Food steaming oven steam intelligent distribution control method and system

Abstract

The present application relates to a kind of food steaming stove steam intelligent distribution control method and system, belong to food processing control technical field.The method includes: automatic identification material information, based on cooking curve database for each partition matches corresponding target cooking curve;Real-time acquisition of the state information of partition, the validity of the collected image information is judged, and the current state vector is output;Current state information is compared with corresponding target cooking curve, and the basic heat demand and target steam quality parameter are obtained;Then based on the thermal interference model between partitions, the basic heat demand is corrected, and the final heat demand value is obtained;According to the final heat demand value and target steam quality parameter of each partition, the steam quality delivered to each partition is dynamically adjusted, and the flow corresponding to each partition is controlled.The present application guarantees the reliability of perception through visual anti-interference, improves the accuracy and energy efficiency of partition coordination by thermal decoupling global optimization, and realizes on-demand steam supply by independently adjusting steam quality.

Description

Technical Field

[0001] This invention belongs to the field of food processing control technology, specifically relating to a method and system for intelligent steam distribution control in a food steamer. Background Technology

[0002] Traditional food steamers typically use a single steam source to supply steam to the entire cavity, which makes it difficult to meet the different steam requirements when cooking different ingredients in multiple zones at the same time.

[0003] While existing technologies have proposed solutions for independent steam supply to different zones, most of them rely solely on temperature as a control parameter, which has the following shortcomings: The lack of perception of ingredient types, initial state, and real-time cooking process makes it impossible to achieve true on-demand steam supply; ignoring the coupling effects of heat conduction and heat radiation between adjacent zones leads to decreased control precision and increased energy consumption; controlling only the steam flow rate without considering the key impact of steam quality (dryness / superheat) on cooking results makes it difficult to provide the most suitable cooking medium for different ingredients; in a steam-rich, enclosed environment, visual perception is easily interfered with by water mist, resulting in distorted image information and affecting the accuracy of decision-making.

[0004] Therefore, there is an urgent need for an intelligent distribution control method and system that can integrate multi-source sensing, decouple thermal interference, and dynamically adjust steam quality. Summary of the Invention

[0005] To address the aforementioned problems in the existing technology, this invention provides a method and system for intelligent steam distribution control in a food steamer. It ensures perception reliability through visual anti-interference, improves the accuracy and energy efficiency of zone coordination through thermal decoupling and global optimization, achieves on-demand steam supply by independently adjusting steam quality, and continuously optimizes control performance through closed-loop self-evolution.

[0006] The objective of this invention can be achieved through the following technical solutions: The first aspect of this disclosure provides a method for intelligent steam distribution control in a food steamer, comprising the following steps: S1. Parameter preset and ingredient recognition: The ingredient information is automatically recognized by the image acquisition unit in each zone, and the target cooking curve is matched for each zone based on the cooking curve database. S2. Acquire multi-source real-time information: Collect the status information of each partition in real time, judge the validity of the collected image information, and output the current status vector; S3. Demand Analysis and Control Decision: Compare the current status information of each zone with the corresponding target cooking curve to obtain the basic heat demand and target steam quality parameters; then, based on the thermal interference model between zones, correct the basic heat demand to obtain the final heat demand value. S4. Dynamic adjustment and allocation execution: Based on the final heat demand value and target steam quality parameters of each zone, dynamically adjust the steam quality delivered to each zone, and independently control the opening degree or flow rate of the steam regulation unit corresponding to each zone. S5. Closed-loop feedback and iteration: After the dynamic adjustment and allocation execution completes the steam allocation execution for each zone, it immediately returns to the step of obtaining real-time information from multiple sources, and starts a new round of closed-loop iteration with a preset control cycle, repeating the iteration until the zone meets the cooking termination condition.

[0007] Furthermore, the parameter preset and ingredient recognition include the following steps: S11. Construct cooking curves: Obtain cooking process parameters of ingredients in real time through the cloud, and construct a cooking curve database, including the change curves of temperature, humidity, and steam quality parameters over time, as well as the corresponding visual feature thresholds. S12. Initial Image Acquisition and Preprocessing: The cameras in each independent zone automatically acquire the initial image of the current food and preprocess the acquired raw images. S13. Multi-dimensional identification of ingredients based on deep learning: Input the pre-processed image into a pre-trained deep learning model to obtain the ingredient type, laying quantity estimation and initial state evaluation results; S14. Target cooking curve matching: Based on the ingredient type, estimated amount, and initial state assessment results, retrieve the most matching cooking curve from the cooking curve database.

[0008] Furthermore, the acquisition of multi-source real-time information includes the following steps: S21. Synchronous acquisition of multi-source status information: According to the preset sampling frequency, synchronously acquire status information including temperature, humidity, pressure and image; S22. Image Information Validity Judgment: For each frame of acquired image information, the image preprocessing module is invoked to perform a validity judgment to detect whether there is image distortion caused by steam condensation and water mist. This includes the following steps: Image quality feature extraction: The original image is preprocessed to extract quantitative indicators for evaluating image distortion. The quantitative indicators include blur and color features. The blur index is obtained by calculating the variance of the Laplacian operator of the image. The color feature index is characterized by converting the image from the RGB color space to the HSV color space and calculating the average value of all pixels in the H channel. Distortion determination and labeling: The extracted quantization index is compared with the preset validity threshold range: if the Laplacian variance is lower than the preset first threshold, the image is determined to be distorted due to excessive blurring; if the average hue value of the H channel continuously exceeds the preset second threshold range, the image is determined to be distorted due to color drift. If both of the above conditions are met at the same time, it is judged as serious distortion; When an image is determined to be distorted, the image data of that frame is marked, and its distortion type and timestamp are recorded. S23. Image Distortion Compensation: For image information that is determined to be distorted, a preset compensation strategy is activated. The Kalman filter is called to extract the visual feature sequence based on the previous consecutive frames of images that were determined to be valid in the same partition, predict the visual feature value at the current moment, and use the predicted value as the current valid visual information. S24. Current state vector output: If the image is determined to be valid, extract the visual features used to characterize the cooking state of the ingredients from the image; if the image is determined to be distorted, output the predicted visual features obtained by image distortion compensation; fuse the visual features of each partition at the current moment with the synchronously collected temperature, humidity and pressure data to construct a multi-dimensional current state vector.

