temperature fluctuations if the root mean square error of RMSE exceeds a preset threshold.

[0099] The mode switching unit is used to immediately trigger an emergency power-off mechanism when abnormal temperature fluctuations are detected, cutting off the heating power supply to the electric oven and switching to PID standby mode.

[0100] The temperature control unit is used to calculate the temperature control quantity based on the deviation between the current temperature feedback value and the set value using a PID controller, generate an adjustment command based on the temperature control quantity, and adjust the power of the heating element and the fan speed based on the adjustment command.

[0101] Specifically,

[0102] (I) Multi-dimensional anomaly judgment system

[0103] A three-level abnormal temperature fluctuation judgment mechanism is constructed, comprehensively considering temperature prediction error, rate of change, and historical fluctuation trends:

[0104] Basic error judgment: Calculate the root mean square error (RMSE) between the predicted temperature curve and the preset normal curve, set dynamic thresholds for different ingredients / cooking modes (the threshold for baking pastries is set to 4℃, and the threshold for roasting meat is set to 6℃), and determine the optimal threshold range through training with historical data.

[0105] Rate of sudden change detection: The temperature change rate (ΔT / Δt) is calculated in real time. If the heating rate exceeds 10℃ / s or the cooling rate exceeds 8℃ / s within 3 seconds (exceeding the normal temperature control range of the oven), an abnormal warning is triggered directly.

[0106] Trend consistency verification: Compare the temperature trend predicted by the Transformer model with the actual sensor data trend. If the opposite direction occurs for 5 consecutive time points (500ms) (predicted temperature rise but actual continuous temperature drop), it is determined to be an abnormal fluctuation caused by model prediction failure.

[0107] (II) Tiered Response Execution Process

[0108] 1. Emergency power failure protection module

[0109] Hardware implementation: A dual-relay parallel structure is adopted, with the main relay controlling the power supply of the heating tube and the auxiliary relay controlling the power supply of the control board. The power failure response time is <20ms. A supercapacitor is equipped to maintain the operation of the sensor and control system for 30 seconds after a power failure, ensuring that status data is not lost.

[0110] Safety linkage: When power is cut off, a buzzer alarm (85dB) is triggered, the oven display shows the "ERROR01" code, and the mobile APP pushes an abnormal notification, including the real-time temperature curve and the fault type.

[0111] 2. PID Backup Mode Switching Logic

[0112] Initialization parameter loading: Based on the current cooking mode (baking, grilling, fermentation), the pre-stored PID parameter group (typical parameters: P=0.8, I=0.2, D=0.15) is automatically called, and the differential coefficient D is dynamically adjusted in combination with real-time humidity and food weight data (D value increases by 0.05 for every 10% increase in humidity).

[0113] Smooth transition control: A feedforward compensation algorithm is used at the moment of switching, and the temperature deviation before power failure is used as the feedforward input to avoid temperature overshoot during the initial adjustment of PID; the heating tube power is adjusted by PWM (pulse width modulation) technology, with an adjustment accuracy of 1% of the rated power.

[0114] Mode exit condition: When the temperature fluctuation is less than ±2℃ for 10 consecutive minutes and the prediction model returns to normal output, it will automatically switch back to intelligent control mode, and the current heating power will remain unchanged during the switching process.

[0115] The oven cooling control module is used to activate the side cooling system to cool the interior of the oven in a targeted manner if it is determined that the food is locally overheated.

[0116] Specifically, this embodiment also includes a directional cooling unit, which is used to activate the lateral cooling system to directionally cool the inside of the oven when it is determined that the food is locally overheated. The lateral cooling system consists of adjustable-angle fans and cold air ducts installed on both sides of the oven.

[0117] The cooling adjustment unit is used to control the system to generate cooling adjustment commands based on the local overheating location determined by image recognition. It automatically adjusts the angle and speed of the fan to direct the cold air towards the overheated area and quickly reduce the temperature of that area.

[0118] (I) Multimodal fusion detection algorithm

[0119] Construct a triple detection system of "image feature recognition + spatial temperature gradient analysis + food deformation monitoring":

[0120] Image feature recognition:

[0121] A lightweight Yolov8 model was used to detect burn marks on food surfaces in real time. The training data included 100,000 food images at different focal lengths (burn mark area ratio 0-30%), with a detection accuracy of 92%.

