Deep learning driven personalized cooking curve generation method and system

Abstract

The application relates to the cooking control field and discloses a personalized cooking curve generation method and system based on deep learning driving, which comprises the following steps: after a user obtains a first baking product by baking a first food material according to a first cooking curve through an oven, first multi-dimensional feedback information of the user's feedback on the first baking product is acquired, the first multi-dimensional feedback information comprises evaluation feedback information of multiple dimensions; through a cooking curve segmentation model, the first cooking curve is segmented into multiple cooking curve segments according to the evaluation feedback information of the first dimension determined by the user; through a first cooking curve optimization model corresponding to the evaluation feedback information of the first dimension, the multiple cooking curve segments are optimized according to the evaluation feedback information of other dimensions except the first dimension, and an optimized cooking curve corresponding to the first cooking curve is obtained. The oven can optimize the heating curve based on deep learning driving, and the cooking effect of the oven is improved.

Description

Technical Field

[0001] This application relates to the field of cooking control technology, and more specifically, to a method and system for generating personalized cooking curves based on deep learning. Background Technology

[0002] In the baking and roasting industry, the precision of an oven's heating control directly affects the taste, texture, and appearance of food. Because the physical structure, moisture content, and main components (such as protein, carbohydrates, and fat) of ingredients like bread, cakes, and meats differ significantly, the biochemical reactions that occur during heating (such as Maillard reaction, starch gelatinization, and protein denaturation) vary in their sensitivity to temperature and time. For example, bread dough undergoes crust hardening, internal starch maturation, and moisture migration during baking; cake batter involves protein structure stabilization, sugar caramelization, and maintaining a light and airy texture; while roasting meat requires balancing surface browning, internal juiciness, and even cooking. These complex culinary demands necessitate ovens with precise heating profile control capabilities to match the heat transfer patterns of different ingredients at different stages.

[0003] When controlling the baking process, taking bread baking as an example, the initial high temperature (approximately 200-230℃) allows the dough surface to quickly dehydrate and form a hard crust. This process, through the Maillard reaction, gives the bread a golden-brown color and a caramelized flavor, while also locking in internal moisture to prevent excessive evaporation and a dry, hard texture. Once the crust has set, the temperature is lowered to 160-180℃ for slow baking, allowing heat to penetrate evenly into the dough. This ensures that the starch is fully gelatinized and the gluten network is fully formed, while preventing the crust from burning due to prolonged high temperatures. Similarly, when baking chiffon cakes, the batter should first be expanded evenly at a medium temperature (150-170℃) before gradually increasing the temperature to fix the structure and prevent collapse. When baking steaks, a high temperature (above 220℃) is used to quickly seal the edges and lock in the juices, followed by slow baking at a low temperature (120-150℃) to the desired doneness. The phased and differentiated heating strategy in the baking process precisely designs the heating curve to meet the characteristic needs of ingredients at different cooking stages, thereby achieving a technological upgrade from "cooking" to "high-quality cooking." However, ovens cannot optimize heating curves for individual users, making it difficult to meet their personalized needs. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for generating personalized cooking curves based on deep learning, which solves the technical problem that ovens cannot perform personalized optimization of heating curves, and achieves the technical effect that ovens can perform personalized optimization of heating curves based on deep learning.

[0005] This application provides a method for generating personalized cooking curves based on deep learning. The method includes: after a user bakes a first ingredient in an oven according to a first cooking curve to obtain a first baked product, the user obtains first multi-dimensional feedback information on the first baked product. The first multi-dimensional feedback information includes evaluation feedback information in multiple dimensions, including evaluation feedback information on doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, charring degree, and cooking efficiency. Using a cooking curve segmentation model, the first cooking curve is segmented into multiple cooking curve segments based on the first dimension of evaluation feedback information determined by the user. Using a first cooking curve optimization model corresponding to the first dimension of evaluation feedback information, the multiple cooking curve segments are optimized based on evaluation feedback information in other dimensions besides the first dimension of evaluation feedback information to obtain an optimized cooking curve corresponding to the first cooking curve.

[0006] In one possible implementation, the method further includes: using a feedback contradiction analysis unit, determining multiple contradictory feedback information groups within the first multi-dimensional feedback information based on the first multi-dimensional feedback information, each contradictory feedback information group including feedback information from two mutually contradictory dimensions; wherein, the multiple contradictory feedback information groups include ripeness and crispness, ripeness and moisture retention, color and burntness, aroma and nutrient retention, texture uniformity and cooking efficiency, crispness and cooking efficiency, and nutrient retention and cooking efficiency; determining the similarity between each contradictory feedback information group and the evaluation feedback information of the first dimension, as a priority index corresponding to each contradictory feedback information group; and prioritizing... The system presents multiple sets of contradictory feedback information to users in descending order of the evaluation index. It also obtains the weights of the contradictory dimensions of user feedback for each set, where the weights represent the importance of the contradictory dimensions. Using the first cooking curve optimization model corresponding to the first dimension's evaluation feedback information, and based on each contradictory feedback information set, the weights of the contradictory dimensions within each set, and evaluation feedback information from other dimensions, the system optimizes multiple cooking curve segments to obtain the optimized cooking curve corresponding to the first cooking curve.

[0007] In another possible implementation, the method further includes: obtaining the oven's operating specifications information and obtaining the relative weight values ​​among multiple contradictory feedback information groups corresponding to the operating specifications information, wherein the relative weight values ​​characterize the mutual influence relationship among the multiple contradictory feedback information groups; and optimizing multiple cooking curve segments through a first cooking curve optimization model corresponding to the first multidimensional feedback information, based on the relative weight values ​​among the multiple contradictory feedback information groups, the weights of feedback information of each contradictory feedback information group and the weights of feedback information of the mutually contradictory dimensions of each contradictory feedback information group, and the evaluation feedback information of other dimensions outside the multiple contradictory feedback information groups, to obtain the optimized cooking curve corresponding to the first cooking curve.

[0008] In another possible implementation, the method further includes: identifying surface contradictory feedback information groups among multiple contradictory feedback information groups using a cooking contradiction identification unit, wherein the feedback information of the two mutually contradictory dimensions of the surface contradictory feedback information group can be eliminated through segmented optimization of the cooking curve; determining the control optimization strategy corresponding to each contradictory feedback information group based on the two mutually contradictory dimensions of the feedback information of each contradictory feedback information group using a knowledge graph-based cooking curve optimization unit; optimizing multiple cooking curve segments according to the control optimization strategy corresponding to each contradictory feedback information group to obtain multiple intermediate cooking curve segments; and optimizing multiple intermediate cooking curve segments according to the relative weight values ​​between multiple contradictory feedback information groups, the weight of each contradictory feedback information group, the weight of the feedback information of the mutually contradictory dimensions of each contradictory feedback information group, and the evaluation feedback information of other dimensions outside the multiple contradictory feedback information groups, to obtain the optimized cooking curve corresponding to the first cooking curve using a first cooking curve optimization model corresponding to the first cooking curve.

[0009] In another possible implementation, the method further includes: after the user bakes the second ingredient in an oven according to the first cooking curve to obtain the second baked product, obtaining second multi-dimensional feedback information from the user on the second baked product. This second multi-dimensional feedback information includes evaluation feedback information from multiple dimensions, including evaluation feedback information on doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, degree of charring, and cooking efficiency. The method also involves obtaining the second ingredient characteristics corresponding to the second ingredient and the first ingredient characteristics corresponding to the first ingredient, and determining the second multi-dimensional feedback information based on these characteristics. The evaluation feedback information from multiple dimensions and the similarity weights corresponding to the evaluation feedback information from multiple dimensions in the first multidimensional feedback information are used. Using a cooking curve segmentation model, the first cooking curve is segmented into multiple cooking curve segments based on the evaluation feedback information from the first dimension. Using a first cooking curve optimization model corresponding to the first multidimensional feedback information, the multiple cooking curve segments are optimized based on the evaluation feedback information from other dimensions besides the first dimension, the evaluation feedback information from multiple dimensions in the second multidimensional feedback information, and the similarity weights corresponding to the evaluation feedback information from multiple dimensions in the second multidimensional feedback information, resulting in an optimized cooking curve corresponding to the first cooking curve.