[0009] Furthermore, the aforementioned demand analysis and control decision-making includes the following steps: S31. Multi-dimensional control deviation calculation: Compare the current state vector with the target value of the target cooking curve of the corresponding partition at the corresponding time step by step to calculate the control deviation vector at the current time. S32. Single-zone basic demand decision: Based on the control deviation vector and combined with the characteristics of the types of ingredients in the zone, obtain the basic heat demand and target steam quality required for each zone. S33. Demand Adjustment Based on Thermal Interference Model: By establishing a thermal interference model, the basic heat demand is adjusted, including: A thermal coupling coefficient matrix between zones is established in advance through experiments, and the actual temperature change rate of each zone is established. Based on the actual temperature change rate of each zone, the basic heat demand is decoupled and corrected, so that the zones achieve the expected temperature rise effect while taking into account the heat contribution of adjacent zones. S34. Global Constraint Optimization: Based on the preliminary heat demand of each zone after correction, obtain the maximum steam supply power of the steam generating unit at the current moment, and set the minimum steam supply of each zone as a constraint condition; construct an optimization function with the goal of minimizing total energy consumption and minimizing tracking deviation; use linear programming, quadratic programming or heuristic optimization algorithm to solve the optimization problem and obtain the final heat demand value of each zone. S35, Decision Output: Package the final heat demand value and target steam quality parameters to generate an execution instruction vector.

[0010] Furthermore, the acquisition of the basic heat demand includes: A multi-input multi-output PID controller is used, taking temperature deviation, humidity deviation and visual characteristic deviation as inputs, and outputting the basic heat demand of the zone. The acquisition of the target steam quality includes: Based on the identified ingredient type and the current cooking stage, the expected steam quality value is read from the target cooking curve, and the target steam quality is obtained based on the dynamic correction of the rule engine. Among them, dynamic correction based on the rule engine includes: Calculate the deviation between the current temperature and the target temperature, and use a vision module to identify the deviation between the surface color of the food and the target color; The deviation is input into the rule base, and temperature hysteresis compensation, visual feedback adjustment and anti-water accumulation adjustment are performed based on preset rules to correct the expected value of steam quality and obtain the target steam quality.

[0011] Furthermore, the dynamic adjustment and allocation execution includes the following steps: S41. Dynamic adjustment of steam quality: An independent micro quality adjustment unit is set on the steam branch corresponding to each zone. The quality adjustment unit on the corresponding branch is controlled according to the target steam quality parameters of each zone so that the steam entering each zone reaches its respective target quality. S42. Heat demand conversion: Based on the actual steam quality parameters entering the zone, look up the steam thermodynamic properties table to obtain the specific enthalpy of the steam and the specific enthalpy of the saturated water under the corresponding steam quality, and calculate the target mass flow rate required for the zone. S43. Steam Flow Regulation Unit Control: The target mass flow rate is sent as the set value to the steam flow regulation unit of the corresponding zone. The steam flow regulation unit adopts a proportional regulating valve and controls the valve opening through analog signals to achieve continuous flow regulation.

[0012] Furthermore, the target mass flow rate required for the computational partition is expressed as: ; In the formula, For target quality flow rate, This is the final heat requirement value. It represents the sum of latent heat and sensible heat released by the condensation of a unit mass of steam.

[0013] Furthermore, the cooking termination condition includes: Time condition: The preset total time to reach the target cooking curve at the current moment; Temperature conditions: The core temperature of the food reaches the target value and is maintained for a period of time exceeding the set threshold; Visual conditions: The deviation between the current visual features and the target visual features is less than the preset allowable range and remains stable; User intervention: The user issues a termination command through the human-computer interface.

[0014] The second aspect of this disclosure provides a smart steam distribution control system for a food steamer, which executes the smart steam distribution control method for a food steamer as described above, including a multi-source information acquisition module, an image preprocessing and compensation module, a main controller, a steam quality adjustment unit, a steam flow adjustment unit, and a human-machine interaction module; The multi-source information acquisition module is set in each independent cooking zone and includes a temperature sensor, a humidity sensor, a pressure sensor and an image acquisition unit, used to collect temperature, humidity, pressure and food image information in the zone in real time; The image preprocessing and compensation module is connected to the image acquisition unit and is used to determine the validity of the acquired image, extract the ambiguity index and color feature index and compare them with a preset threshold to determine whether the image is distorted due to steam and water vapor; for the distorted image, the Kalman filter is called to perform prediction compensation based on the historical valid visual feature sequence, and the valid visual features at the current moment are output. The steam quality adjustment unit is located at the inlet of each zone's steam branch and includes a miniature superheater and a micro-spray nozzle. It is used to independently adjust the dryness or superheat of the corresponding branch steam according to the target steam quality parameters output by the main controller, so that the steam entering each zone reaches the required thermodynamic state. The steam flow regulation unit is installed in each zone's steam branch and uses a proportional regulating valve to calculate the target mass flow rate based on the final heat demand value output by the main controller and the specific enthalpy corresponding to the current steam quality, and to achieve dynamic regulation of the steam flow rate by controlling the valve opening. The human-computer interaction module is connected to the main controller and is used to receive cooking instructions, ingredient information confirmation, and cooking preference settings input by the user, and to display the cooking status of each zone and completion prompts.

[0015] In a preferred embodiment of the present invention, the main controller is connected to a multi-source information acquisition module, an image preprocessing and compensation module, a steam quality regulation unit, and a steam flow regulation unit, respectively, and is used to execute the following control logic: Target curve matching: Based on the type, quantity, and initial state of ingredients identified in the initial image, match or generate personalized target cooking curves for the target cooking curve database; Deviation calculation and demand decision: The current state vector is compared with the target curve dimension by dimension to calculate the deviation of temperature, humidity and visual features. The basic heat demand is obtained through the PID controller, and the target steam quality is determined by combining the type of food and the rule engine. Thermal interference correction: A thermal interference model is established based on a pre-stored thermal coupling coefficient matrix, and the basic heat demand is decoupled and corrected to obtain the preliminary heat demand; Global optimization: Under the constraint of total steam supply power, the optimization problem is solved with the goal of minimizing energy consumption and tracking deviation, and the final heat demand value of each zone is output. Closed-loop iteration: Repeat the above process with a preset control cycle until the cooking termination condition is met; Adaptive optimization: Record data throughout the entire process and update the thermal coupling coefficient matrix, target curve parameters, and controller coefficients periodically.