[0122] Color space conversion: Convert the preprocessed grayscale image to HSV color space, set the focal spot feature threshold (H∈[0,10], S∈[0.3,1], V∈[0.1,0.5]), and mark the region as a suspected overheated area when the pixel ratio of the region meets the threshold > 15%.

[0123] Space temperature gradient analysis:

[0124] Define the overheating zone determination rule: With the temperature sensor as the center, calculate the temperature standard deviation σ in the 3x3 grid. When σ > 3℃ and the temperature of the center sensor > the preset upper limit (set temperature + 20℃), lock the area as a physical overheating zone.

[0125] Food deformation monitoring: The judgment is aided by the sudden change in the slope of the weight sensor data (moisture evaporation rate > 0.5g / s), avoiding misjudgment caused by sensor failure.

[0126] (II) Intelligent directional cooling system

[0127] 1. Hardware Architecture Design

[0128] Components of the air conditioning system:

[0129] Three sets of rotatable fans are deployed on each side (speed 0-3000RPM, angle adjustment range ±45°), and equipped with NTC temperature sensors to monitor the cold air outlet temperature in real time (controlled at 25±2℃).

[0130] The air duct adopts a guide fin structure, which ensures that the cold air coverage area error is <2cm and the wind speed uniformity is >90%.

[0131] Actuator control: The fan angle is driven by a stepper motor, with an angle adjustment accuracy of 1° and a response time of <500ms; the fan speed is controlled by pulse width modulation technology, with a minimum adjustment unit of 10RPM.

[0132] Time sequence control: The "pulse cooling" strategy is adopted, with each air supply lasting 10-30 seconds and a 5-second interval for detecting temperature changes to avoid excessive temperature difference between the inside and outside of the food due to sudden drops in local temperature; when the temperature of the overheated area drops to the set temperature +10℃, it automatically switches to low-speed maintenance mode (wind speed 2m / s).

[0133] (III) Feedback Mechanism for Processing Results

[0134] During the cooling process, the oven status prediction index is updated in real time, refreshing every 2 seconds; if the overheated area is not eliminated within 1 minute (the area of ​​the charred spot does not decrease or the temperature does not drop), a secondary response is triggered: the overall heating power is reduced by 30%, and the user is prompted on the display screen that "the food is locally overheated and the protection mode has been activated".

[0135] After processing, an abnormal event log is generated, which includes the time of overheating, the location of the affected area, the processing time, and the temperature change curve, for users to view and for system optimization training.

[0136] Its beneficial effects lie in obtaining multimodal oven data by acquiring temperature, humidity, weight, and image data from the internal sensor array of the oven; preprocessing the multimodal oven data to obtain initial multimodal oven data; extracting local spatial features based on a CNN convolutional neural network and identifying color changes in the oven food, using a Transformer to capture global temporal dependencies, and integrating multimodal data to predict temperature curves to establish a CNN-Transformer oven state prediction model; inputting the initial multimodal oven data into the CNN-Transformer oven state prediction model for identification to obtain the oven predicted state; judging based on the oven predicted state, if it is determined to be an abnormal temperature fluctuation, an emergency power cut is triggered and the system switches to PID standby mode; if it is determined to be local overheating of the food, the side cooling system is activated to directionally cool the oven interior. 1. Compared with the traditional single temperature control method, the temperature curve can be dynamically adjusted according to the characteristics of the food and the temperature and humidity distribution inside the oven, significantly improving the accuracy and consistency of cooking and meeting the diverse needs of users for different foods and different cooking effects. 2. Based on the oven's status prediction index, the system can promptly identify abnormal temperature fluctuations and localized overheating of food, and quickly take corresponding measures, shifting from passive control to proactive intelligent decision-making, effectively avoiding cooking failures caused by improper temperature control. 3. When abnormal temperature fluctuations are detected, the heating power is immediately cut off and PID control is activated to prevent temperature runaway and potential fires or other safety accidents. For locally overheated food, a lateral cooling system provides targeted cooling to prevent over-burning and the generation of harmful substances, while simultaneously protecting the oven's internal structure from high-temperature damage, ensuring comprehensive user safety. 4. It can adapt to the cooking needs of different types and quantities of food, providing precise temperature control whether baking bread, meat, or pastries. Furthermore, the intelligent control and rapid response functions reduce the tedious manual adjustments required by users. Users only need to set basic cooking parameters, and the oven will automatically optimize the temperature control process, providing a convenient, efficient, and intelligent cooking experience.