[0010] In another possible implementation, the method further includes: obtaining the first number of times the user bakes the first ingredient in an oven according to the first cooking curve, and obtaining the second number of times the user bakes the second ingredient in an oven according to the first cooking curve; determining the quotient of the second number of cookings divided sequentially by the first number of cookings and the standard number of cookings, as a cooking number adjustment factor; multiplying the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information by the cooking number adjustment factor, so as to adjust the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information.

[0011] In another possible implementation, the method further includes: obtaining multiple cooking health concerns input by the user; determining the cooking health adjustment factors corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information for each of the multiple cooking health concerns; and multiplying the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information by the cooking health adjustment factors to adjust the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information.

[0012] In another possible implementation, the method further includes: obtaining multiple first key dimension feedback information from the first multidimensional feedback information and multiple second key dimension feedback information from the second multidimensional feedback information, wherein the first key dimension feedback information is the evaluation feedback information of the dimension in which the first ingredient plays a key role in baking, and the second key dimension feedback information is the evaluation feedback information of the dimension in which the second ingredient plays a key role in baking; taking the intersection of the multiple first key dimension feedback information and the multiple second key dimension feedback information to obtain multiple overlapping key dimension feedback information; and optimizing multiple cooking curve segments using the first cooking curve optimization model corresponding to the first multidimensional feedback information, based on the evaluation feedback information of other dimensions besides the first dimension evaluation feedback information, the multiple overlapping key dimension feedback information, and the similarity weights corresponding to the multiple overlapping key dimension feedback information, to obtain the optimized cooking curve corresponding to the first cooking curve.

[0013] In another possible implementation, obtaining multiple first key dimension feedback information from the first multidimensional feedback information and multiple second key dimension feedback information from the second multidimensional feedback information includes: determining multiple first ingredient key dimension feedback information corresponding to the first ingredient in the first multidimensional feedback information based on the ingredient characteristics of the first ingredient; obtaining multiple first historical multidimensional feedback information from user feedback on the first baked product in historical cooking data; determining multiple second ingredient key dimension feedback information corresponding to the second ingredient in the second multidimensional feedback information based on the ingredient characteristics of the second ingredient; obtaining multiple second historical multidimensional feedback information from user feedback on the second baked product in historical cooking data; taking the intersection of the multiple first ingredient key dimension feedback information and the multiple first historical multidimensional feedback information to obtain multiple first key dimension feedback information from the first multidimensional feedback information; taking the intersection of the multiple second ingredient key dimension feedback information and the multiple second historical multidimensional feedback information to obtain multiple second key dimension feedback information from the second multidimensional feedback information.

[0014] This application also provides a deep learning-driven personalized cooking curve generation system, including a unit for performing the method described in any of the preceding claims.

[0015] The beneficial effects of the embodiments in this application compared with the prior art are:

[0016] This application provides a method for generating personalized cooking curves based on deep learning. The method includes: after a user bakes a first ingredient in an oven according to a first cooking curve to obtain a first baked product, obtaining first multi-dimensional feedback information from the user on the first baked product. This first multi-dimensional feedback information includes evaluation feedback information from multiple dimensions, including cookedness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, degree of charring, and cooking efficiency. Using a cooking curve segmentation model, the first cooking curve is segmented into multiple cooking curve segments based on the first-dimensional evaluation feedback information determined by the user. Using a first cooking curve optimization model corresponding to the first-dimensional evaluation feedback information, the multiple cooking curve segments are optimized based on evaluation feedback information from other dimensions besides the first-dimensional evaluation feedback information to obtain an optimized cooking curve corresponding to the first cooking curve. This application embodiment can personalize the cooking curve based on the multi-dimensional evaluation feedback information provided by the user, improving the personalization of the oven's operation and enabling the food baked in the oven to meet the user's personalized taste requirements. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the first method for generating personalized cooking curves based on deep learning, provided in this application embodiment;

[0019] Figure 2 A schematic diagram of the workflow of the first deep learning-driven personalized cooking curve generation method provided in the embodiments of this application;

[0020] Figure 3 A schematic diagram showing the comparison of cooking curves before and after optimization using the deep learning-driven personalized cooking curve generation method provided in the embodiments of this application;

[0021] Figure 4 A flowchart illustrating the second method for generating personalized cooking curves based on deep learning, provided in an embodiment of this application.

[0022] Figure 5 A schematic diagram of the workflow of the second deep learning-driven personalized cooking curve generation method provided in the embodiments of this application;

[0023] Figure 6A flowchart illustrating the third method for generating personalized cooking curves based on deep learning, provided in this application embodiment;

[0024] Figure 7 A flowchart illustrating the fourth method for generating personalized cooking curves based on deep learning, provided in this application embodiment;

[0025] Figure 8 This is a schematic diagram of the logical structure of a personalized cooking curve generation system based on deep learning, provided in an embodiment of this application. Detailed Implementation

[0026] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0027] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0028] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0029] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0030] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0031] Existing ovens cannot optimize heating curves for individual users, making it difficult to meet their personalized needs.

[0032] Based on the above reasons, this application provides a method for generating personalized cooking curves based on deep learning. The method includes: after a user bakes a first ingredient in an oven according to a first cooking curve to obtain a first baked product, obtaining first multi-dimensional feedback information from the user on the first baked product. This first multi-dimensional feedback information includes evaluation feedback information from multiple dimensions, including evaluation feedback information on doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, degree of charring, and cooking efficiency. Using a cooking curve segmentation model, the first cooking curve is segmented into multiple cooking curve segments based on the first-dimensional evaluation feedback information determined by the user. Using a first cooking curve optimization model corresponding to the first-dimensional evaluation feedback information, the multiple cooking curve segments are optimized based on evaluation feedback information from other dimensions besides the first-dimensional evaluation feedback information to obtain an optimized cooking curve corresponding to the first cooking curve. This application embodiment can personalize the cooking curve based on the multi-dimensional evaluation feedback information provided by the user, improving the personalization of the oven's operation and enabling the food baked in the oven to meet the user's personalized taste requirements.

[0033] In some scenarios, the deep learning-driven personalized cooking curve generation method of this application embodiment can be applied to the personalized control of ovens, which can optimize the cooking curve of the oven in a personalized way and improve the oven's performance.

[0034] The following describes in detail, with specific examples, a method for generating personalized cooking curves based on deep learning provided in the embodiments of this application.

[0035] Figure 1 A flowchart illustrating the first deep learning-driven personalized cooking curve generation method provided in this application embodiment is shown below. Figure 1 As shown, the above method includes S110 to S120, and S110 to S120 will be described in detail below.

[0036] S110. After the user bakes the first ingredient in the oven according to the first cooking curve to obtain the first baked product, the user obtains the first multi-dimensional feedback information on the first baked product. The first multi-dimensional feedback information includes evaluation feedback information in multiple dimensions, including evaluation feedback information on doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, charring degree, and cooking efficiency.