[0016] The beneficial effects of this invention are as follows: This invention solves the problem of visual sensor distortion in steam environments by acquiring multi-source information in real time and judging visual validity, ensuring the reliability of state perception. It decouples and corrects the thermal coupling between adjacent zones through a thermal interference model, and achieves optimal energy allocation under total power constraints by combining global optimization, improving the accuracy and energy efficiency of multi-zone collaborative control. By independently adjusting the micro-quality units of the steam branches in each zone, the control dimension is expanded from the traditional two-dimensional flow / time to a three-dimensional flow-time-quality, enabling the system to dynamically provide the most suitable steam thermodynamic state for different ingredients (such as pastries needing to be moist and meat needing to be dry). At the same time, through closed-loop feedback and data-driven adaptive optimization, the control model and cooking curve are continuously evolved. Attached Figure Description

[0017] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0018] Figure 1 This is a schematic diagram illustrating the steps of an intelligent steam distribution control method for a food steamer provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the parameter preset and food ingredient recognition steps provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the steps for acquiring multi-source real-time information provided in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the steps of demand analysis and control decision-making provided in an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the steps of dynamic adjustment and allocation execution provided in an embodiment of the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0020] Example 1 This embodiment provides a method for intelligent steam distribution control in a food steamer, such as... Figure 1 As shown, it includes the following steps: S1. Parameter Preset and Ingredient Recognition: Ingredient information is automatically identified through image acquisition units within each zone, and a corresponding target cooking curve is matched for each zone, such as... Figure 2 As shown, it includes the following steps: S11. Construct cooking curves: Obtain cooking process parameters of ingredients in real time through the cloud, and construct a cooking curve database, including the change curves of temperature, humidity, and steam quality parameters over time, as well as the corresponding visual feature thresholds. The data includes: a temperature-time curve describing the ideal temperature within a cooking zone over time, typically stored as discrete time points or a function; a humidity-time curve describing the ideal relative humidity within a zone over time; a steam quality-time curve describing the ideal quality (dryness or superheat) of the steam delivered to the zone over time; and a visual feature threshold sequence containing expected or threshold values ​​for the visual characteristics of the ingredients corresponding to time or cooking stage, such as the RGB range of the ingredient's surface color, shrinkage rate, and degree of oil exudation, used for state comparison during subsequent visual closed-loop control. These thresholds are usually derived through extensive experimental statistics and indexed by timestamps or percentages of cooking progress.

[0021] S12. Initial Image Acquisition and Preprocessing: The cameras in each independent zone automatically acquire the initial image of the current food and preprocess the acquired raw image, including noise reduction, contrast enhancement and geometric correction.

[0022] It should be noted that the lens anti-fog heating is automatically triggered during the initial image acquisition process to avoid image blurring caused by fogging of the cold lens.

[0023] S13. Multi-dimensional identification of ingredients based on deep learning: Input the pre-processed image into a pre-trained deep learning model to obtain the ingredient type, laying quantity estimation and initial state evaluation results; The ingredient category results include the ingredient's class (e.g., "steamed sea bass," "frozen buns," "boiled shrimp," etc.) and a confidence score. If the confidence score is below a set threshold, the system prompts the user to manually confirm or select to ensure the accuracy of the category information. The quantity estimation uses image segmentation technology to identify ingredient areas and, combined with monocular vision or known reference object dimensions, estimates the area, volume, or quantity of ingredients (e.g., the number of buns, the length of fish). The quantity directly affects the total heat required and cooking time. The initial state assessment analyzes the initial physical state of the ingredients, such as freshness (color, gloss), degree of thawing (whether ice crystals remain), and size uniformity. This information can be used to fine-tune the cooking curve; for example, incompletely thawed ingredients require increased preheating time in the low-temperature section.

[0024] Specifically, building a deep learning model includes the following steps: Model architecture: The network structure includes a shared feature extraction backbone network and three task output branches, where: Shared feature extraction backbone network: Employing lightweight convolutional neural networks, such as MobileNetV3 or EfficientNet-Lite, to uniformly scale the RGB images captured by the camera to a shared feature extraction backbone network. Pixels are used as network input, and the output size is The shared feature maps (the specific dimensions depend on the selected network) contain rich spatial and semantic information; The three task output branches include an ingredient type identification branch, a placement quantity estimation branch, and an initial state evaluation branch, where: The food category recognition branch: After sharing the feature map, a global average pooling layer is connected to convert the feature map into a 1280-dimensional feature vector, followed by two fully connected layers with dimensions of 512 and 1212 respectively. ( (This represents the total number of food categories). The last layer uses the Softmax activation function to output the probability distribution of each category; the confidence score for each food category. ,satisfy The system takes the category corresponding to the highest score as the recognition result. If the highest score is lower than a preset threshold... (This embodiment takes) If this happens, the user confirmation process will be triggered.

[0025] The placement estimation branch uses a U-Net or DeepLabV3+ semantic segmentation network, upsampling and skip connections are applied to the shared feature map, and the output is a pixel-level segmentation map with the same resolution as the input image. Each pixel is classified as either "ingredients" or "background". The total number of pixels labeled as "ingredients" in the segmentation map is counted. By combining known camera intrinsic parameters and installation height, the pixel area is converted into the actual physical area using monocular visual geometry. (Unit: cm²): ; in, The distance from the food surface to the camera (obtained through the preset steamer rack height). Where is the lens focal length, and pixel_size is the physical size of a single pixel.

[0026] For discrete ingredients (such as steamed buns and shrimp), a second regression branch is connected, and the quantity estimate is directly output through a fully connected layer. Alternatively, the volume can be estimated by multiplying the average volume of a single ingredient by the quantity, and the regression branch is trained using mean squared error loss.

[0027] Initial state evaluation branch: After sharing the feature map, separate convolutional and fully connected layers are connected to output regression values ​​or classification results for multiple state parameters, including: Freshness rating: Output continuous values 1 indicates the freshest, calculated by the distance between color distribution features (such as RGB histogram) and standard features of fresh ingredients.

