Cooking robot menu automatic correction method and system based on artificial intelligence

Abstract

The invention discloses a cooking robot menu automatic correction method and system based on artificial intelligence, and relates to the technical field of intelligent kitchens.The system collects food material data and a food material video set through a sensor set and an industrial high-definition camera, extracts foaming delay data through the food material video set, and sends the foaming delay data to a server; performing dimensionless processing to obtain a food material thermal change data set; calculating a structure thermal deformation index vsi based on the food material thermal change data set to perform food material structure adaptability evaluation; when the food material structure adaptability is evaluated as thermal response texture sensitivity and insensitivity, response behavior data is collected, a thermal seasoning release interaction index htr is calculated, and a flavor consistency index feq is calculated in combination with the thermal seasoning release interaction index htr for flavor conformity evaluation; the system can dynamically adjust the cooking strategy according to the evaluation result, self-adaptive correction of the menu execution process is achieved, and the intelligence and fine control level and the dish output consistency of the cooking robot are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent kitchen technology, and in particular to an artificial intelligence-based cooking robot recipe automatic correction method and system. Background Art

[0002] With the development of smart kitchen technology, cooking robots are gradually entering home and commercial kitchens, replacing traditional manual operations and achieving standardized cooking processes. These devices usually automatically complete steps such as adding ingredients, stir-frying, and heat control through preset programs, improving efficiency and consistency of dishes. However, when faced with complex and changing raw materials, fixed recipe programs are difficult to fully adapt to ingredients from different batches, different textures, or different temperature states, resulting in flavor imbalances in some dishes. To this end, the industry has gradually transformed from "programmed control" to "data-driven automatic recipe correction", further developing dynamic correction technology for flavor output. This is achieved by automatically adjusting and optimizing the flavor dimension by collecting the coupling relationship between the thermal response of ingredients and seasoning behavior in real time.

[0003] In the Chinese invention application with application publication number CN118861689A, an intelligent cooking processing data monitoring method and system based on machine learning are disclosed, which includes: collecting data on flavor parameters during food processing, as well as thermocouple temperature measurement data and infrared thermal imaging data; processing the data to obtain a historical training data set; using the historical training data set to perform model training to obtain a trained comprehensive analysis network model; based on the comprehensive analysis network model, during the current food processing process, predicting the pot body temperature and flavor parameters to obtain a predicted temperature value and a flavor parameter predicted value; calculating a first deviation between the standard temperature of the pot body and the predicted temperature value, and calculating a second deviation between the standard value of the flavor parameter of the pot body and the predicted value of the flavor parameter, and adjusting the current food processing process parameter value according to the first deviation and the second deviation, as well as the weight coefficients of the flavor parameter and the temperature.

[0004] In combination with the existing technology, the above application still has the following deficiencies: First, although the above applications integrate thermocouple temperature measurement, infrared thermal imaging and flavor parameter analysis, and use machine learning to establish a pot temperature and flavor prediction model, the overall application still remains at the general closed-loop control logic of data collection, model prediction and error feedback, lacking analysis and intervention of the dynamic changes of the physical structure during the cooking process; secondly, the above applications do not establish a behavioral and thermal coupling relationship model between seasoning behavior and pot temperature changes, and are unable to determine "whether the current operation constitutes flavor deviation". Control is only based on the deviation of temperature and flavor prediction values. Although the logical closed loop is complete, it lacks deep perception and regulation capabilities, and the overall control strategy is manifested as "feedback compensation". Rather than "feedforward adaptation", it lacks the ability to logically distinguish between the seasoning triggering timing and the thermal state in the pot, making it easy for flavor control to be excessive or delayed; finally, the above-mentioned control method for flavor deviation relies on the deviation between the temperature and the flavor prediction value and is adjusted by weighting the weight coefficient. This control mechanism based on "error feedback" lacks feedforward judgment and dynamic response capabilities, and cannot intervene in the deviation trend in advance or perform personalized strategy classification in actual operation, which leads to delayed control measures, missed seasoning windows, or too single execution path for flavor correction, and lacks iterative optimization and multi-strategy linkage control at the recipe level. Summary of the Invention

[0005] In view of the deficiencies in the prior art, the present invention provides a cooking robot recipe automatic correction method and system based on artificial intelligence, which solves the problems in the above-mentioned background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an artificial intelligence-based cooking robot recipe automatic correction system, including a data acquisition module, an ingredient structure analysis module, a seasoning incentive module, a comprehensive evaluation module and a graded instruction module; The data acquisition module is used to collect food data and food video sets in real time, and perform video processing and feature extraction on the food video sets to obtain foaming delay data, and obtain food thermal change data sets after dimensionless processing; The food structure analysis module is used to calculate the structural thermal deformation index vsi based on the food thermal deformation data group, and to evaluate the food structure adaptability with the preset first structural thermal inertia deformation threshold A and second structural thermal inertia deformation threshold B; The seasoning incentive module is used to collect response behavior data in real time, perform data analysis after dimensionless conversion, obtain a thermal control time series data group, and then calculate the thermal seasoning release interaction index htr based on the thermal control time series data group; The comprehensive evaluation module is used to calculate the flavor consistency index feq and perform flavor consistency evaluation with the preset first flavor consistency threshold M and second flavor consistency threshold N; The grading instruction module is used to execute corresponding recipe correction instructions according to the evaluation results.

[0007] Preferably, the data acquisition module includes a data acquisition unit, an image processing unit and a data preprocessing unit; The data acquisition unit is used to collect food data and food video sets in real time based on the sensor group and industrial high-definition camera installed on the cooking robot; The sensor group includes a laser vision instrument and a near-infrared moisture monitor; The laser vision instrument is used to be installed under the knife holder of the slicing robot arm, and after the food is placed on the cutting tray for slicing, it is used to collect the thickness ds of the food slices in real time when it passes through the feeding channel; The near-infrared moisture monitor is used to be installed at the outlet of the feeding channel to collect the moisture content hl of the food in real time; The image processing unit is used to capture a video of the food in real time using an industrial high-definition camera installed in the center above the pot cavity when the food is put into the pot, extract each frame of the video and mark each frame with a timestamp using the image processing library, perform contrast enhancement using the Retinex algorithm, perform feature extraction on the processed video, and obtain foaming delay data; The feature extraction is used to build a bubble recognition model based on a convolutional neural network, and a large number of bubble images are collected and input into the bubble recognition model for model training. The processed frame images are then input into the bubble recognition model to automatically identify the bubble feature images in the frame images. When the timestamp of the first bubble frame image is identified and the timestamp of the time when the food is put into the pot are subtracted, the bubble delay time tb is obtained; The data preprocessing unit is used to perform dimensionless processing on the food material data and the foaming characteristic data to obtain a food material thermal change data group; The food thermal change data group includes food slice thickness ds, food moisture content hl and foaming delay time tb.