[0037] In this implementation, in order to optimize the cooking curve of the oven, after the user bakes the first ingredient in the oven according to the first cooking curve to obtain the first baked product, the user's first multi-dimensional feedback information on the first baked product can be obtained. The first multi-dimensional feedback information includes evaluation feedback information of multiple dimensions. When obtaining the multi-dimensional evaluation feedback information, the user can submit multi-dimensional evaluation feedback information on the first baked product through the interactive interface.

[0038] For example, multi-dimensional evaluation feedback can be provided through the oven's built-in touch interface or a networked mobile application, allowing users to easily provide detailed evaluations of the different sensory characteristics and process indicators of the first baked product.

[0039] For example, the evaluation feedback information of multiple dimensions may include evaluation feedback information of doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, charring degree and cooking efficiency. The evaluation feedback information may specifically include the score value corresponding to each indicator, and the score value corresponding to each indicator is represented by numbers.

[0040] For example, when obtaining evaluation feedback information from multiple dimensions, evaluation criteria such as doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, charring degree, and cooking efficiency can be published to users. Then, the evaluation feedback information from users from multiple dimensions can be obtained to minimize the impact of subjective factors on the feedback information of doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, charring degree, and cooking efficiency.

[0041] S120. Using the cooking curve segmentation model, based on the user-defined first-dimensional evaluation feedback information, the first cooking curve is segmented into multiple cooking curve segments. Using the first cooking curve optimization model corresponding to the first-dimensional evaluation feedback information, based on evaluation feedback information from other dimensions besides the first-dimensional evaluation feedback information, the multiple cooking curve segments are optimized to obtain the optimized cooking curve corresponding to the first cooking curve.

[0042] Figure 2 A schematic diagram of the workflow of the first deep learning-driven personalized cooking curve generation method provided in this application embodiment is shown below. Figure 2 As shown, when optimizing the cooking curve, a deep learning-driven cooking curve segmentation model can be used to divide the first cooking curve into stages based on the evaluation feedback information of the first dimension actively selected by the user, resulting in multiple cooking curve segments obtained by segmenting the first cooking curve.

[0043] For example, when users use ripeness rating as a key dimension, the model can analyze the feedback data corresponding to the ripeness dimension, identify the key time points that affect the ripeness change curve during the cooking process, and thus divide the continuous cooking process into multiple cooking curve segments with different heating rate characteristics. These multiple cooking curve segments can specifically include the pretreatment stage, the core heating stage, and the post-ripening stage. This dynamic segmentation method can determine multiple cooking curve segments that are directly related to user preferences for different optimization objectives.

[0044] like Figure 2 As shown, after obtaining multiple cooking curve segments from the first cooking curve, the first cooking curve optimization model corresponding to the first dimension evaluation feedback information can be used to optimize the multiple cooking curve segments based on evaluation feedback information from other dimensions besides the first dimension evaluation feedback information, to obtain the optimized cooking curve corresponding to the first cooking curve. The optimized cooking curve is a cooking curve optimized according to user preferences, and the cooking process of the first ingredient can be optimized based on the optimized cooking curve.

[0045] Figure 3 This illustration shows a comparison of the cooking curve before and after optimization using the deep learning-driven personalized cooking curve generation method provided in the embodiments of this application. Figure 3 The image above shows the first cooking curve. Figure 3 The evaluation feedback information corresponding to the first dimension in the above figure can be maturity; Figure 3 The image below shows the optimized cooking curve corresponding to the first cooking curve. Figure 3 The image below corresponds to the optimized cooking curve, which is the first cooking curve optimized based on factors such as crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, degree of charring, and cooking efficiency. Figure 3 The optimized cooking curve in the figure below improves the heating power in the initial heating stage and extends the heating time in the maximum power stage. The optimized cooking curve comprehensively improves the cooking effect corresponding to the cooked rice's doneness and other evaluation feedback information.

[0046] In this implementation, after the user determines different evaluation feedback information of the first dimension, the cooking curve is optimized according to the first cooking curve optimization model corresponding to the evaluation feedback information of the first dimension. This allows for targeted optimization of the cooking curve based on evaluation feedback information of other dimensions besides the first dimension, thereby improving the personalization of the optimization process.

[0047] For example, when the selected first dimension of evaluation feedback information is the doneness dimension, the heating parameters of each cooking curve segment can be optimized based on evaluation feedback information from other dimensions besides the first dimension, such as crispness, color, and nutrient retention.

[0048] For example, the cooking curve segmentation model can be a deep learning model based on neural networks. The cooking curve segmentation model can be trained using the evaluation feedback information of the first dimension of the samples and multiple sample cooking curve segments. The trained cooking curve segmentation model can divide the first cooking curve into stages according to the evaluation feedback information of the first dimension actively selected by the user, and obtain multiple cooking curve segments obtained by segmenting the first cooking curve.

[0049] When optimizing the heating parameters for each cooking curve segment, the first cooking curve optimization model based on a deep neural network can be used to perform multi-dimensional optimization of parameters such as temperature gradient, heating duration, and humidity control threshold to find a combination of cooking parameters that meets multi-dimensional requirements.

[0050] For example, when users are particularly concerned about nutrient retention, the system can automatically adjust the duration of the high-temperature segment while ensuring that the degree of gelatinization meets the standard, thereby improving the cooking effect.

[0051] For example, the first cooking curve optimization model can be trained by a neural network-based deep learning model, and the first cooking curve optimization model can be trained by sample evaluation feedback information and sample optimization cooking curves in dimensions other than the first dimension evaluation feedback information.

[0052] It should be noted that when optimizing the heating parameters of each cooking curve segment, the optimized cooking curve segments must meet the parameter smooth transition conditions between adjacent segments when re-splicing, ensuring the feasibility of the optimized cooking curve.

[0053] The beneficial effect of the above implementation method is that, through the full-dimensional collection of multi-dimensional feedback information and the combined application of different dimensions, it is possible to accurately analyze the user's focus and sensitivity to various cooking effect indicators, thereby generating an optimized cooking plan that not only meets individual taste preferences but also takes into account the overall quality of the dish.

[0054] The beneficial effect of the above implementation method is that, through the dynamic combination of the cooking curve segmentation model and the first cooking curve optimization model, it is possible to strengthen and adjust local parameters for a single dimension that users focus on, and to achieve global balance optimization based on feedback data from other dimensions, ultimately achieving a systematic improvement in multi-dimensional indicators of cooking effect.

[0055] Figure 4 A flowchart illustrating the second deep learning-driven personalized cooking curve generation method provided in this application embodiment is shown below. Figure 4 As shown, the above method also includes S210 to S230, which will be explained in detail below.

[0056] S210. Through the feedback contradiction analysis unit, based on the first multi-dimensional feedback information, determine multiple contradictory feedback information groups in the first multi-dimensional feedback information. Each contradictory feedback information group includes feedback information from two mutually contradictory dimensions. Among them, the multiple contradictory feedback information groups include: doneness and crispness, doneness and moisture retention, color and charring, aroma and nutrient retention, texture uniformity and cooking efficiency, crispness and cooking efficiency, and nutrient retention and cooking efficiency.

[0057] When users provide feedback on the baking results, there may be contradictory feedback information groups. When optimizing the cooking curve, the decision on the cooking curve may be affected by the contradictory feedback information. Therefore, after users submit their evaluation of the first baked product through multi-dimensional feedback information, the contradiction analysis unit can be used to identify the contradiction relationship of the multi-dimensional feedback information, and optimize the cooking curve for the multi-dimensional feedback information with contradictory relationships.