[0028] Thawing degree: Three-category output (not thawed / partially thawed / completely thawed), determined by texture features and surface reflectivity, for example, ice crystal areas are characterized as high reflectivity areas.

[0029] Uniformity index: Calculates the variance of color and texture within the segmented region and outputs a uniformity score. .

[0030] This branch employs multi-task learning, with the total loss being the weighted sum of the losses from each subtask.

[0031] Model pre-training: Collect over 100,000 food images, covering the following dimensions: Ingredient variety: at least 50 common steamed ingredients, each including different varieties, colors and sizes.

[0032] Laying conditions: different quantities, arrangement densities, and degrees of overlap.

[0033] Initial state: different freshness (from freshly picked to slightly rotten), different degree of thawing (frozen, partially thawed, fully thawed), different surface humidity (dry, moist, with water droplets).

[0034] Environmental interference: different lighting conditions, presence or absence of water mist on the lens, and different background patterns (simulating steam grid texture).

[0035] The following information is manually labeled for each image: Category label Segmentation mask (Pixel-level labeling of ingredient areas), actual quantity / area values or Status labels: Freshness rating (continuous value), degree of thawing (categorized), and uniformity rating (continuous value).

[0036] The AdamW optimizer is used, with a cosine annealing strategy for learning rate scheduling. The learning rate decays according to a cosine function after each iteration, and early stopping is employed. This is done when the validation set loss is continuous. Stop training if the wheel does not descend.

[0037] Loss function selection: Category recognition branch: Cross-entropy loss In the formula, This represents the total number of food categories, i.e., the number of all possible food categories that the model needs to distinguish (e.g., 50 categories). For the first The true label of a category is an indicator variable, when the true category of a sample is hour, ;otherwise It adopts a one-hot encoding format; The sample predicted by the model belongs to the first... The probability of each class, output by the Softmax function, satisfies... and .

[0038] Split Branch: A combination of Dice loss and cross-entropy loss ;in: ; ; In the formula, The total number of pixels in the image. The model predicts the first The probability that a pixel belongs to the "food" category (usually activated by Sigmoid, with a value range of [0,1]). For the first The real label of each pixel This indicates that the pixel belongs to the food ingredient area. Indicates the background. For smoothing terms (e.g.) The Dice loss is used to avoid a zero denominator and to make the loss function differentiable. Based on the Dice coefficient, the Dice loss measures the degree of overlap between the predicted and actual segmentations. The smaller the loss value, the higher the overlap.

[0039] Lay-out regression branch: Mean squared error loss or In the formula, The number of ingredients predicted by the model (e.g., the number of steamed buns). This represents the actual quantity of ingredients. The area (in cm²) or volume (in cm³) of food to be laid out as predicted by the model. The actual area or volume of the tiled surface is manually marked.

[0040] Condition assessment branch: Freshness score uses mean squared error; thawing degree uses cross-entropy; uniformity score uses mean squared error.

[0041] Total loss: ,in The weights for each task are determined through a grid search.

[0042] The dataset is divided into training, validation, and test sets in an 8:1:1 ratio for training. After each round of training, the accuracy / error of each task is calculated on the validation set to monitor overfitting.

[0043] S14. Target cooking curve matching: Based on the ingredient type, estimated amount, and initial state assessment results, retrieve the most matching cooking curve from the cooking curve database.

[0044] Specifically, retrieving the best-matching cooking curve involves the following strategies: If a matching entry exists in the database (of the same type and within the allowable error range), the standard curve is directly used as the target cooking curve for that partition.

[0045] If there is no perfect match, but multiple curves exist for the same type of ingredient with different layering amounts, the system generates a new curve suitable for the current layering amount through linear interpolation or nonlinear fitting methods. For example, given the cooking curves for 500g and 1000g fish, a curve for 750g fish can be generated using time axis scaling or a heat compensation algorithm.

[0046] For new ingredients not included in the database, the system migrates and corrects the curve of the closest ingredient based on the similarity between the ingredient's attributes (such as density, water content, and fat content) and known ingredients, or generates an initial curve by combining the cooking parameters manually entered by the user (such as "steam over high heat for 15 minutes").

[0047] The curve parameters are further fine-tuned based on the initial state assessment results. For example, if the initial temperature of the food is low (such as when it has just been taken out of the refrigerator), a preheating stage is added to the beginning of the curve; if there is a lot of moisture on the surface of the food, the initial steam dryness is appropriately increased to prevent the surface from becoming soft and mushy.

[0048] Ultimately, each zone obtains a personalized target cooking curve, which uses time as the independent variable and includes a sequence of expected values ​​for temperature, humidity, and steam quality, as well as a series of staged feature thresholds for visual verification.

[0049] S2. Acquire multi-source real-time information: Collect the status information of each partition in real time, determine the validity of the collected image information, and output the current status vector, such as... Figure 3 As shown, it includes the following steps: S21. Synchronous acquisition of multi-source status information: According to the preset sampling frequency, the status information including temperature, humidity, pressure and image is collected synchronously. The sampling frequency of the image is independently set to be lower than the sampling frequency of other status information in order to balance the system's computing load and data real-time performance.

[0050] S22. Image Information Validity Judgment: For each frame of acquired image information, the image preprocessing module is invoked to perform a validity judgment to detect whether there is image distortion caused by steam condensation and water mist. This includes the following steps: Image quality feature extraction: The original image is preprocessed to extract quantitative indicators for evaluating image distortion, including blur and color features.

[0051] It should be noted that the blurriness index is obtained by calculating the variance of the Laplacian operator in the image. The lower the Laplacian variance value, the less edge information the image has, and the higher the degree of blurriness. The color feature index is characterized by converting the image from the RGB color space to the HSV color space and calculating the average value and distribution variance of all pixels in the H channel (hue). When there is severe steam or water vapor, the image appears white overall, the H value distribution narrows, and the mean shifts.

[0052] Distortion assessment and labeling: The extracted quantitative indicators are compared with a preset validity threshold range. If the Laplacian variance is lower than a preset first threshold (e.g., If the image is too blurry, it is determined that the image is distorted. If the average hue value of the H channel continuously exceeds the preset second threshold range (for example, or If the color shift is detected, the image is determined to be distorted due to color drift. If both of the above conditions are met simultaneously, it is judged as severely distorted.