[0008] Preferably, the food structure analysis module includes a thermal deformation analysis unit and an adaptability assessment unit; The thermal deformation analysis unit is used to summarize and calculate the food thermal deformation data group to obtain the structural thermal deformation index vsi, which is used to analyze the degree of synergistic adaptability between the structural characteristics of the food and the thermal reaction of the cookware. The specific formula is as follows: ; Where x represents the integral variable in the thickness direction of the food, dx represents the thickness differential, e represents the exponential function, and f represents the thermal response relaxation time constant, which is used to control the time scale of thermal deformation and can be set by the user according to specific circumstances.

[0009] Preferably, the adaptability evaluation unit is used to preset a first structural thermal hysteresis deformation threshold value A and a second structural thermal hysteresis deformation threshold value B, and to evaluate the adaptability of the food structure based on the structural thermal deformation index vsi obtained in real time. The specific evaluation scheme is as follows; When the structural thermal deformation index vsi is less than or equal to the first structural thermal inertia deformation threshold A, it indicates that the thermal response texture is sensitive. At this time, after executing the first adaptive control instruction, the seasoning analysis instruction is executed; When the first structural thermal inertia deformation threshold value A is less than the structural thermal deformation index vsi and less than the second structural thermal inertia deformation threshold value B, it indicates that the thermal response texture adaptability is normal. At this time, the default cooking program is activated and the current parameters are recorded. When the structural thermal deformation index vsi ≥ the second structural thermal inertia deformation threshold B, it indicates that the thermal response texture is insensitive. At this time, after executing the second adaptive control instruction, the seasoning analysis instruction is executed.

[0010] Preferably, the seasoning incentive module is used to execute the seasoning analysis instruction when the food structure adaptability is assessed as thermal response texture sensitive or insensitive, and specifically includes a response behavior perception unit, a data processing unit and a thermal seasoning analysis unit; The response behavior sensing unit is used to collect and extract response behavior data during the cooking process in real time based on the sensor group installed above the pot and the electronic recipe; The response behavior data includes the initial temperature of the ingredients fs, the initial temperature of the pot bottom st, the time series of the temperature inside the pot W and the time point ts at which the seasoning operation is performed; The sensor group includes a non-contact infrared temperature measuring probe; The non-contact infrared temperature measuring probe is used to be fixed at the end of the feeding channel. When the heating in the pot is started, the initial temperature of the food fs and the initial temperature of the pot bottom st are collected, and the collection timestamp is set to every second. The time series W of the temperature in the pot is collected in real time. The specific form of the time series W of the temperature in the pot is: W={gw0, gw1,,,,gw i}, where gw represents the temperature inside the pot; Obtain the seasoning operation execution time ts based on the time it takes for the seasoning to be first added as recorded in the electronic recipe; The data processing unit is used to perform data analysis on the response behavior data after dimensionless processing to obtain a thermal control time series data group; The data analysis includes temperature rise rate analysis and temperature rise amplitude analysis; The temperature rise rate analysis is used to calculate the pot temperature rise rate T within 5 seconds based on the pot temperature time series W and the pot temperature gw in the previous 5 seconds. 5s , specifically: , where t i Indicates the timestamp of the i-th second, Indicates the average time value of the first 5 seconds, Indicates the average temperature of the previous 5 seconds, gw i represents the temperature inside the pot at the i-th second; The temperature rise analysis is used to record the timestamp of the first bubbling in the pot, and extract the current pot temperature gw according to the pot temperature time series W k , and then subtract the value from the initial temperature st of the pot bottom to obtain the total temperature rise in the pot T, specifically: T=gw k -st; The thermal control time series data group includes the pot temperature rise rate T within 5 seconds 5s , seasoning operation execution time ts, initial temperature of ingredients fs, total temperature rise in the pot T and the initial temperature of the pot bottom st.

[0011] Preferably, the thermal seasoning analysis unit is used to perform summary calculations based on the thermal control time series data group to obtain the thermal seasoning release interaction index htr, which is used to analyze the coupling strength between the temperature change rate of the pot surface, the seasoning carrying capacity of the food structure, and the seasoning behavior time series. The specific formula is as follows: ; Where, log e represents the natural logarithm function, ε represents a stability constant that prevents the denominator from being zero, and its value is 0.01.

[0012] Preferably, the comprehensive evaluation module includes a compliance analysis unit and a consistency evaluation unit; The conformity analysis unit is used to perform a comprehensive calculation based on the structural thermal deformation index vsi and the thermal seasoning release interaction index htr to obtain the flavor consistency index feq, which measures the multi-dimensional deviation in timing, action and rhythm between the actual cooking behavior and the standard recipe model. The specific formula is as follows: ; In the formula, cos represents the cosine function, Represents pi, with two decimal places.

[0013] Preferably, the consistency evaluation unit is used to preset a first flavor matching threshold value M and a second flavor matching threshold value N, and perform flavor consistency evaluation with the flavor consistency index feq obtained in real time. The specific evaluation scheme is as follows; When the flavor consistency index feq is less than the first flavor matching threshold M, it indicates that the flavor of the food has deviated, and the first flavor adjustment instruction is executed; When the first flavor matching threshold M ≤ flavor consistency index feq ≤ flavor path matching threshold N, it indicates that the flavor of the recipe is met, and the seasoning process is performed normally and normal monitoring is maintained; When the flavor consistency index feq>the flavor path matching threshold N, it indicates that the flavor of the food has deviated, and the second flavor adjustment instruction is executed.

[0014] Preferably, the grading instruction module is used to execute corresponding recipe modification instructions according to the evaluation results, as follows; The first adaptive control instruction: the cooking robot lowers the pot temperature by 15%, and uses intermittent heating to control the heat, heating for 3 seconds every 2 seconds, and returns to normal after three rounds; The second adaptive control instruction: increase the pot temperature by 30% and increase the stir-fry interval time by 50%; The first flavor adjustment instruction: the cooking robot delays the seasoning release by 5 seconds and performs iterative analysis and evaluation through the seasoning incentive module until the flavor matches the recipe; The second flavor adjustment instruction: the cooking robot releases the seasoning 5 seconds in advance, and performs iterative analysis and evaluation through the seasoning incentive module until it meets the flavor of the recipe.