[0058] When identifying contradictory relationships in multidimensional feedback information, a feedback contradiction analysis unit can be used to determine multiple contradictory feedback information groups in the first multidimensional feedback information based on the first multidimensional feedback information. Each contradictory feedback information group includes feedback information from two mutually contradictory dimensions. The feedback contradiction analysis unit can be constructed based on a preset dimension association rule base and user historical preference data. The feedback contradiction analysis unit can automatically detect multiple contradictory feedback information groups existing in the first multidimensional feedback information.

[0059] For example, each contradictory feedback information group consists of two evaluation dimensions with a negative correlation. For instance, the evaluations of ripeness and crispness may have an inverse influence trend. The system identifies such contradictory relationships by analyzing the synergistic change patterns of the score values ​​of different dimensions.

[0060] For example, multiple contradictory feedback information groups may include contradictory feedback information groups such as 1. doneness and crispness; 2. doneness and moisture retention; 3. color and charring; 4. aroma and nutrient retention; 5. texture uniformity and cooking efficiency; 6. crispness and cooking efficiency; 7. nutrient retention and cooking efficiency.

[0061] S220. Determine the similarity between each contradictory feedback information group and the evaluation feedback information of the first dimension, using this as the priority index corresponding to each contradictory feedback information group. Present multiple contradictory feedback information groups to the user in descending order of priority index, and obtain the weights of the user's feedback information on the contradictory dimensions for each contradictory feedback information group. The weights of the feedback information on the contradictory dimensions represent the importance of the user's feedback information on the contradictory dimensions.

[0062] In this implementation, to ensure the personalized optimization effect of the cooking curve, the similarity between each contradictory feedback information group and the evaluation feedback information of the first dimension can be determined as the priority index corresponding to each contradictory feedback information group. The similarity between each contradictory feedback information group and the evaluation feedback information of the first dimension represents the relevance between each contradictory feedback information group and the evaluation feedback information of the first dimension determined by the user. In this way, multiple contradictory feedback information groups can be fed back in order of priority index from high to low to improve the user experience.

[0063] For example, after identifying multiple conflicting feedback information groups, the similarity between each conflicting group and the evaluation feedback information of the first dimension can be calculated using a vector space model.

[0064] After determining the similarity between each contradictory feedback information group and the evaluation feedback information of the first dimension, multiple contradictory feedback information groups can be displayed to the user in descending order of priority index corresponding to each contradictory feedback information group. The weights of the contradictory dimensions of the user's feedback information in each contradictory feedback information group are obtained. The weights of the contradictory dimensions of the user's feedback information represent the importance of the contradictory dimensions of the user's feedback information. Therefore, multiple cooking curve segments can be optimized based on the importance of the contradictory dimensions of the user's feedback information.

[0065] For example, after obtaining the priority ranking of conflict feedback information groups, the system presents the corresponding optimization conflict options for each conflict group to the user through a graphical interface, in descending order of priority index. For instance, when the conflict group of ripeness and crispness is identified as the highest priority, the user can set the weight ratio of the ripeness dimension and the crispness dimension respectively by sliding the adjustment bar on the touch screen.

[0066] For example, the weight setting interface can simultaneously display the predicted impact of different weight combinations on the optimization results. For instance, increasing the brittleness weight may cause the system to preferentially extend the heating time in the high-temperature stage.

[0067] S230. Using the first cooking curve optimization model corresponding to the first dimension of evaluation feedback information, and based on the weights of the feedback information of each contradictory feedback information group, the mutually contradictory dimensions of each contradictory feedback information group, and the evaluation feedback information of other dimensions outside of multiple contradictory feedback information groups, optimize multiple cooking curve segments to obtain the optimized cooking curve corresponding to the first cooking curve.

[0068] Figure 5 A schematic diagram of the workflow of the second deep learning-driven personalized cooking curve generation method provided in this application embodiment is shown below. Figure 5As shown, after obtaining the weights of the feedback information of the mutually contradictory dimensions in each contradictory feedback information group, the first cooking curve optimization model corresponding to the first dimension's evaluation feedback information can be used to jointly optimize the temperature, humidity, and time parameters of multiple cooking curve segments through the first cooking curve optimization model, based on the weights of the feedback information of the mutually contradictory dimensions of each contradictory feedback information group and the evaluation feedback information of other dimensions outside of multiple contradictory feedback information groups. This optimization model yields the optimized cooking curve corresponding to the first cooking curve.

[0069] For example, when optimizing a set of contradictory feedback information including doneness and crispness, the heating rate and duration configuration of the core heating stage can be automatically balanced according to the weights of the contradictory dimensions of the feedback information set by the user, so as to improve the cooking effect.

[0070] In this implementation, the evaluation feedback information of other dimensions outside of the multiple contradictory feedback information groups is the evaluation data of non-contradictory dimensions. The evaluation feedback information of other dimensions outside of the multiple contradictory feedback information groups can be used as auxiliary constraints to participate in the optimization calculation. For example, when adjusting the ripeness-related parameters, the nutrient retention score will limit the maximum allowable duration of the high-temperature segment.

[0071] The beneficial effect of the above implementation method is that by automatically identifying and quantifying the contradictory relationships in multidimensional feedback information, it can effectively analyze the user's trade-off needs between different cooking effect indicators, thereby transforming subjective preferences into calculable optimization constraints.

[0072] The beneficial effects of the above implementation method are that, through dynamic priority sorting and visual weight setting mechanism, users can intuitively participate in the process of solving key contradictions, which not only ensures that the optimization direction matches the user's real needs, but also lowers the threshold for users to participate in technical decision-making, thus comprehensively improving the user experience.

[0073] The beneficial effect of the above implementation method is that by integrating multiple conflict feedback information groups and feedback data from other dimensions into the optimization model, it achieves synergistic optimization of local conflict resolution and global effect improvement, ensuring that the generated cooking curve meets the main conflict adjustment needs while taking into account the optimization effect of cooking quality corresponding to feedback information from other dimensions.

[0074] In some implementations, the above method also includes S240 to S250, which will be described in detail below.

[0075] S240. Obtain the working specification information of the oven, and obtain the relative weight values ​​between multiple contradictory feedback information groups corresponding to the working specification information. The relative weight values ​​represent the mutual influence relationship between multiple contradictory feedback information groups.

[0076] In this implementation, after obtaining multi-dimensional feedback information from users and performing contradiction group analysis, the working specifications information of the oven equipment can be further combined to optimize parameter adaptation, so as to improve the optimization effect of the cooking curve of ovens with different working specifications.

[0077] When obtaining the relative weight values ​​between multiple conflicting feedback information groups corresponding to work specification information, the relative weight values ​​between different conflicting feedback information groups can be determined through an empirical value table.

[0078] For example, by reading the technical parameter database corresponding to the oven model, the current oven's heating power range, temperature adjustment accuracy, humidity control capability, and other working specifications information can be obtained. By matching the oven's working specifications information with a preset cooking parameter constraint library, the relative weight values ​​between different contradictory feedback information groups can be obtained. The relative weight values ​​represent the mutual influence relationship between multiple contradictory feedback information groups.

[0079] For example, for ovens with low hot air circulation efficiency, the system may automatically increase the relative weight of the contradictory pair of texture uniformity and cooking efficiency in order to balance equipment performance and limit its impact on cooking results.

[0080] For example, when determining the relative weight values ​​between conflicting feedback information groups, a multi-dimensional weight relationship mapping model can be established based on the correlation analysis of work specification parameters and user historical operation data, and the relative weight values ​​between multiple conflicting feedback information groups corresponding to work specification information can be determined according to the weight relationship mapping model.

[0081] For example, when the oven's maximum temperature limit is detected to be low, the relative weight of the contradictory pair of color and burntness can be automatically increased, thereby prioritizing burntness control during the optimization process.