[0137] Please see Figure 2 In an adaptive temperature control method for an electric oven, the method includes the following steps:

[0138] Temperature, humidity, weight, and image data from the sensor array inside the oven are acquired to obtain multimodal oven data; the multimodal oven data is preprocessed to obtain initial multimodal oven data;

[0139] Based on CNN convolutional neural network to extract local spatial features and identify color changes in oven food, Transformer is used to capture global temporal dependencies, and multimodal data is integrated to predict temperature curves, thus establishing a CNN-Transformer oven state prediction model.

[0140] The initial multimodal oven data is input into the CNN-Transformer oven state prediction model for identification, and the predicted oven state is obtained.

[0141] The oven's predicted status is used to determine if there is an abnormal temperature fluctuation. If so, an emergency power outage is triggered and the oven switches to PID standby mode.

[0142] If it is determined that the food is overheated in a localized area, the side cooling system will be activated to cool the interior of the oven in a targeted manner.

[0143] Please see Figure 3 This is a schematic diagram of an electric oven structure, which is an adaptive temperature control system for an electric oven.

[0144] 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 preferred examples and are not intended to limit 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 the present invention is defined by the appended claims and their equivalents.

Claims

1. An adaptive temperature control system for an electric oven, characterized in that, The temperature control system of the electric oven includes the following modules: The oven data acquisition module is used to acquire temperature data, humidity data, weight data, and image data from the sensor array inside the oven to obtain multimodal oven data; the multimodal oven data is preprocessed to obtain initial multimodal oven data; The prediction model building module is used to extract local spatial features based on CNN convolutional neural network, identify color changes in oven food, capture global temporal dependencies using Transformer, integrate multimodal data and predict temperature curves, and build a CNN-Transformer oven state prediction model. The oven state prediction module is used to input the initial multimodal oven data into the CNN-Transformer oven state prediction model for identification, and obtain the oven predicted state. The oven temperature control module is used to make judgments based on the predicted state of the oven. If an abnormal temperature fluctuation is detected, an emergency power outage is triggered and the system switches to PID standby mode. The oven cooling control module is used to activate the side cooling system to cool the interior of the oven in a targeted manner if it is determined that the food is locally overheated.

2. The adaptive temperature control system for an electric oven as described in claim 1, characterized in that, The oven data acquisition module includes the following units: The temperature data processing unit is used to use a median filtering algorithm to take five consecutive time points of data collected by each temperature sensor to form a sliding window, calculate the median of the data within the window as the effective temperature value at the current moment, and obtain the initial temperature data. A humidity data processing unit is used to perform mean filtering on humidity data to obtain filtered humidity data, and to perform normalization processing on the filtered humidity data to obtain initial humidity data. The weight data processing unit is used to fit the trend term of the weight data using the least squares method, subtract the trend term from the original data to obtain weight fluctuation data, and standardize the weight fluctuation data to obtain initial weight data. The image data processing unit is used to perform grayscale processing on the image, convert the color image into a grayscale image, enhance the grayscale image, and improve the contrast of the image using a histogram equalization method to obtain initial image data.

3. The adaptive temperature control system for an electric oven as described in claim 1, characterized in that, The CNN convolutional neural network includes: The model building layer is used to construct a multi-layer CNN convolutional neural network, where the input layer is used to receive preprocessed image data; The first convolutional layer uses 32 3x3 convolutional kernels with a stride of 1 and padding of 1 to perform convolution operations on the input image and extract local features of edges and textures. The first pooling layer is used for the max pooling method. The pooling kernel size is 2x2 and the stride is 2. It downsamples the output of convolutional layer 1 to reduce the amount of data and retain the main features. The second convolutional layer uses 64 3x3 convolutional kernels with a stride of 1 and padding of 1 to convolve the output of pooling layer 1 again. The color recognition layer is used to establish a color recognition layer by connecting a fully connected layer after the convolutional layer, corresponding to the three channels of the RGB color space.