[0015] An artificial intelligence-based cooking robot recipe automatic correction method comprises the following steps: S1. Real-time collection of food data and food video sets, and video processing and feature extraction of the food video sets to obtain foaming delay data, and after dimensionless processing, obtain a food thermal change data set; S2. Calculating a structural thermal deformation index vsi based on the food thermal deformation data set, and evaluating the food structural adaptability using a preset first structural thermal hysteresis deformation threshold value A and a preset second structural thermal hysteresis deformation threshold value B; S3. Real-time collection of response behavior data, and data analysis after dimensionless conversion to obtain a thermal control time series data set, and then calculation of the thermal seasoning release interaction index htr based on the thermal control time series data set; S4, calculating the flavor consistency index feq, and performing flavor consistency evaluation with a preset first flavor consistency threshold M and a second flavor consistency threshold N; S5. Execute corresponding recipe modification instructions based on the evaluation results.

[0016] The present invention provides a method and system for automatically correcting cooking robot recipes based on artificial intelligence. It has the following beneficial effects: (1) The data acquisition module of this system uses a laser vision instrument installed below the slicing robot's blade holder and a near-infrared moisture detector at the outlet of the feeding channel to obtain the food slice thickness ds and food moisture content hl in real time. An industrial high-definition camera installed above the pot cavity collects a video of the food being cooked, extracts the food foaming delay time tb, and completes dimensionless processing to form a food thermal change data set. This provides a structural responsiveness and thermal behavior data foundation for subsequent modules, ensuring the automation of the acquisition process and the multidimensionalization of parameters, thus opening up the input channel for the robot to understand the thermal behavior of food.

[0017] (2) The food structure analysis module of the system calculates the structural thermal deformation index vsi through the thermal deformation analysis unit, and the adaptability evaluation unit combines the preset first structural thermal inertia deformation threshold A and second structural thermal inertia deformation threshold B to evaluate the food structure adaptability and determine the food thermal response texture state. When the food structure adaptability is evaluated as thermal response texture sensitive or insensitive, the seasoning stimulation module executes the seasoning analysis instruction. Based on the sensor group installed above the pot and the electronic recipe, the response behavior data during the cooking process is collected and extracted in real time. After dimensionless conversion, the data is analyzed to form a thermal control time series data group. The system further calculates the thermal seasoning release interaction index htr based on the data group, captures the coupling degree between the temperature change speed in the pot, the seasoning timing and the thermal energy accumulation, accurately describes the interactive relationship between the pot thermal response behavior and the seasoning release window, and provides key quantitative indicators for seasoning rhythm optimization.

[0018] (3) The system's comprehensive evaluation module calculates the flavor consistency index feq by integrating the structural thermal deformation index vsi and the thermal seasoning release interaction index htr through the conformity analysis unit and the consistency evaluation unit, and performs flavor conformity evaluation with the preset first flavor conformity threshold M and second flavor conformity threshold N to determine whether the current cooking state meets the flavor specifications. Once a deviation is detected, the system immediately issues corresponding instructions from the hierarchical instruction module. In the case of structural adaptability deviation, the system performs pot temperature adjustment and heating strategy change; in the case of flavor conformity abnormality, the system performs seasoning time adjustment and iterative incentive analysis. Through the series connection of the above modules, the system realizes the whole process control from data acquisition, structural recognition, behavior perception, interactive calculation to adaptive correction. It not only improves the response sensitivity of the cooking process to the state of ingredients and the control accuracy of seasoning behavior, but also enhances the consistency of flavor reproduction and the autonomous correction ability of robot operation, solving the problem that the traditional fixed recipe cooking system cannot adapt to the actual state changes of ingredients. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of an automatic recipe correction system for a cooking robot based on artificial intelligence according to the present invention; Figure 2This is a schematic diagram of the steps of a cooking robot recipe automatic correction method based on artificial intelligence of the present invention; Figure 3 This is a schematic diagram of the operating principle of an artificial intelligence-based automatic recipe correction system for a cooking robot according to the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0021] Example 1

[0022] See also Figure 1 The present invention provides an artificial intelligence-based automatic recipe correction system for a cooking robot. To achieve the above purpose, the present invention is implemented through the following technical solutions: including a data acquisition module, an ingredient structure analysis module, a seasoning incentive module, a comprehensive evaluation module and a graded instruction module; The data acquisition module is used to collect food data and food video sets in real time, and perform video processing and feature extraction on the food video sets to obtain foaming delay data, and obtain food thermal change data sets after dimensionless processing; The food structure analysis module is used to calculate the structural thermal deformation index vsi based on the food thermal deformation data group, and to evaluate the food structure adaptability with the preset first structural thermal inertia deformation threshold A and second structural thermal inertia deformation threshold B; The seasoning incentive module is used to collect response behavior data in real time, perform data analysis after dimensionless conversion, obtain a thermal control time series data group, and then calculate the thermal seasoning release interaction index htr based on the thermal control time series data group; The comprehensive evaluation module is used to calculate the flavor consistency index feq and perform flavor consistency evaluation with the preset first flavor consistency threshold M and second flavor consistency threshold N; The grading instruction module is used to execute corresponding recipe correction instructions according to the evaluation results.