[0053] When an image is determined to be distorted, the image data of that frame is marked, its distortion type (blur distortion / color distortion) and timestamp are recorded, and it is marked as "unusable for visual feature extraction".

[0054] S23. Image Distortion Compensation: For image information that is judged to be distorted, a preset compensation strategy is activated. The Kalman filter is called to extract the visual feature sequence (such as the trend of food color change and the trend of shrinkage rate change) based on the previous consecutive frames of images judged to be valid in the same partition. The predicted value is then used as the current valid visual information.

[0055] It should be noted that in visual feature prediction, the Kalman filter constructs a system state vector containing visual features and their rates of change (such as the rate of change of color values), and uses a constant velocity or acceleration model as the state transition equation to make a prior estimate of the feature values ​​at the current moment. Subsequently, when new valid image frames are acquired and the actual observation values ​​are extracted, the filter calculates the residual between the prior estimate and the actual observation, and combines the Kalman gain to make a weighted correction on the estimated value, thereby outputting the optimal posterior estimate that integrates historical trends and current observations, and realizing real-time compensation for visual feature data that is missing or delayed due to water mist.

[0056] S24. Current state vector output: If the image is determined to be valid, extract the visual features used to characterize the cooking state of the ingredients from the image; if the image is determined to be distorted, output the predicted visual features obtained by image distortion compensation; fuse the visual features (extracted or predicted) of each partition at the current moment with the synchronously collected temperature, humidity and pressure data to construct a multi-dimensional current state vector.

[0057] The visual features include at least: the average color value of the food surface, the uniformity of color distribution, the sharpness of the food outline edge, the gloss of the food surface, or the area of ​​the water accumulation area.

[0058] S3. Demand Analysis and Control Decision: Compare the current status information of each zone with the corresponding target cooking curve to obtain the basic heat demand and target steam quality parameters; then, based on the inter-zone thermal interference model, correct the basic heat demand to obtain the final heat demand value, such as... Figure 4 As shown, it includes the following steps: S31. Multi-dimensional control deviation calculation: The current state vector is compared with the target value of the target cooking curve of the corresponding partition at the corresponding time in a dimension-by-dimensional manner to calculate the control deviation vector at the current time, including temperature deviation, humidity deviation and visual feature deviation. Specifically, assuming the current state vector ,time target value Then the temperature deviation , humidity deviation Visual feature deviation ,in This is a feature distance function used to quantify the difference between the current visual features and the target visual features (e.g., the Euclidean distance in the color space).

[0059] in, For the first i The temperature of the zone, For the first i Humidity of each zone For the first i Partition pressure, For the first i The visual feature vector of the partition. For the target temperature, For target humidity, For target visual features.

[0060] S32. Single-zone basic demand decision: Based on the control deviation vector and combined with the characteristics of the types of ingredients in the zone, obtain the basic heat demand and target steam quality required for each zone.

[0061] The acquisition of basic calorie requirements includes: A multi-input multi-output PID controller is used to control temperature deviation. , humidity deviation and visual feature deviation As input, the output zone's basic heat demand .

[0062] Specifically, taking a PID controller as an example, its calculation formula is as follows: In the formula, For the PID coefficients of temperature control, and The weighting coefficients for humidity and visual characteristics are dynamically adjusted based on the cooking characteristics of different ingredients.

[0063] Target steam quality acquisition includes: Based on the identified ingredient type and the current cooking stage, the expected steam quality value is read from the target cooking curve, and the target steam quality is obtained based on the dynamic correction of the rule engine.

[0064] The steam quality parameters are expressed by steam dryness (range 0-1) or superheat (unit: °C). For example, for pasta-type foods (such as steamed buns), a high humidity environment is required in the initial stage of steaming to prevent the surface from drying and cracking; the target steam dryness is set to... For meat products (such as ribs), high-temperature dry steaming is required in the later stages to force out the fat; the target superheat is set to [value missing]. (That is, the steam temperature is 20°C higher than the saturation temperature at the current pressure).

[0065] Among them, dynamic correction based on the rule engine includes: Calculate the deviation between the current temperature and the target temperature, and use a vision module to identify the deviation between the surface color of the food and the target color; The deviation is input into the rule base, and temperature hysteresis compensation, visual feedback adjustment, and anti-water accumulation adjustment are performed based on preset rules to correct the expected value of steam quality and obtain the target steam quality.

[0066] Specifically, the preset rules are as follows: Rule Example 1 (Temperature Lag Compensation): IF (Cooking Stage = "Rapid Heating Period") AND ( (And lasts for 30 seconds) THEN Target Steam Quality = Preset value + 10% superheat (temporarily increases steam temperature to accelerate heating); Rule Example 2 (Visual Feedback Adjustment): IF (Ingredient Type = "Steamed Fish") AND (Surface Color) Browning rate > threshold) THEN = Preset dryness value - 5% (reduces dryness, increases humidity, and prevents the surface from becoming too dry). Rule Example 3 (Anti-water accumulation adjustment): IF (visual recognition of water accumulation area at the bottom of the zone > threshold) AND (food type = "baozi") THEN = Preset dryness value + 8% (increases dryness and reduces condensation).

[0067] S33. Demand Adjustment Based on Thermal Interference Model: By establishing a thermal interference model, the basic heat demand is adjusted to eliminate the coupling effect between adjacent zones, including: The thermal coupling coefficient matrix between partitions was established in advance through experiments. The matrix is matrix( (total number of partitions), where elements Indicates the first Partition to the first The thermal impact weight of the partition. The actual temperature change rate of the zone can be expressed as: ; In the formula, For the first The heat capacity coefficient of the zone, For actual supply The heat of each zone The temperature rise caused by the steam supply to this zone The contribution to heat exchange caused by temperature difference between adjacent zones (positive value indicates heat transfer from other zones to this zone).

[0068] A decoupling correction is applied to the basic heat demand. The goal of this correction is to ensure that, while considering the heat contributions from adjacent zones, the first... The zones achieved the expected temperature rise. Revised preliminary heat demand. satisfy: ; When adjacent partitions The temperature is higher than that of this zone hour, This indicates a partition. To partition Heat was transferred, so the system actively reduced the impact on the partition. Steam supply ( Energy saving can be achieved by utilizing this thermal interference; conversely, when the temperature of an adjacent zone is lower than that of this zone, the steam supply can be increased to compensate for the heat loss.