4. The adaptive temperature control system for an electric oven as described in claim 1, characterized in that, The prediction model building module includes: The data integration unit is used to integrate temperature data, humidity data, weight data, and image features extracted by CNN according to time series to obtain multimodal time series data; A data input unit is used to input the multimodal time series data into a Transformer model, wherein the Transformer model includes at least an encoder and a decoder; The data computation unit is used to calculate the attention scores between multimodal data at different time points through a self-attention mechanism.

5. The adaptive temperature control system for an electric oven as described in claim 1, characterized in that, The oven temperature control module includes: The status judgment unit is used to determine if the oven prediction status is abnormal, and then calculate the root mean square error (RMSE) between the predicted temperature curve and the set normal temperature curve. The threshold calculation unit is used to determine abnormal temperature fluctuations if the root mean square error of RMSE exceeds a preset threshold.

6. The adaptive temperature control system for an electric oven as described in claim 1, characterized in that, The oven temperature control module includes: The mode switching unit is used to immediately trigger an emergency power-off mechanism when abnormal temperature fluctuations are detected, cutting off the heating power supply to the electric oven and switching to PID standby mode. The temperature regulation unit is used to calculate the temperature control quantity based on the deviation between the current temperature feedback value and the set value using a PID controller, generate an adjustment command based on the temperature control quantity, and adjust the power of the heating element and the fan speed based on the adjustment command.

7. The adaptive temperature control system for an electric oven as described in claim 1, characterized in that, The oven cooling control module includes: The directional cooling unit is used to activate the side cooling system to cool the inside of the oven in a directional manner when it is determined that the food is locally overheated. The side cooling system consists of adjustable-angle fans and cold air ducts installed on both sides of the oven. The cooling adjustment unit is used to control the system to generate cooling adjustment commands based on the local overheating location determined by image recognition. It automatically adjusts the angle and speed of the fan to direct the cold air towards the overheated area and quickly reduce the temperature of that area.

8. A method for adaptively adjusting the temperature of an electric oven, characterized in that, The electric oven temperature control method includes the following steps: Temperature data, humidity data, weight data, and image data from the sensor array inside the oven are acquired to obtain multimodal oven data; the multimodal oven data is preprocessed to obtain initial multimodal oven data; Based on CNN convolutional neural network to extract local spatial features and identify color changes in oven food, Transformer is used to capture global temporal dependencies, and multimodal data is integrated to predict temperature curves, thus establishing a CNN-Transformer oven state prediction model. The initial multimodal oven data is input into the CNN-Transformer oven state prediction model for identification, and the predicted oven state is obtained. Based on the oven's predicted status, if an abnormal temperature fluctuation is detected, an emergency power outage is triggered and the system switches to PID standby mode. If it is determined that the food is overheated in a localized area, the side cooling system will be activated to cool the interior of the oven in a targeted manner.

9. The adaptive temperature control method for an electric oven as described in claim 8, characterized in that, The step of judging based on the oven's predicted state, and if an abnormal temperature fluctuation is determined, triggering an emergency power outage and switching to PID standby mode, includes: If an abnormal temperature fluctuation is detected, the system immediately triggers an emergency power-off mechanism, cutting off the heating power to the electric oven and switching to PID standby mode. The PID controller calculates the temperature control quantity based on the deviation between the current temperature feedback value and the set value, generates an adjustment command based on the temperature control quantity, and adjusts the power of the heating element and the fan speed based on the adjustment command.

10. The adaptive temperature control method for an electric oven as described in claim 8, characterized in that, If it is determined that the food is locally overheated, the side cooling system will be activated to cool the interior of the oven in a targeted manner, including: If it is determined that the food is locally overheated, the side cooling system will be activated to cool the inside of the oven in a directional manner. The side cooling system consists of adjustable fans and cooling ducts installed on both sides of the oven. Based on the location of localized overheating determined by image recognition, the control system generates a cooling adjustment command, automatically adjusts the angle and speed of the fan, and directs the cool air toward the overheated area to quickly reduce the temperature of that area.

Images