[0023] In this embodiment, the data acquisition module utilizes multiple sensors and video processing to acquire the structural parameters and thermal response behavior of ingredients in real time. It then constructs a data set of thermal changes in the ingredients through dimensionless processing, providing a high-quality input foundation for thermal behavior modeling. The ingredient structure analysis module calculates the data set to obtain the structural thermal deformation index (vsi). This index is then used to evaluate the structural adaptability of the ingredients using a preset first and second structural thermal hysteresis deformation thresholds (A and B). The seasoning stimulation module collects response behavior data during the cooking process when the ingredient structure adaptability is assessed as either thermally sensitive or insensitive. After dimensionless processing, the data is analyzed to obtain a thermal control time series data set. This data is then used to calculate the thermal seasoning release interaction index (htr), accurately characterizing the dynamic coupling relationship between the temperature rise behavior of the cookware and the seasoning operation. The comprehensive evaluation module calculates the flavor consistency index (feq) based on the structural thermal deformation index (vsi) and the thermal seasoning release interaction index (htr). This index is then combined with the preset first and second flavor consistency thresholds (M and N) to assess flavor consistency, reflecting the degree of deviation between the actual cooking behavior and the target recipe. Finally, the hierarchical instruction module implements refined recipe corrections based on the evaluation results, enabling dynamic optimization of wok temperature control and seasoning time adjustment. This system, through a nested feedback mechanism combining structural thermal response, seasoning interaction, and flavor evaluation, empowers the robot to dynamically identify and control ingredient variations and thermal behavior within the wok, significantly enhancing its adaptability in complex environments and its intelligent decision-making capabilities for seasoning. The resulting effect is not only improved flavor reproduction stability, but also significantly enhanced recipe execution accuracy, heat control rationality, and seasoning release rhythm. This overcomes the flavor deviation and imbalance issues that can occur in traditional solutions due to the lack of a structure-behavior linkage evaluation mechanism, driving the evolution of cooking robots toward highly intelligent, adaptive, and authentic interaction.

[0024] Example 2

[0025] Please refer to Figure 1 ,Specifically: the data acquisition module includes a data acquisition unit, an image processing unit and a data preprocessing unit; The data acquisition unit is used to collect food data and food video sets in real time based on the sensor group and industrial high-definition camera installed on the cooking robot; The sensor group includes a laser vision instrument and a near-infrared moisture monitor; The laser vision instrument is used to be installed under the blade holder of the slicing robot arm. After the food is placed on the cutting tray for slicing, it collects the food slice thickness ds in real time when it passes through the feeding channel. It represents the vertical cutting thickness of a single piece of vegetable and reflects the path required for heat penetration. The greater the thickness, the slower the heat diffusion. The near-infrared moisture detector is used to be installed at the outlet of the feeding channel to collect the moisture content of the food in real time, which represents the total percentage of free water and bound water in the food. More moisture means faster thermal conductivity and faster deformation. The image processing unit is used to capture a video of the food in real time using an industrial high-definition camera installed in the center above the pot cavity when the food is put into the pot. The image processing library is used to extract each frame of the video and mark each frame with a timestamp. The image processing unit is then used to perform contrast enhancement using a Retinex algorithm to highlight subtle disturbances in the oil surface texture. Feature extraction is then performed on the processed video to obtain foaming delay data. The feature extraction is used to build a bubble recognition model based on a convolutional neural network, and a large number of bubble images are collected and input into the bubble recognition model for model training. The processed frame images are then input into the bubble recognition model to automatically identify bubble feature images in the frame images. When the timestamp when the first bubble frame image is identified is subtracted from the timestamp when the food is put into the pot, the bubble delay time tb is obtained, which represents the time delay for bubbles to appear on the oil surface after the food is put into the pot. It is used to analyze the critical point of the chemical reaction between the oil temperature and the food; The data preprocessing unit is used to perform dimensionless processing on the food material data and the foaming characteristic data to obtain a food material thermal change data group; The food thermal change data group includes food slice thickness ds, food moisture content hl and foaming delay time tb.

[0026] In this embodiment, by deploying laser vision instruments, near-infrared moisture detectors, and industrial high-definition cameras above the pot cavity at different locations on the cooking robot, food data and food video sets are collected in real time. Image processing algorithms and a convolutional neural network-based bubble recognition model are used to extract oil surface bubble characteristics and obtain bubble delay data. The food data and bubble feature data are uniformly dimensionlessly processed by the data processing unit to construct a food thermal change data set, providing a stable and reliable input basis for subsequent thermal structure analysis and seasoning behavior modeling. This module implementation achieves high-precision, automated, and contactless collection of key factors of structural thermal behavior. This not only breaks the traditional system's reliance on static preset parameters, but also significantly improves the system's ability to perceive the actual thermal response state of the food, thereby providing technical support for dynamic recipe modification and personalized heat control, and overall improving the cooking robot's intelligence level and cooking accuracy.

[0027] Example 3

[0028] Please refer to Figure 1 ,Specifically: the food structure analysis module includes a thermal deformation analysis unit and an adaptability evaluation unit; The thermal deformation analysis unit is used to summarize and calculate the food thermal deformation data group to obtain the structural thermal deformation index vsi, which is used to analyze the degree of synergistic adaptability between the structural characteristics of the food and the thermal reaction of the cookware. The specific formula is as follows: ; Where x represents the integral variable in the thickness direction of the food, dx represents the thickness calculus, e represents the exponential function, and f represents the thermal response relaxation time constant, which is used to control the time scale of thermal deformation and is set by the user according to the specific situation. It represents the total amount of structural thermal inertia of the food layer at unit moisture content. The square of the thermal response demand at each thickness point is weighted to form a nonlinear effect. The thicker the food and the lower the moisture content, the higher the thermal energy required to complete the cooking. It indicates the delay intensity of the temperature rise of the pot surface. The longer the bubbling delay time tb is, the slower the temperature rise rate of the pot surface is, and the pot temperature is insufficient.

[0029] The adaptability evaluation unit is used to preset a first structural thermal inertia deformation threshold value A and a second structural thermal inertia deformation threshold value B, and to evaluate the adaptability of the food structure based on the structural thermal deformation index vsi obtained in real time. The specific evaluation scheme is as follows; When the structural thermal deformation index vsi is less than or equal to the first structural thermal inertia deformation threshold A, it indicates that the thermal response texture is sensitive. At this time, after executing the first adaptive control instruction, the seasoning analysis instruction is executed; When the first structural thermal inertia deformation threshold value A is less than the structural thermal deformation index vsi and less than the second structural thermal inertia deformation threshold value B, it indicates that the thermal response texture adaptability is normal. At this time, the default cooking program is activated and the current parameters are recorded. When the structural thermal deformation index vsi ≥ the second structural thermal inertia deformation threshold B, it indicates that the thermal response texture is insensitive. At this time, after executing the second adaptive control instruction, the seasoning analysis instruction is executed.