[0082] S250. Using the first cooking curve optimization model corresponding to the first multi-dimensional feedback information, based on the relative weight values ​​between multiple contradictory feedback information groups, the weight of feedback information of each contradictory feedback information group, the weight of the mutually contradictory dimensions of each contradictory feedback information group, and the evaluation feedback information of other dimensions outside of multiple contradictory feedback information groups, multiple cooking curve segments are optimized to obtain the optimized cooking curve corresponding to the first cooking curve.

[0083] In the cooking curve optimization stage, multi-objective joint optimization can be performed using a deep learning model based on the relative weight values ​​between multiple contradictory feedback information groups, the weight of feedback information of each contradictory feedback information group, the weight of feedback information of the mutually contradictory dimensions of each contradictory feedback information group, and the evaluation feedback information of other dimensions outside of multiple contradictory feedback information groups.

[0084] For example, by combining the relative weight values ​​of the contradiction between moisture retention and doneness corresponding to the oven's working specifications, the baking temperature in the core heating stage can be adjusted in a restrictive manner to ensure that the generated cooking curve not only conforms to the actual working capacity of the equipment, but also meets the user's personalized needs to the greatest extent.

[0085] The beneficial effect of the above implementation method is that by integrating equipment operating specification information and user feedback data through multi-dimensional correlation analysis, it can effectively solve the adaptation contradiction between personalized needs and equipment performance boundaries, and improve the equipment execution reliability of the optimization solution.

[0086] The beneficial effect of the above implementation method is that by establishing a dynamic weight fusion mechanism, the influence of equipment technical parameters on cooking effect can be quantified into calculable optimization constraints, thereby achieving an organic unity between objective equipment characteristics and subjective user preferences and improving the optimization effect of cooking curves.

[0087] In some implementations, the above method also includes S260 to S270, which will be described in detail below.

[0088] S260. Through the cooking contradiction identification unit, the surface contradiction feedback information group among multiple contradiction feedback information groups is identified. The feedback information of the two mutually contradictory dimensions of the surface contradiction feedback information group can be eliminated through the segmented optimization of the cooking curve.

[0089] In this implementation, when associating the weights of equipment specification information and conflict feedback information groups, the cooking conflict identification unit can also perform in-depth analysis of the conflict feedback information groups to reduce the number of conflict feedback information groups and improve the accuracy of optimizing and controlling the cooking curve.

[0090] When conducting in-depth analysis of contradictory feedback information groups, the cooking contradiction identification unit can be used to identify the surface contradictory feedback information groups among multiple contradictory feedback information groups. Based on the historical optimization case library and the cooking mechanism knowledge base, the cooking contradiction identification unit can conduct a reconciliation assessment of each contradictory feedback information group and screen out the surface contradictory feedback information groups that can be eliminated through segmented optimization of the cooking curve.

[0091] For example, when determining a group of surface contradictions, it can be identified by indicators such as the physical correlation strength of the feedback information of two mutually contradictory dimensions in multiple contradictory feedback information groups, the historical optimization success rate, and the adjustable space of equipment performance.

[0092] For example, the contradiction between ripeness and moisture retention can be identified as a superficial contradiction because the moisture retention effect can be improved without affecting the overall ripeness by adjusting the temperature rise rate in the preheating stage and the heating cycle in the core heating stage.

[0093] S270. Using a knowledge graph-based cooking curve optimization unit, the control optimization strategy corresponding to each contradictory feedback information group is determined based on the feedback information of the two mutually contradictory dimensions of each contradictory feedback information group. Based on the control optimization strategy corresponding to each contradictory feedback information group, multiple cooking curve segments are optimized to obtain multiple intermediate cooking curve segments. Using the first cooking curve optimization model corresponding to the first multi-dimensional feedback information, based on the relative weight values ​​between multiple contradictory feedback information groups, the weights of each contradictory feedback information group and the feedback information of the mutually contradictory dimensions of each contradictory feedback information group, and the evaluation feedback information of other dimensions besides the multiple contradictory feedback information groups, the optimized cooking curve corresponding to the first cooking curve is obtained.

[0094] After obtaining feedback information from two mutually contradictory dimensions for each contradictory feedback information group, a targeted control strategy can be generated for the identified surface contradictory feedback information group through a cooking curve optimization unit based on a knowledge graph.

[0095] For example, the relationship between cooking parameters and effect indicators can be stored in the knowledge graph, such as the causal chain of "high temperature short time heating → increased crispness + nutrient loss".

[0096] For example, when dealing with the conflict between ripeness and crispness, the optimization unit can resolve the strategy of "extending the low-temperature baking time in the post-ripening stage", so that the cooking strategy can improve crispness through surface crisping treatment while ensuring the core ripeness.

[0097] After obtaining the control optimization strategy corresponding to each set of conflicting feedback information, multiple cooking curve segments can be optimized based on the control optimization strategy corresponding to each set of conflicting feedback information to obtain multiple intermediate cooking curve segments. Each control strategy can generate a set of intermediate cooking curve segments. For example, by dividing the core heating stage into a rapid heating segment and a constant temperature penetration segment, parameters can be configured for cookedness control and crispness formation, respectively.

[0098] After obtaining multiple intermediate cooking curve segments, the relative weight values ​​between multiple contradictory feedback information groups, the weights of feedback information of each contradictory feedback information group, the dimensions of mutual contradictions in each contradictory feedback information group, and the evaluation feedback information of other dimensions outside of multiple contradictory feedback information groups can be integrated to optimize multiple intermediate cooking curve segments and obtain the optimized cooking curve corresponding to the first cooking curve. This achieves the integration of equipment-related weights, user-set weights, and other dimension data, and realizes the comprehensive optimization of the cooking curve.

[0099] The beneficial effect of the above implementation method is that by differentiating between surface contradictions and deep contradictions through a differentiated processing mechanism, it is possible to prioritize the resolution of contradiction groups that can be quickly reconciled through technical means, which can significantly improve optimization efficiency and reduce computational resource consumption, thereby improving the economic efficiency of oven control optimization.

[0100] The beneficial effect of the above implementation method is that by utilizing the cooking mechanism association rules stored in the knowledge graph, a scientific basis is provided for the generation of strategies for conflict groups, avoiding local optima caused by simply relying on data-driven approaches, and improving the cooking optimization effect.

[0101] The beneficial effect of the above implementation method is that by adopting a phased optimization architecture, first resolving decomposable contradictions through the intermediate segment, and then fine-tuning global parameters, it is possible to achieve a hierarchical integration of equipment adaptability optimization and user personalized needs, thereby improving the optimization effect of the oven.

[0102] Figure 6 A flowchart illustrating the third deep learning-driven personalized cooking curve generation method provided in this application embodiment is shown below. Figure 6 As shown, the above method also includes S310 to S320, which will be described in detail below.

[0103] S310. After the user bakes the second ingredient in the oven according to the first cooking curve to obtain the second baked product, the user obtains the second multi-dimensional feedback information on the second baked product. The second multi-dimensional feedback information includes evaluation feedback information in multiple dimensions, including evaluation feedback information on doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, charring degree, and cooking efficiency.

[0104] In this implementation, after optimizing the cooking curve of the first ingredient, feedback data from multiple ingredients can be further integrated to optimize the cooking curve across ingredients, thereby improving the optimization effect of the cooking curve.

[0105] After the user bakes the second ingredient in the oven according to the first cooking curve to obtain the second baked product, the user's feedback information on the second baked product can be obtained, and then the cooking curve can be optimized based on the second multidimensional feedback information of the second baked product.