[0069] It should be noted that the thermal coupling coefficient matrix The determination was carried out through a step response experiment combined with steady-state thermal equilibrium analysis: First, only a single target partition was considered. Apply constant heating power While maintaining zero input for other partitions, record the temperature rise of all partitions when they reach steady state. Based on the heat balance equation (Or by fitting the transient response curve) solve for other partitions. For partitions thermal coupling coefficient ; Repeat the above experiment, using all partitions as heat sources in sequence, and finally obtain all the calibrated coefficients. Assembled by row and column indexes into complete Coupling matrix diagonal elements It is usually normalized to 1 or separately calibrated according to the self-heating effect.

[0070] S34. Global Constraint Optimization: Based on the corrected preliminary heat demand of each zone, obtain the maximum steam supply power of the steam generating unit at the current moment, and set the minimum steam supply of each zone (to prevent the control dead zone from being caused by the valves being completely closed) as a constraint; construct an optimization function with the objectives of minimizing total energy consumption and minimizing tracking deviation: ; in, The final heat requirement to be solved; To apply The predicted temperature at the next moment; , which is a weighting factor used to balance the importance between minimizing energy consumption and temperature tracking accuracy (the second term). For the future The target temperature at any given time. To control the period or predict the length of the time domain.

[0071] The constraints include: Total steam supply does not exceed the limit; Steam supply to each zone shall not be lower than the minimum value.

[0072] Under the premise of satisfying the constraints, the above optimization problem is solved using linear programming, quadratic programming, or heuristic optimization algorithms (such as particle swarm optimization) to obtain the final heat demand values ​​for each zone. The solution process will not be described in detail here.

[0073] S35, Decision Output: Package the final heat demand value and target steam quality parameters to generate an execution instruction vector.

[0074] S4. Dynamic Adjustment and Distribution Execution: Based on the final heat demand and target steam quality parameters of each zone, dynamically adjust the steam quality delivered to each zone, and independently control the opening or flow rate of the steam regulating unit corresponding to each zone, such as... Figure 5 As shown, it includes the following steps: S41. Dynamic adjustment of steam quality: Each zone is equipped with an independent micro quality adjustment unit (such as a micro superheater or micro spray nozzle) on the corresponding steam branch. The quality adjustment unit on the corresponding branch is controlled according to the target steam quality parameters of each zone, so that the steam entering each zone reaches its respective target quality. Understandably, the micro-quality control unit is typically integrated at the steam branch inlet of each zone, consisting of two core actuators—a micro-electric superheater and a micro-spray nozzle—connected in parallel or series. Its working principle is as follows: based on the deviation between the target steam quality parameters of the zone and the current reference steam quality entering the branch, the system independently controls the heating power of the micro-superheater to increase the steam superheat or dryness, while simultaneously controlling the opening of the micro-spray nozzle to inject atomized water to reduce dryness. Through a PID algorithm, these two actuators are decoupled and coordinated, ensuring that the final steam quality entering the zone tracks the target value in real time. This allows each zone to achieve a completely personalized steam thermodynamic state while sharing the same steam source.

[0075] S42. Heat Demand Conversion: Based on the actual steam quality parameters entering the zone, consult the steam thermodynamic properties table to obtain the specific enthalpy of the steam at the corresponding steam quality. and the specific enthalpy of saturated water The sum of latent heat and sensible heat released by the condensation of a unit mass of steam is: .

[0076] Calculate the first Target quality flow rate required for each partition for: ; The above process transforms energy demand into flow control targets.

[0077] S43, Steam Flow Regulation Unit Control: Controls the target mass flow rate. As a set value, it is sent to the corresponding number. Each zone has a steam flow regulation unit. The steam flow regulation unit uses a proportional control valve, which controls the valve opening through an analog signal and uses the valve characteristic curve to achieve continuous flow regulation.

[0078] S5. Closed-loop feedback and iteration: After the dynamic adjustment and allocation execution completes the steam allocation execution for each zone, immediately return to the step of obtaining real-time information from multiple sources, and start a new round of closed-loop iteration with a preset control cycle (e.g., 1 second), repeating the iteration until the zone meets the cooking termination condition.

[0079] It should be noted that in each closed-loop iteration, the system determines in real time whether each zone meets the cooking termination conditions. The termination conditions include, but are not limited to: Time condition: Current moment The preset total time to reach the target cooking curve; Temperature conditions: The core temperature of the food reaches the target value and is maintained for a period of time exceeding the set threshold; Visual conditions: The deviation between the current visual features and the target visual features is less than the preset allowable range (e.g., color value difference is less than 5%), and remains stable. User intervention: The user issues a termination command through the human-computer interface.

[0080] Example 2 This embodiment also provides an intelligent steam distribution control system for a food steamer, including a multi-source information acquisition module, an image preprocessing and compensation module, a main controller, a steam quality adjustment unit, a steam flow adjustment unit, and a human-machine interaction module.

[0081] The multi-source information acquisition module is set in each independent cooking zone and includes a temperature sensor, a humidity sensor, a pressure sensor and an image acquisition unit, used to collect temperature, humidity, pressure and food image information in the zone in real time.

[0082] The image preprocessing and compensation module is connected to the image acquisition unit and is used to determine the validity of the acquired image, extract the ambiguity index and color feature index and compare them with the preset threshold to determine whether the image is distorted due to steam and water mist; for the distorted image, the Kalman filter is called to perform prediction compensation based on the historical valid visual feature sequence, and the valid visual features at the current moment are output.

[0083] The main controller is connected to the multi-source information acquisition module, the image preprocessing and compensation module, the steam quality regulation unit, and the steam flow regulation unit, and is used to execute the following control logic: Target curve matching: Based on the type, quantity, and initial state of ingredients identified in the initial image, match or generate personalized target cooking curves for the target cooking curve database; Deviation calculation and demand decision: The current state vector is compared with the target curve dimension by dimension to calculate the deviation of temperature, humidity and visual features. The basic heat demand is obtained through the PID controller, and the target steam quality is determined by combining the type of food and the rule engine. Thermal interference correction: A thermal interference model is established based on a pre-stored thermal coupling coefficient matrix, and the basic heat demand is decoupled and corrected to obtain the preliminary heat demand; Global optimization: Under the constraint of total steam supply power, the optimization problem is solved with the goal of minimizing energy consumption and tracking deviation, and the final heat demand value of each zone is output. Closed-loop iteration: Repeat the above process with a preset control cycle until the cooking termination condition is met; Adaptive optimization: Record data throughout the entire process and update the thermal coupling coefficient matrix, target curve parameters, and controller coefficients periodically.