[0030] In this embodiment, the food structure analysis module calculates the structural thermal deformation index (vsi) based on the collected food thermal deformation data set using a nonlinear weighted formula in integral form through the thermal deformation analysis unit. This index comprehensively reflects the structural thermal inertia and thermal response requirements of food layers of different thicknesses at a unit moisture content, and considers the impact of the bubbling delay time tb on the temperature rise delay of the pot surface, thereby achieving accurate modeling of the synergistic adaptability between the food structure and the thermal response of the pot. The physical meaning of the formula is to measure the degree to which the structural deformation of the food during the heat treatment process matches the thermal response of the pot, reflecting the synergistic adaptation relationship between the food structure, water content and thermal behavior in the pot. It represents the total amount of thermal inertia per unit moisture content under the thickness distribution of the food. The thicker the food, the stronger the thermal inertia of the food. The moisture content hl is used as the denominator to participate in the square scaling, which means that the lower the moisture content, the more difficult it is for the structure to deform due to heat transfer, and a higher heat drive is required. The multiplication term It represents the hysteresis effect of the temperature rise of the pot surface on the thermal response of the food, that is, the hysteresis contribution of the bubbling delay time tb of the pot surface to the overall thermal behavior. The exponential function is used to simulate the accumulation process of energy transfer from the pot temperature to the surface of the food. For the overall structure, the larger the structural thermal deformation index vsi, the more significant the thermal conduction deformation response and thermal inertia are under the coupling of thickness, moisture and pot temperature.

[0031] Based on the preset first and second structural thermal hysteresis deformation thresholds (A and B), the adaptability assessment unit categorizes the ingredient state into three categories: sensitive to thermal response texture, normal adaptability, and insensitive. This module then determines whether to trigger the adaptive control and seasoning analysis process. This module quantitatively assesses and categorizes the thermal response characteristics of ingredients, effectively improving the cooking robot's accuracy in identifying the structural behavior of different ingredients and program scheduling flexibility. This overcomes the traditional system's weak ability to perceive individual ingredient differences and single heating strategies, enabling refined recipe adaptation and intelligent adjustment of thermal control paths. This significantly enhances the system's ability to control texture changes and maintain flavor stability.

[0032] Example 4

[0033] Please refer to Figure 1 , specifically: the seasoning incentive module is used to execute the seasoning analysis instruction when the food structure adaptability is evaluated as thermal response texture sensitive or insensitive, and specifically includes a response behavior perception unit, a data processing unit and a thermal seasoning analysis unit; The response behavior sensing unit is used to collect and extract response behavior data during the cooking process in real time based on the sensor group installed above the pot and the electronic recipe; The response behavior data includes the initial temperature of the ingredients fs, the initial temperature of the pot bottom st, the time series of the temperature inside the pot W and the time point ts at which the seasoning operation is performed; The sensor group includes a non-contact infrared temperature measuring probe; The non-contact infrared temperature measuring probe is used to be fixed at the end of the feeding channel. When the heating in the pot is started, the initial temperature of the food fs and the initial temperature of the pot bottom st are collected, and the collection timestamp is set to every second. The time series W of the temperature in the pot is collected in real time. The specific form of the time series W of the temperature in the pot is: W={gw0, gw1,,,,gw i}, where gw represents the temperature inside the pot; The initial temperature of the food, fs, represents the temperature of the food before it enters the pot cavity and determines the initial value boundary of the temperature rise control strategy; The initial temperature of the pot bottom st represents the initial temperature before the pot cavity starts to heat; According to the time when the seasoning is first added recorded in the electronic recipe, the seasoning operation execution time point ts is obtained, which represents the time from the start of heating in the pot to the first addition of the seasoning; The data processing unit is used to perform data analysis on the response behavior data after dimensionless processing to obtain a thermal control time series data group; The data analysis includes temperature rise rate analysis and temperature rise amplitude analysis; The temperature rise rate analysis is used to calculate the pot temperature rise rate T within 5 seconds based on the pot temperature time series W and the pot temperature gw in the previous 5 seconds. 5s , which indicates the temperature rise rate of the pot bottom within 5 seconds after heating begins, reflecting the heating efficiency of the pot per unit time, specifically: , where t i Indicates the timestamp of the i-th second, Indicates the average time value of the first 5 seconds, Indicates the average temperature of the previous 5 seconds, gw i represents the temperature inside the pot at the i-th second; The temperature rise analysis is used to record the timestamp of the first bubbling in the pot, and extract the current pot temperature gw according to the pot temperature time series W k , and then subtract the value from the initial temperature st of the pot bottom to obtain the total temperature rise in the pot T represents the net increase in pot bottom temperature from the initial heating to the appearance of bubbles on the oil surface. It reflects the overall thermal response capability of the pot. Specifically: T=gw k -st; The thermal control time series data group includes the pot temperature rise rate T within 5 seconds 5s , seasoning operation execution time ts, initial temperature of ingredients fs, total temperature rise in the pot T and the initial temperature of the pot bottom st.

[0034] The thermal seasoning analysis unit is used to perform summary calculations based on the thermal control time series data set to obtain the thermal seasoning release interaction index htr, which is used to analyze the coupling strength between the temperature change rate of the pot surface, the seasoning carrying capacity of the food structure, and the seasoning behavior time series. The specific formula is as follows: ; Where, log e represents the natural logarithm function, ε represents the stability constant to prevent the denominator from being zero, and its value is 0.01. It represents the interactive coupling between the heating rate of the pot and the seasoning trigger rhythm, and is used to analyze the coupling rate index of thermal induction and seasoning time point. It represents the time lag and the pot heat accumulation term, reflecting the historical accumulation of heat in the pot. The later the seasoning is, the more likely it is that a large amount of heat energy will accumulate on the pot surface. The natural logarithm function ensures that the growth of this term is limited and will not lead to an exponential explosion. Indicates the degree of heat surge in the pot, affects the dissolution and diffusion of seasoning molecules, and is used to control the timing selection accuracy of the seasoning release window.

[0035] In this embodiment, the system activates when the ingredient structure is assessed as either sensitive or insensitive to thermal response. A non-contact infrared temperature probe mounted above the pot captures the initial ingredient temperature fs, the initial pot bottom temperature st, and the pot internal temperature time series W in real time. Combined with the seasoning execution time ts recorded in the electronic recipe, response behavior data reflecting temperature rise and seasoning timing are extracted. A data processing unit non-dimensionalizes and analyzes this response behavior data to obtain a thermal control time series data set. The thermal seasoning analysis unit calculates the thermal seasoning release interaction index htr based on this data set, comprehensively characterizing the coupling strength between the pot surface temperature rise efficiency, seasoning timing, and thermal energy accumulation. Through this module's dynamic identification and indexed analysis, the system achieves precise alignment of seasoning behavior with the pot's thermal response process. This not only optimizes the accuracy of seasoning release window selection, but also enhances the stability of flavor restoration and the system's adaptive adjustment capabilities to complex thermal behaviors. This resolves the misalignment between traditional fixed-time seasoning methods and the actual pot temperature control state, significantly improving temperature control accuracy and seasoning rhythm matching during cooking.