[0106] S320. Obtain the second ingredient features corresponding to the second ingredient and the first ingredient features corresponding to the first ingredient. Based on the second ingredient features and the first ingredient features, determine the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multi-dimensional feedback information and the evaluation feedback information of multiple dimensions in the first multi-dimensional feedback information. Using a cooking curve segmentation model, divide the first cooking curve into multiple cooking curve segments based on the evaluation feedback information of the first dimension. Using a first cooking curve optimization model corresponding to the evaluation feedback information of the first dimension, optimize the multiple cooking curve segments based on the evaluation feedback information of other dimensions besides the first dimension, the evaluation feedback information of multiple dimensions in the second multi-dimensional feedback information, and the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multi-dimensional feedback information, to obtain the optimized cooking curve corresponding to the first cooking curve.

[0107] After obtaining the evaluation feedback information of multiple dimensions in the second multidimensional feedback information, the second ingredient features corresponding to the second ingredient and the first ingredient features corresponding to the first ingredient can be obtained. Based on the second ingredient features and the first ingredient features, the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information and the evaluation feedback information of multiple dimensions in the first multidimensional feedback information can be determined. Then, the cooking curve of the first ingredient can be optimized based on the cooking curve of the second ingredient.

[0108] For example, when acquiring the features of the second ingredient and the features of the first ingredient, the physicochemical characteristics of the first and second ingredients can be analyzed separately using the ingredient feature extraction module. For instance, for meat and root vegetables, the system can extract feature parameters such as moisture content, thermal conductivity, and protein denaturation temperature threshold to establish an ingredient feature vector library for calculating similarity weights.

[0109] For example, after obtaining multidimensional feedback information of the second ingredient, the similarity weights corresponding to the evaluation feedback information of multiple dimensions corresponding to the features of the two ingredients can be calculated using a feature space mapping algorithm. Specifically, for each evaluation dimension, a correlation model between the ingredient features and the feedback score can be constructed, and the similarity weights corresponding to the evaluation feedback information of multiple dimensions can be determined based on the correlation model between the ingredient features and the feedback score.

[0110] When optimizing the cooking curve, the first cooking curve can be divided into multiple cooking curve segments based on the evaluation feedback information of the first dimension using a cooking curve segmentation model, and then the cooking curve can be optimized for multiple cooking curve segments.

[0111] When optimizing cooking curves, the multidimensional feedback data of the first ingredient and the weighted feedback data of the second ingredient can be fused together. The first cooking curve optimization model corresponding to the first dimension evaluation feedback information can be used to optimize multiple cooking curve segments based on the similarity weights of the evaluation feedback information of other dimensions besides the first dimension, the evaluation feedback information of multiple dimensions in the second multidimensional feedback information, and the evaluation feedback information of multiple dimensions in the second multidimensional feedback information. This results in the optimized cooking curve corresponding to the first cooking curve. This cross-ingredient feedback fusion mechanism allows the optimization process to inherit the verified and effective parameter configurations while also making adaptive adjustments based on the characteristics of new ingredients.

[0112] For example, when optimizing the temperature parameters of the core heating stage, the model can simultaneously refer to the evaluation feedback information of the doneness of the first ingredient and the evaluation feedback information of the doneness of the second ingredient after the similarity weight adjustment, in order to optimize the cooking curve segment of the core heating stage.

[0113] The beneficial effect of the above implementation method is that, through cross-ingredient feature analysis and feedback data fusion, it is possible to break through the limitations of single ingredient optimization, form a transferable cooking parameter optimization strategy, and significantly improve the accuracy of predicting the cooking effect of new ingredients.

[0114] The beneficial effect of the above implementation method is that it uses a dynamic similarity weight mechanism to achieve appropriate generalization of personalized feedback data, which not only preserves user preference characteristics but also avoids overfitting to specific ingredients, thereby enhancing the applicability of the optimization scheme.

[0115] In some implementations, the above method also includes S330 to S340, which will be described in detail below.

[0116] S330: Obtain the first number of times the user bakes the first ingredient in the oven according to the first cooking curve, and obtain the second number of times the user bakes the second ingredient in the oven according to the first cooking curve.

[0117] In this implementation, after completing the fusion of cross-ingredient feedback data, the accuracy of weight adjustment can be further improved through cooking frequency analysis to enhance the optimization effect of the cooking curve. During optimization, the historical operation logs of users using specific cooking curves to process different ingredients can be recorded, and the number of times the first cooking curve bakes the first ingredient can be counted as the first cooking count, and the number of times the second cooking curve bakes the second ingredient can be counted as the second cooking count.

[0118] For example, when a user bakes chicken (first ingredient) a total of 20 times using the pizza mode (first cooking curve) and at the same time bakes vegetables (second ingredient) a total of 5 times using the same first cooking curve, the system will record these two values ​​as the first cooking count and the second cooking count, respectively.

[0119] S340. Determine the ratio of the second cooking count to the first cooking count, as the cooking count adjustment factor. Multiply the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information by the cooking count adjustment factor to adjust the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information.

[0120] When optimizing the cooking curve, after determining the first and second cooking times, the ratio of the second cooking times to the first cooking times can be calculated as a cooking times adjustment factor. After obtaining the cooking times adjustment factor, the cooking times adjustment factor can be multiplied by the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information to adjust the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information.

[0121] For example, when the standard number of cooking times is set to 10, the adjustment factor for the vegetable roasting case in the example above can be calculated as 5 / 20 = 0.25. The cooking number adjustment factor can comprehensively reflect the relative relationship of the current food roasting experience value. The cooking number adjustment factor can be used to dynamically calibrate the credibility weight of cross-food feedback data.

[0122] During the similarity weight adjustment process, the multidimensional feedback similarity weight of the second ingredient can be adjusted using a cooking count adjustment factor. For example, when the number of times the second ingredient is cooked is significantly lower than that of the first ingredient, the system will automatically reduce the weighting of sensitive dimensions such as crispness to prevent optimization bias caused by insufficient data samples.

[0123] The beneficial effect of the above implementation method is that, through dynamic analysis of cooking frequency data, it is possible to effectively identify differences in users' cooking experience with different ingredients, thereby improving the scientific rigor and reliability of cross-ingredient parameter transfer. By adjusting the similarity weights through a cooking frequency adjustment factor, a dynamic balance between data validity and user experience can be achieved.

[0124] In some implementations, the above method also includes S350 to S360, which will be described in detail below.

[0125] S350. Obtain multiple cooking health concerns input by the user, and determine the cooking health adjustment factors corresponding to the evaluation feedback information of multiple dimensions in the second multi-dimensional feedback information for each of the multiple cooking health concerns.

[0126] In this implementation, after adjusting the similarity weights by adjusting the cooking frequency factor, user health preference data can be further introduced to optimize the cooking parameters.

[0127] When optimizing based on user health preference data, multiple cooking health concerns actively input by users can be obtained through the interactive interface, such as specific dietary needs like low salt intake, low fat retention, and high fiber preservation.

[0128] After obtaining multiple cooking health concerns input by the user, it is possible to determine the cooking health adjustment factors corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information for each of the multiple cooking health concerns. This enables the construction of the association mapping relationship between each health concern and the multidimensional feedback evaluation dimensions. For example, the need for low-fat retention may correspond to the enhanced attention to the dimensions of nutrient retention and moisture retention.

[0129] For example, based on a health knowledge graph and a nutrition rule base, the cooking health adjustment factors corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information for multiple cooking health concerns can be automatically determined.

[0130] For example, when a user selects "control Maillard reaction products" as a health concern, the system will analyze the correlation path between the user's needs and dimensions such as color, charring, and nutrient retention, and generate corresponding cooking health adjustment factors.