[0084] The steam quality adjustment unit is located at the inlet of each zone's steam branch and includes a miniature superheater and a micro-spray nozzle. It is used to independently adjust the dryness or superheat of the corresponding branch steam according to the target steam quality parameters output by the main controller, so that the steam entering each zone reaches the required thermodynamic state.

[0085] The steam flow regulation unit is installed in each zone's steam branch and uses a proportional regulating valve. It is used to calculate the target mass flow rate based on the final heat demand value output by the main controller and the specific enthalpy corresponding to the current steam quality, and to achieve dynamic regulation of the steam flow rate by controlling the valve opening.

[0086] The human-computer interaction module is connected to the main controller and is used to receive cooking instructions, ingredient information confirmation, and cooking preference settings input by the user, and to display the cooking status of each zone and completion prompts.

[0087] This invention solves the problem of visual sensor distortion in steam environments by acquiring multi-source information in real time and judging visual validity, ensuring the reliability of state perception. It decouples and corrects the thermal coupling between adjacent zones through a thermal interference model, and achieves optimal energy allocation under total power constraints by combining global optimization, improving the accuracy and energy efficiency of multi-zone collaborative control. By independently adjusting the micro-quality units of the steam branches in each zone, the control dimension is expanded from the traditional two-dimensional flow / time to a three-dimensional flow-time-quality, enabling the system to dynamically provide the most suitable steam thermodynamic state for different ingredients (such as pastries needing to be moist and meat needing to be dry). At the same time, through closed-loop feedback and data-driven adaptive optimization, the control model and cooking curve are continuously evolved.

[0088] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for intelligent steam distribution control in a food steamer, characterized in that: Includes the following steps: Parameter preset and ingredient recognition: The ingredient information is automatically identified by the image acquisition unit in each zone, and the corresponding target cooking curve is matched for each zone based on the cooking curve database; Acquire multi-source real-time information: Collect the status information of each partition in real time, judge the validity of the collected image information, and output the current status vector; Demand analysis and control decision-making: The current status information of each zone is compared with the corresponding target cooking curve to obtain the basic heat demand and target steam quality parameters; then, based on the thermal interference model between zones, the basic heat demand is corrected to obtain the final heat demand value. Dynamic adjustment and allocation execution: Based on the final heat demand value and target steam quality parameters of each zone, dynamically adjust the steam quality delivered to each zone, and independently control the opening degree or flow rate of the steam regulating unit corresponding to each zone. Closed-loop feedback and iteration: After the dynamic adjustment and allocation execution completes the steam distribution execution for each zone, it immediately returns to the step of obtaining real-time information from multiple sources, and starts a new round of closed-loop iteration with a preset control cycle, repeating the iteration until the zone meets the cooking termination conditions.

2. The intelligent steam distribution control method for a food steamer according to claim 1, characterized in that: The parameter preset and ingredient recognition include the following steps: Constructing cooking curves: By acquiring the cooking process parameters of ingredients in real time through the cloud, a cooking curve database is constructed, including the change curves of temperature, humidity, and steam quality parameters over time, as well as the corresponding visual feature thresholds. Initial image acquisition and preprocessing: The cameras in each independent zone automatically acquire the initial image of the current food and preprocess the acquired raw images; Multi-dimensional food identification based on deep learning: The pre-processed image is input into a pre-trained deep learning model to obtain the food type, the estimated amount of food to be laid, and the initial state evaluation results. Target cooking curve matching: Based on the ingredient type, estimated amount, and initial state assessment results, the most matching cooking curve is retrieved from the cooking curve database.

3. The intelligent steam distribution control method for food steamers according to claim 1, characterized in that: The acquisition of multi-source real-time information includes the following steps: Multi-source status information synchronous acquisition: Based on the preset sampling frequency, status information including temperature, humidity, pressure and image is acquired synchronously; Image information validity assessment: For each frame of acquired image information, the image preprocessing module is invoked to assess its validity in order to detect whether image distortion is caused by steam condensation and water mist. This includes the following steps: Image quality feature extraction: The original image is preprocessed to extract quantitative indicators for evaluating image distortion. The quantitative indicators include blur and color features. The blur index is obtained by calculating the variance of the Laplacian operator of the image. The color feature index is characterized by converting the image from the RGB color space to the HSV color space and calculating the average value of all pixels in the H channel. Distortion determination and labeling: The extracted quantization index is compared with the preset validity threshold range: if the Laplacian variance is lower than the preset first threshold, the image is determined to be distorted due to excessive blurring; if the average hue value of the H channel continuously exceeds the preset second threshold range, the image is determined to be distorted due to color drift. If both of the above conditions are met at the same time, it is judged as serious distortion; When an image is determined to be distorted, the image data of that frame is marked, and its distortion type and timestamp are recorded. Image distortion compensation: For image information that is determined to be distorted, a preset compensation strategy is activated. The Kalman filter is called to extract the visual feature sequence based on the previous consecutive frames of images that were determined to be valid in the same partition, predict the visual feature value at the current moment, and use the predicted value as the current valid visual information. Current state vector output: If the image is determined to be valid, visual features representing the cooking state of the ingredients are extracted from the image; if the image is determined to be distorted, the predicted visual features obtained by image distortion compensation are output. The visual features of each partition at the current moment are fused with the synchronously collected temperature, humidity and pressure data to construct a multi-dimensional current state vector.