[0036] The physical meaning of the formula is to measure the thermal interaction intensity between the temperature rise behavior of the cookware, the seasoning operation sequence and the heat load characteristics of the food. This term represents the strength of the rhythmic coupling between the pot surface heating rate and the seasoning timing. The faster the pot surface heats up and the earlier the seasoning operation is performed, the larger this term is. This indicates that the thermal surge is more likely to strongly drive the seasoning process, and the rhythmic matching of the seasoning window needs to be strengthened. Indicates the thermal conductivity readiness of the food. A larger initial temperature fs indicates a higher initial temperature, and the food structure is more susceptible to thermal response and deformation. This nonlinear growth is controlled, simulating the slow accumulation relationship between actual thermal conductivity and the softening state of the food. It indicates the influence of the degree of heat energy accumulation on the pot surface on the seasoning behavior. ΔT represents the total temperature rise in the pot from the start of heating to bubbling. st is the initial temperature of the pot bottom. The ratio of the two describes the intensity of net heat accumulation and the thermal inertia of the pot body, controlling the physical window for the dissolution of seasoning substances and the release of flavor. The thermal seasoning release interaction index htr reflects the degree of matching and adaptation between the thermal state in the pot, the seasoning timing and the thermal response ability of the ingredients, providing a basis for controlling the triggering, rhythm and intensity of the seasoning release window.

[0037] Example 5

[0038] Please refer to Figure 1 ,Specifically: the comprehensive evaluation module includes a compliance analysis unit and a consistency ,evaluation unit; The conformity analysis unit is used to perform a comprehensive calculation based on the structural thermal deformation index vsi and the thermal seasoning release interaction index htr to obtain the flavor consistency index feq, which measures the multi-dimensional deviation in timing, action and rhythm between the actual cooking behavior and the standard recipe model. The specific formula is as follows: ; In the formula, cos represents the cosine function, Represents pi, with two decimal places. The joint Euclidean distance of thermal risk and seasoning strategy represents the overall deviation of the structural thermal deformation index vsi and the thermal seasoning release interaction index htr strategy variables. is the strategy matching period function term, which represents the impact of seasoning behavior on the precise matching thermal behavior.

[0039] The consistency evaluation unit is used to preset a first flavor matching threshold value M and a second flavor matching threshold value N, and perform flavor consistency evaluation with the flavor consistency index feq obtained in real time. The specific evaluation scheme is as follows; When the flavor consistency index feq is less than the first flavor matching threshold M, it indicates that the flavor of the food has deviated, and the first flavor adjustment instruction is executed; When the first flavor matching threshold M ≤ flavor consistency index feq ≤ flavor path matching threshold N, it indicates that the flavor of the recipe is met, and the seasoning process is performed normally and normal monitoring is maintained; When the flavor consistency index feq>the flavor path matching threshold N, it indicates that the flavor of the food has deviated, and the second flavor adjustment instruction is executed.

[0040] In this embodiment, a nonlinear strategy deviation function is constructed using cosine functions and π values ​​to accurately quantify the overall deviation between actual cooking behavior and the standard recipe model in terms of timing, movement, and seasoning rhythm. This generates a flavor consistency index (feq). This index then evaluates flavor consistency against a preset first flavor consistency threshold (M) and a second flavor consistency threshold (N), enabling intelligent determination and dynamic regulation of flavor deviations. This module significantly improves the accuracy of identifying deviations in the coupling between seasoning behavior and thermal control status during recipe execution. This enables the system to accurately execute corresponding adjustment instructions upon detecting flavor anomalies, effectively correcting flavor imbalances and ensuring the accuracy and stability of dish flavor reproduction. This results in improved flavor reproduction consistency and enhanced responsiveness of the system's intelligent feedback.

[0041] The physical meaning of the formula is to measure the degree of deviation between the actual cooking behavior and the preset standard recipe model, reflecting the consistency of the behavior of properly adding seasonings under the pot temperature conditions. The core purpose is to dynamically measure the comprehensive offset of the actual thermal behavior and seasoning behavior compared with the standard recipe model during the execution of the cooking robot, and guide the adaptation of the seasoning behavior and the correction of the recipe. represents the Euclidean distance term, which is used to describe the overall deviation between the structural thermal deformation index vsi and the thermal seasoning release interaction index htr during cooking; the periodic cosine harmonic term The cyclical rhythmic relationship between the thermal seasoning release interaction index htr and the structural thermal deformation index vsi is expressed. The introduction of a cosine structure exponentially modulates the synchronization effect of seasoning behavior, reflecting the coordination between seasoning timing and the thermal environment in the pot. The denominator introduces 1+vsi, incorporating the structural thermal deformation index vsi into the cycle adjustment model to achieve dynamic stretching of the seasoning trigger cycle. This enables the system to dynamically adjust the rhythm judgment standard based on the thermal response speed of the food structure. The overall reciprocal is calculated, so that a larger value indicates a more coordinated flavor behavior. The structural thermal deformation index vsi indicates the response difficulty of the structural end and reflects the impedance degree of heat energy transmission. The thermal seasoning release interaction index htr indicates the release coordination of the operating end and reflects whether the seasoning behavior is synchronized with heat accumulation. The two together describe the complete physical logic closed loop from how the heat from the pot enters the ingredients to when the seasoning is released. The distance and rhythm coupling are used to jointly determine the comprehensive deviation degree of the current cooking behavior in the three-dimensional space of the structural layer, timing layer, and action layer. Only when the structure is more adaptable to heat conduction and the seasoning behavior is coordinated with the heat accumulation rhythm can a good flavor restoration effect be achieved. The flavor consistency index feq is used to uniformly evaluate the degree of deviation between the thermal response behavior and the seasoning behavior during the cooking process.