[0131] S360. Multiply the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information by a cooking health adjustment factor to adjust the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information.

[0132] After obtaining the cooking health adjustment factor, the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information can be multiplied by the cooking health adjustment factor to adjust the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information.

[0133] For example, when optimizing the cooking curve of fish fillets, if the user sets the health concern of "preserving omega-3 fatty acids", the similar weight of the nutrient retention dimension can be multiplied by the dual adjustment of frequency confidence decay and health demand enhancement in S340. The adjusted comprehensive weight will guide the optimization model to prioritize the retention effect of specific nutrients, and can reduce the oxidative decomposition of unsaturated fatty acids by reducing the peak temperature of the core heating stage.

[0134] The beneficial effect of the above implementation method is that by integrating users' proactive health needs and objective feedback data, it is possible to achieve a precise match between nutritional goals and cooking process parameters, thereby enhancing the health value of personalized cooking solutions.

[0135] The beneficial effect of the above implementation method is that, by multiplying the similarity weights of the evaluation feedback information of multiple dimensions in the second multidimensional feedback information by the cooking health adjustment factor, and through the multi-factor coupling weight adjustment mechanism, the synergistic optimization of health goals, user experience and equipment performance can be achieved, forming an intelligent cooking solution that takes into account nutrition, taste and cooking efficiency.

[0136] Figure 7 A flowchart illustrating the fourth method for generating personalized cooking curves based on deep learning, as provided in this application embodiment, is shown below. Figure 7 As shown, the above method also includes S410 to S420, which are described in detail below.

[0137] S410. Obtain multiple first key dimension feedback information from the first multidimensional feedback information and multiple second key dimension feedback information from the second multidimensional feedback information. The first key dimension feedback information is the evaluation feedback information of the dimension in which the first ingredient plays a key role in baking, and the second key dimension feedback information is the evaluation feedback information of the dimension in which the second ingredient plays a key role in baking.

[0138] In this implementation, the universality of the optimization scheme can be further improved through key dimension intersection analysis. Specifically, multiple first key dimension feedback information in the first multidimensional feedback information and multiple second key dimension feedback information in the second multidimensional feedback information can be obtained to extract the first key dimension feedback information that plays a key role in the baking process of the first ingredient and the second key dimension feedback information corresponding to the second ingredient.

[0139] For example, for steak (the primary ingredient), doneness and charring might be identified as key dimensions, while for salmon (the secondary ingredient), nutrient retention and moisture retention might be identified as key dimensions. The determination of key dimensions can be based on the evaluation of the heat-sensitive characteristic parameters of the ingredients.

[0140] S420. Take the intersection of multiple first key dimension feedback information and multiple second key dimension feedback information to obtain multiple overlapping key dimension feedback information. Through the first cooking curve optimization model corresponding to the first multi-dimensional feedback information, optimize multiple cooking curve segments based on the evaluation feedback information of other dimensions besides the first dimension evaluation feedback information, the multiple overlapping key dimension feedback information, and the similarity weights corresponding to the multiple overlapping key dimension feedback information, to obtain the optimized cooking curve corresponding to the first cooking curve.

[0141] After obtaining multiple first key dimension feedback information from the first multidimensional feedback information and multiple second key dimension feedback information from the second multidimensional feedback information, the intersection of the multiple first key dimension feedback information and the multiple second key dimension feedback information can be taken to obtain multiple overlapping key dimension feedback information.

[0142] For example, when the key dimensions of steak include doneness and charring, and the key dimensions of salmon include doneness and moisture retention, the system will determine that doneness is the overlapping key dimension.

[0143] When optimizing the cooking curve, the first cooking curve optimization model corresponding to the first multi-dimensional feedback information can be used. Based on the evaluation feedback information of other dimensions besides the first dimension, the feedback information of multiple overlapping key dimensions, and the similarity weights corresponding to the feedback information of multiple overlapping key dimensions, multiple cooking curve segments can be optimized to obtain the optimized cooking curve corresponding to the first cooking curve, thus achieving the cooking curve optimization effect based on the feedback information of multiple overlapping key dimensions.

[0144] For example, regarding the overlapping dimension of doneness, priority can be given to ensuring the stability of parameter configuration for the overlapping dimension, maintaining the temperature gradient required for basic doneness formation during the preheating stage, and making differentiated fine adjustments based on the characteristics of different ingredients during the core heating stage.

[0145] The beneficial effects of the above implementation method are that by identifying the intersection features of key dimensions across ingredients, an optimization framework that takes into account both generality and specificity can be established, which can significantly improve the efficiency of parameter adjustment in multi-ingredient scenarios; by using overlapping dimensions as optimization anchor points, the core contradictions in cross-ingredient parameter migration can be effectively resolved, and the basic quality requirements of different ingredients can be met simultaneously.

[0146] In some implementations, S410 above involves obtaining multiple first key dimension feedback information from the first multidimensional feedback information and multiple second key dimension feedback information from the second multidimensional feedback information, including S411 to S412. S411 to S412 will be explained in detail below.

[0147] S411. Based on the characteristics of the first ingredient, determine multiple key dimension feedback information of the first ingredient corresponding to the first ingredient in the first multidimensional feedback information. Obtain multiple first historical multidimensional feedback information of the user's feedback on the first baked product in historical cooking data. Based on the characteristics of the second ingredient, determine multiple key dimension feedback information of the second ingredient corresponding to the second ingredient in the second multidimensional feedback information. Obtain multiple second historical multidimensional feedback information of the user's feedback on the second baked product in historical cooking data.

[0148] In this implementation, when determining the key dimension feedback information of the ingredients, dual verification can be achieved by combining the objective characteristics of the ingredients and the user's historical feedback data. First, the physicochemical characteristics of the first ingredient can be analyzed based on the ingredient composition analysis module, and then the multiple key dimensions of the ingredients that need to be focused on during the baking process can be determined. Based on the characteristics of the first ingredient, the multiple key dimension feedback information of the first ingredient corresponding to the first ingredient in the first multi-dimensional feedback information can be determined.

[0149] For example, for steaks with high fat content, the system may automatically determine the degree of charring and nutrient retention as key dimensions related to the characteristics of the ingredients. These dimensions are directly related to fat oxidation and the risk of nutrient loss.

[0150] Similarly, based on the characteristics of the second ingredient, multiple key dimension feedback information of the second ingredient can be determined in the second multidimensional feedback information; and multiple second historical multidimensional feedback information of the user's feedback on the second baked product in historical cooking data can be obtained.

[0151] S412. Take the intersection of multiple first-ingredient key dimension feedback information and multiple first-historical multi-dimensional feedback information to obtain multiple first-key dimension feedback information in the first multi-dimensional feedback information. Take the intersection of multiple second-ingredient key dimension feedback information and multiple second-historical multi-dimensional feedback information to obtain multiple second-key dimension feedback information in the second multi-dimensional feedback information.

[0152] While obtaining feedback information on multiple key dimensions of the first ingredient based on the characteristics of the ingredient, the system can retrieve the user's historical cooking feedback records for the same ingredient, thereby obtaining multiple first historical multidimensional feedback information on the first baked product from the historical cooking data. The system can then take the intersection of the multiple key dimension feedback information of the first ingredient and the multiple first historical multidimensional feedback information to obtain multiple first key dimension feedback information in the first multidimensional feedback information.

[0153] For example, when doneness and texture uniformity correspond to multiple key dimensions of the primary ingredient feedback information, when users provide feedback on the doneness and texture uniformity of steak products in the past three months, these two dimensions can be identified as historical key dimensions that users pay attention to. By taking the intersection of charring degree (characteristic dimension) and doneness (historical dimension) in the steak case, a key dimension combination that meets both the objective requirements of the ingredient and matches the user's subjective preferences can be selected.