4. The intelligent steam distribution control method for a food steamer according to claim 1, characterized in that: The demand analysis and control decision-making process includes the following steps: Multi-dimensional control deviation calculation: The current state vector is compared with the target value of the target cooking curve of the corresponding partition at the corresponding time in a dimension-by-dimensional manner to calculate the control deviation vector at the current time. Single-zone basic demand decision: Based on the control deviation vector and combined with the characteristics of the types of ingredients in the zone, the basic heat requirement and target steam quality required for each zone are obtained respectively; Demand Adjustment Based on Thermal Disturbance Model: By establishing a thermal disturbance model, the basic heat demand is adjusted, including: A thermal coupling coefficient matrix between zones is established in advance through experiments, and the actual temperature change rate of each zone is established. Based on the actual temperature change rate of each zone, the basic heat demand is decoupled and corrected, so that the zones achieve the expected temperature rise effect while taking into account the heat contribution of adjacent zones. Global constraint optimization: Based on the preliminary heat demand of each zone after correction, obtain the maximum steam supply power of the steam generating unit at the current moment, and set the minimum steam supply of each zone as a constraint condition; construct an optimization function with the objectives of minimizing total energy consumption and minimizing tracking deviation; use linear programming, quadratic programming or heuristic optimization algorithm to solve the optimization problem and obtain the final heat demand value of each zone; Decision output: Package the final heat demand value and target steam quality parameters to generate an execution instruction vector.

5. The intelligent steam distribution control method for a food steamer according to claim 4, characterized in that: The acquisition of the basic heat requirement includes: A multi-input multi-output PID controller is used, taking temperature deviation, humidity deviation and visual characteristic deviation as inputs, and outputting the basic heat demand of the zone. The acquisition of the target steam quality includes: Based on the identified ingredient type and the current cooking stage, the expected steam quality value is read from the target cooking curve, and the target steam quality is obtained based on the dynamic correction of the rule engine. Among them, dynamic correction based on the rule engine includes: Calculate the deviation between the current temperature and the target temperature, and use a vision module to identify the deviation between the surface color of the food and the target color; The deviation is input into the rule base, and temperature hysteresis compensation, visual feedback adjustment and anti-water accumulation adjustment are performed based on preset rules to correct the expected value of steam quality and obtain the target steam quality.

6. The intelligent steam distribution control method for a food steamer according to claim 1, characterized in that: The dynamic adjustment and allocation are executed. Includes the following steps: Dynamic adjustment of steam quality: An independent micro quality adjustment unit is set on the steam branch corresponding to each zone. The quality adjustment unit on the corresponding branch is controlled according to the target steam quality parameters of each zone, so that the steam entering each zone reaches its respective target quality. Heat demand conversion: Based on the actual steam quality parameters entering the zone, look up the steam thermodynamic properties table to obtain the specific enthalpy of the steam and the specific enthalpy of the saturated water under the corresponding steam quality, and calculate the target mass flow rate required for the zone. Steam flow regulation unit control: The target mass flow rate is sent as the set value to the steam flow regulation unit of the corresponding zone. The steam flow regulation unit adopts a proportional regulating valve and controls the valve opening through analog signals to achieve continuous flow regulation.

7. The intelligent steam distribution control method for a food steamer according to claim 6, characterized in that: The target quality flow rate required for the computational partition is expressed as: ； In the formula, For target quality flow rate, This is the final heat requirement value. It represents the sum of latent heat and sensible heat released by the condensation of a unit mass of steam.

8. The intelligent steam distribution control method for a food steamer according to claim 1, characterized in that: The cooking termination conditions include: Time condition: The preset total time to reach the target cooking curve at the current moment; Temperature conditions: The core temperature of the food reaches the target value and is maintained for a period of time exceeding the set threshold; Visual conditions: The deviation between the current visual features and the target visual features is less than the preset allowable range and remains stable; User intervention: The user issues a termination command through the human-computer interface.

9. A smart steam distribution control system for a food steamer, executing the smart steam distribution control method for a food steamer as described in any one of claims 1-8, characterized in that: It includes a multi-source information acquisition module, an image preprocessing and compensation module, a main controller, a steam quality regulation unit, a steam flow regulation unit, and a human-machine interaction module; The multi-source information acquisition module is set in each independent cooking zone and includes a temperature sensor, a humidity sensor, a pressure sensor and an image acquisition unit, used to collect temperature, humidity, pressure and food image information in the zone in real time; The image preprocessing and compensation module is connected to the image acquisition unit and is used to determine the validity of the acquired image, extract the ambiguity index and color feature index and compare them with a preset threshold to determine whether the image is distorted due to steam and water vapor; for the distorted image, the Kalman filter is called to perform prediction compensation based on the historical valid visual feature sequence, and the valid visual features at the current moment are output. The steam quality adjustment unit is located at the inlet of each zone's steam branch and includes a miniature superheater and a micro-spray nozzle. It is used to independently adjust the dryness or superheat of the corresponding branch steam according to the target steam quality parameters output by the main controller, so that the steam entering each zone reaches the required thermodynamic state. The steam flow regulation unit is installed in each zone's steam branch and uses a proportional regulating valve to calculate the target mass flow rate based on the final heat demand value output by the main controller and the specific enthalpy corresponding to the current steam quality, and to achieve dynamic regulation of the steam flow rate by controlling the valve opening. The human-computer interaction module is connected to the main controller and is used to receive cooking instructions, ingredient information confirmation, and cooking preference settings input by the user, and to display the cooking status of each zone and completion prompts.

10. The intelligent steam distribution control system for a food steamer according to claim 9, characterized in that: The main controller is connected to the multi-source information acquisition module, the image preprocessing and compensation module, the steam quality regulation unit, and the steam flow regulation unit, and is used to execute the following control logic: Target curve matching: Based on the type, quantity, and initial state of ingredients identified in the initial image, match or generate personalized target cooking curves for the target cooking curve database; Deviation calculation and demand decision: The current state vector is compared with the target curve dimension by dimension to calculate the deviation of temperature, humidity and visual features. The basic heat demand is obtained through the PID controller, and the target steam quality is determined by combining the type of food and the rule engine. Thermal interference correction: A thermal interference model is established based on a pre-stored thermal coupling coefficient matrix, and the basic heat demand is decoupled and corrected to obtain the preliminary heat demand; Global optimization: Under the constraint of total steam supply power, the optimization problem is solved with the goal of minimizing energy consumption and tracking deviation, and the final heat demand value of each zone is output. Closed-loop iteration: Repeat the above process with a preset control cycle until the cooking termination condition is met; Adaptive optimization: Record data throughout the entire process and update the thermal coupling coefficient matrix, target curve parameters, and controller coefficients periodically.

Images