[0042] Example 6

[0043] Please refer to Figure 1 ,Specifically: the grading instruction module is used to execute the corresponding recipe modification instruction according to the evaluation results, as follows; The first adaptive control instruction: the cooking robot lowers the pot temperature by 15%, and uses intermittent heating to control the heat, heating for 3 seconds every 2 seconds, and returns to normal after three rounds; The second adaptive control instruction: increase the pot temperature by 30% and increase the stir-fry interval time by 50%; The first flavor adjustment instruction: the cooking robot delays the seasoning release by 5 seconds and performs iterative analysis and evaluation through the seasoning incentive module until the flavor matches the recipe; The second flavor adjustment instruction: the cooking robot releases the seasoning 5 seconds in advance, and performs iterative analysis and evaluation through the seasoning incentive module until it meets the flavor of the recipe.

[0044] In this embodiment, the hierarchical instruction module executes corresponding recipe correction instructions based on the evaluation results: If the thermal response is texture-sensitive, the first adaptive control instruction lowers the wok temperature by 15% and implements an intermittent heating strategy to prevent excessive heating that results in ingredients being cooked on the outside but raw on the inside. If the texture is insensitive, the second adaptive control instruction increases the wok temperature by 30% and extends the stir-fry interval to enhance heat penetration. If flavor deviation is detected, the first or second flavor adjustment instruction is executed depending on the deviation direction. By adjusting the seasoning release time and combining iterative incentive analysis, the seasoning behavior is closely aligned with the thermal state within the wok. This instruction system implements closed-loop control of the entire process, from structural response to flavor regulation. This not only enhances the cooking robot's recipe execution flexibility and behavioral adaptability, but also effectively improves seasoning accuracy, heat uniformity, and overall flavor reproduction. It significantly improves the flavor inconsistency problem caused by traditional fixed execution logic, and enhances the stability and intelligence level of the intelligent cooking system.

[0045] Example 7

[0046] Please refer to Figure 2 , a cooking robot recipe automatic correction method based on artificial intelligence, comprising the following steps: S1. Real-time collection of food data and food video sets, and video processing and feature extraction of the food video sets to obtain foaming delay data, and after dimensionless processing, obtain a food thermal change data set; S2. Calculating a structural thermal deformation index vsi based on the food thermal deformation data set, and evaluating the food structural adaptability using a preset first structural thermal hysteresis deformation threshold value A and a preset second structural thermal hysteresis deformation threshold value B; S3. Real-time collection of response behavior data, and data analysis after dimensionless conversion to obtain a thermal control time series data set, and then calculation of the thermal seasoning release interaction index htr based on the thermal control time series data set; S4, calculating the flavor consistency index feq, and performing flavor consistency evaluation with a preset first flavor consistency threshold M and a second flavor consistency threshold N; S5. Execute corresponding recipe modification instructions based on the evaluation results.

[0047] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An artificial intelligence-based cooking robot recipe automatic correction system, characterized by: It includes data acquisition module, food structure analysis module, seasoning incentive module, comprehensive evaluation module and graded instruction module; The data acquisition module is used to collect food data and food video sets in real time, and perform video processing and feature extraction on the food video sets to obtain foaming delay data, and obtain food thermal change data sets after dimensionless processing; The food structure analysis module is used to calculate the structural thermal deformation index vsi based on the food thermal deformation data group, and to evaluate the food structure adaptability with the preset first structural thermal inertia deformation threshold A and second structural thermal inertia deformation threshold B; The seasoning incentive module is used to collect response behavior data in real time, perform data analysis after dimensionless conversion, obtain a thermal control time series data group, and then calculate the thermal seasoning release interaction index htr based on the thermal control time series data group; The comprehensive evaluation module is used to calculate the flavor consistency index feq and perform flavor consistency evaluation with the preset first flavor consistency threshold M and second flavor consistency threshold N; The grading instruction module is used to execute corresponding recipe correction instructions according to the evaluation results.

2. The artificial intelligence-based cooking robot recipe automatic correction system according to claim 1, characterized in that: The data acquisition module includes a data acquisition unit, an image processing unit and a data preprocessing unit; The data acquisition unit is used to collect food data and food video sets in real time based on the sensor group and industrial high-definition camera installed on the cooking robot; The sensor group includes a laser vision instrument and a near-infrared moisture monitor; The laser vision instrument is used to be installed under the knife holder of the slicing robot arm, and after the food is placed on the cutting tray for slicing, it is used to collect the thickness ds of the food slices in real time when it passes through the feeding channel; The near-infrared moisture monitor is used to be installed at the outlet of the feeding channel to collect the moisture content hl of the food in real time; The image processing unit is used to capture a video of the food in real time using an industrial high-definition camera installed in the center above the pot cavity when the food is put into the pot, extract each frame of the video and mark each frame with a timestamp using the image processing library, perform contrast enhancement using the Retinex algorithm, perform feature extraction on the processed video, and obtain foaming delay data; The feature extraction is used to build a bubble recognition model based on a convolutional neural network, and a large number of bubble images are collected and input into the bubble recognition model for model training. The processed frame images are then input into the bubble recognition model to automatically identify the bubble feature images in the frame images. When the timestamp of the first bubble frame image is identified and the timestamp of the time when the food is put into the pot are subtracted, the bubble delay time tb is obtained; The data preprocessing unit is used to perform dimensionless processing on the food material data and the foaming characteristic data to obtain a food material thermal change data group; The food thermal change data group includes food slice thickness ds, food moisture content hl and foaming delay time tb.

3. The artificial intelligence-based cooking robot recipe automatic correction system according to claim 2, characterized in that: The food material structure analysis module includes a thermal deformation analysis unit and an adaptability evaluation unit; The thermal deformation analysis unit is used to summarize and calculate the food thermal deformation data group to obtain the structural thermal deformation index vsi, which is used to analyze the degree of synergistic adaptability between the structural characteristics of the food and the thermal reaction of the cookware. The specific formula is as follows: ； Where x represents the integral variable in the thickness direction of the food, dx represents the thickness differential, e represents the exponential function, and f represents the thermal response relaxation time constant, which is used to control the time scale of thermal deformation and can be set by the user according to specific circumstances.

4. The artificial intelligence-based cooking robot recipe automatic correction system according to claim 3, characterized in that: The adaptability evaluation unit is used to preset a first structural thermal inertia deformation threshold value A and a second structural thermal inertia deformation threshold value B, and to evaluate the adaptability of the food structure based on the structural thermal deformation index vsi obtained in real time. The specific evaluation scheme is as follows; When the structural thermal deformation index vsi is less than or equal to the first structural thermal inertia deformation threshold A, it indicates that the thermal response texture is sensitive. At this time, after executing the first adaptive control instruction, the seasoning analysis instruction is executed; When the first structural thermal inertia deformation threshold value A is less than the structural thermal deformation index vsi and less than the second structural thermal inertia deformation threshold value B, it indicates that the thermal response texture adaptability is normal. At this time, the default cooking program is activated and the current parameters are recorded. When the structural thermal deformation index vsi ≥ the second structural thermal inertia deformation threshold B, it indicates that the thermal response texture is insensitive. At this time, after executing the second adaptive control instruction, the seasoning analysis instruction is executed.

5. The artificial intelligence-based cooking robot recipe automatic correction system according to claim 4, characterized in that: The seasoning incentive module is used to execute the seasoning analysis instruction when the food structure adaptability is assessed as thermal response texture sensitive or insensitive, and specifically includes a response behavior perception unit, a data processing unit and a thermal seasoning analysis unit; The response behavior sensing unit is used to collect and extract response behavior data during the cooking process in real time based on the sensor group installed above the pot and the electronic recipe; The response behavior data includes the initial temperature of the ingredients fs, the initial temperature of the pot bottom st, the time series of the temperature inside the pot W and the time point ts at which the seasoning operation is performed; The sensor group includes a non-contact infrared temperature measuring probe; The non-contact infrared temperature measuring probe is used to be fixed at the end of the feeding channel. When the heating in the pot is started, the initial temperature of the food fs and the initial temperature of the pot bottom st are collected, and the collection timestamp is set to every second. The time series W of the temperature in the pot is collected in real time. The specific form of the time series W of the temperature in the pot is: W={gw0, gw1,,,,gw i }, where gw represents the temperature inside the pot; Obtain the seasoning operation execution time ts based on the time it takes for the seasoning to be first added as recorded in the electronic recipe; The data processing unit is used to perform data analysis on the response behavior data after dimensionless processing to obtain a thermal control time series data group; The data analysis includes temperature rise rate analysis and temperature rise amplitude analysis; The temperature rise rate analysis is used to calculate the pot temperature rise rate T within 5 seconds based on the pot temperature time series W and the pot temperature gw in the previous 5 seconds. 5s , specifically: , where t i Indicates the timestamp of the i-th second, Indicates the average time value of the first 5 seconds, Indicates the average temperature of the previous 5 seconds, gw i represents the temperature inside the pot at the i-th second; The temperature rise analysis is used to record the timestamp of the first bubbling in the pot, and extract the current pot temperature gw according to the pot temperature time series W k , and then subtract the value from the initial temperature st of the pot bottom to obtain the total temperature rise in the pot T, specifically: T=gw k -st; The thermal control time series data group includes the pot temperature rise rate T within 5 seconds 5s , seasoning operation execution time ts, initial temperature of ingredients fs, total temperature rise in the pot T and the initial temperature of the pot bottom st.

6. The artificial intelligence-based cooking robot recipe automatic correction system according to claim 5, characterized in that: The thermal seasoning analysis unit is used to perform summary calculations based on the thermal control time series data set to obtain the thermal seasoning release interaction index htr, which is used to analyze the coupling strength between the temperature change rate of the pot surface, the seasoning carrying capacity of the food structure, and the seasoning behavior time series. The specific formula is as follows: ； Where, log e represents the natural logarithm function, ε represents a stability constant that prevents the denominator from being zero, and its value is 0.

01.

7. The artificial intelligence-based cooking robot recipe automatic correction system according to claim 6, characterized in that: The comprehensive assessment module includes a compliance analysis unit and a consistency assessment unit; The conformity analysis unit is used to perform a comprehensive calculation based on the structural thermal deformation index vsi and the thermal seasoning release interaction index htr to obtain the flavor consistency index feq, which measures the multi-dimensional deviation in timing, action and rhythm between the actual cooking behavior and the standard recipe model. The specific formula is as follows: ； In the formula, cos represents the cosine function, Represents pi, with two decimal places.

8. The artificial intelligence-based automatic recipe correction system for a cooking robot according to claim 7, characterized in that: The consistency evaluation unit is used to preset a first flavor matching threshold value M and a second flavor matching threshold value N, and perform flavor consistency evaluation with the flavor consistency index feq obtained in real time. The specific evaluation scheme is as follows; When the flavor consistency index feq is less than the first flavor matching threshold M, it indicates that the flavor of the food has deviated, and the first flavor adjustment instruction is executed; When the first flavor matching threshold M ≤ flavor consistency index feq ≤ flavor path matching threshold N, it indicates that the flavor of the recipe is met, and the seasoning process is performed normally and normal monitoring is maintained; When the flavor consistency index feq>the flavor path matching threshold N, it indicates that the flavor of the food has deviated, and the second flavor adjustment instruction is executed.

9. The artificial intelligence-based cooking robot recipe automatic correction system according to claim 8, characterized in that: The grading instruction module is used to execute corresponding recipe modification instructions according to the evaluation results, as follows; The first adaptive control instruction: the cooking robot lowers the pot temperature by 15%, and uses intermittent heating to control the heat, heating for 3 seconds every 2 seconds, and returns to normal after three rounds; The second adaptive control instruction: increase the pot temperature by 30% and increase the stir-fry interval time by 50%; The first flavor adjustment instruction: the cooking robot delays the seasoning release by 5 seconds and performs iterative analysis and evaluation through the seasoning incentive module until the flavor matches the recipe; The second flavor adjustment instruction: the cooking robot releases the seasoning 5 seconds in advance, and performs iterative analysis and evaluation through the seasoning incentive module until it meets the flavor of the recipe.

10. An artificial intelligence-based cooking robot recipe automatic correction method, applied to the artificial intelligence-based cooking robot recipe automatic correction system according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1. Real-time collection of food data and food video sets, and video processing and feature extraction of the food video sets to obtain foaming delay data, and after dimensionless processing, obtain a food thermal change data set; S2. Calculating a structural thermal deformation index vsi based on the food thermal deformation data set, and evaluating the food structural adaptability using a preset first structural thermal hysteresis deformation threshold A and a preset second structural thermal hysteresis deformation threshold B; S3. Real-time collection of response behavior data, and data analysis after dimensionless conversion to obtain a thermal control time series data set, and then calculation of the thermal seasoning release interaction index htr based on the thermal control time series data set; S4, calculating the flavor consistency index feq, and performing flavor consistency evaluation with a preset first flavor consistency threshold M and a second flavor consistency threshold N; S5. Execute corresponding recipe modification instructions based on the evaluation results.

Images