[0154] Similarly, multiple second key dimension feedback information in the second multidimensional feedback information can be determined in a similar way.

[0155] The beneficial effect of the above implementation method is that by integrating the objective characteristics of ingredients and the historical feedback of users, a key dimension identification mechanism that balances scientificity and personalization can be established, thereby enhancing the practical application value of cooking curve optimization.

[0156] This application also provides a deep learning-driven personalized cooking curve generation system, including a unit for performing the method described in any of the preceding claims.

[0157] Figure 8 A schematic diagram of the logical structure of a deep learning-driven personalized cooking curve generation system provided in this application embodiment is shown below. Figure 8 As shown, the system 1 of this embodiment includes a processing unit 11, a storage unit 12, and a transceiver unit 13. The processing unit 11 is used to process data, the storage unit 12 is used to store data, and the transceiver unit 13 is used to send and receive data. The processing unit 11, the storage unit 12, and the transceiver unit 13 cooperate with each other to implement the above-described method. The beneficial effects of the embodiments of this application have been described in the above-described method and will not be repeated here.

[0158] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0159] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0160] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0161] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0162] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0163] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0164] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0165] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for generating personalized cooking curves based on deep learning, characterized in that, The method includes: After the user bakes the first ingredient in the oven according to the first cooking curve to obtain the first baked product, the user obtains the first multi-dimensional feedback information on the first baked product. The first multi-dimensional feedback information includes evaluation feedback information from multiple dimensions, including evaluation feedback information on doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, degree of charring, and cooking efficiency. Using a cooking curve segmentation model, the first cooking curve is divided into multiple cooking curve segments based on the user-defined first-dimensional evaluation feedback information. Using a first cooking curve optimization model corresponding to the first-dimensional evaluation feedback information, the multiple cooking curve segments are optimized based on evaluation feedback information from other dimensions besides the first-dimensional evaluation feedback information to obtain the optimized cooking curve corresponding to the first cooking curve. The method further includes: Through the feedback contradiction analysis unit, based on the first multidimensional feedback information, multiple contradictory feedback information groups are determined in the first multidimensional feedback information. Each contradictory feedback information group includes feedback information from two mutually contradictory dimensions. Among them, the multiple contradictory feedback information groups include: doneness and crispness, doneness and moisture retention, color and charring, aroma and nutrient retention, texture uniformity and cooking efficiency, crispness and cooking efficiency, and nutrient retention and cooking efficiency. Determine the similarity between each contradictory feedback information group and the evaluation feedback information of the first dimension, and use it as the priority index corresponding to each contradictory feedback information group; according to the priority index from high to low, show multiple contradictory feedback information groups to users, and obtain the weight of the user's feedback information on the contradictory dimensions for each contradictory feedback information group. The weight of the feedback information on the contradictory dimensions represents the importance of the user's feedback information on the contradictory dimensions. The first cooking curve optimization model, based on the evaluation feedback information of the first dimension, optimizes multiple cooking curve segments according to the weights of the feedback information of each contradictory feedback information group, the mutually contradictory dimensions of each contradictory feedback information group, and the evaluation feedback information of other dimensions outside of the multiple contradictory feedback information groups, to obtain the optimized cooking curve corresponding to the first cooking curve.

2. The method as described in claim 1, characterized in that, The method further includes: Obtain the oven's operating specifications information and the relative weight values ​​among multiple conflicting feedback information groups corresponding to the operating specifications information. The relative weight values ​​represent the mutual influence relationship among multiple conflicting feedback information groups. The first cooking curve optimization model corresponding to the first multidimensional feedback information is used to optimize multiple cooking curve segments based on the relative weight values ​​between multiple contradictory feedback information groups, the weight of feedback information of each contradictory feedback information group, the weight of the mutually contradictory dimensions of each contradictory feedback information group, and the evaluation feedback information of other dimensions outside of multiple contradictory feedback information groups, so as to obtain the optimized cooking curve corresponding to the first cooking curve.

3. The method as described in claim 2, characterized in that, The method further includes: The cooking contradiction identification unit identifies the surface contradiction feedback information group among multiple contradiction feedback information groups. The feedback information of the two mutually contradictory dimensions of the surface contradiction feedback information group can be eliminated through the segmented optimization of the cooking curve. Using a knowledge graph-based cooking curve optimization unit, the control optimization strategy corresponding to each contradictory feedback information group is determined based on the feedback information from two mutually contradictory dimensions of each contradictory feedback information group. Based on the control optimization strategy corresponding to each contradictory feedback information group, multiple cooking curve segments are optimized to obtain multiple intermediate cooking curve segments. Using the first cooking curve optimization model corresponding to the first multi-dimensional feedback information, the multiple intermediate cooking curve segments are optimized based on the relative weight values ​​between multiple contradictory feedback information groups, the weights of each contradictory feedback information group and the feedback information from the mutually contradictory dimensions of each contradictory feedback information group, and the evaluation feedback information from other dimensions outside the multiple contradictory feedback information groups, resulting in the optimized cooking curve corresponding to the first cooking curve.

4. The method as described in claim 3, characterized in that, The method further includes: After the user bakes the second ingredient in the oven according to the first cooking curve to obtain the second baked product, the user's feedback on the second baked product is obtained. The second multi-dimensional feedback information includes evaluation feedback information from multiple dimensions, including evaluation feedback information on doneness, crispness, color, aroma, texture uniformity, moisture retention, nutrient retention, degree of charring, and cooking efficiency. Obtain the second ingredient features corresponding to the second ingredient and the first ingredient features corresponding to the first ingredient. Based on the second ingredient features and the first ingredient features, determine the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information and the evaluation feedback information of multiple dimensions in the first multidimensional feedback information. Using a cooking curve segmentation model, divide the first cooking curve into multiple cooking curve segments based on the evaluation feedback information of the first dimension. Using a first cooking curve optimization model corresponding to the first multidimensional feedback information, optimize the multiple cooking curve segments based on the evaluation feedback information of other dimensions besides the first dimension, the evaluation feedback information of multiple dimensions in the second multidimensional feedback information, and the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information, to obtain the optimized cooking curve corresponding to the first cooking curve.

5. The method as described in claim 4, characterized in that, The method further includes: The system obtains the first number of times the user bakes the first ingredient in the oven according to the first cooking curve, and the second number of times the user bakes the second ingredient in the oven according to the first cooking curve. The quotient of the second cooking count divided by the first cooking count and the standard cooking count is determined as the cooking count adjustment factor. The cooking count adjustment factor is multiplied by the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information to adjust the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information.

6. The method as described in claim 5, characterized in that, The method further includes: Get multiple cooking health concerns input by the user, and determine the cooking health adjustment factors corresponding to the evaluation feedback information of multiple dimensions in the second multi-dimensional feedback information for each of the multiple cooking health concerns. In the second multidimensional feedback information, the similarity weights corresponding to the evaluation feedback information of multiple dimensions are multiplied by a cooking health adjustment factor to adjust the similarity weights corresponding to the evaluation feedback information of multiple dimensions in the second multidimensional feedback information.

7. The method as described in claim 6, characterized in that, The method further includes: Obtain multiple first key dimension feedback information from the first multidimensional feedback information and multiple second key dimension feedback information from the second multidimensional feedback information. The first key dimension feedback information is the evaluation feedback information of the dimension in which the first ingredient plays a key role in baking, and the second key dimension feedback information is the evaluation feedback information of the dimension in which the second ingredient plays a key role in baking. The intersection of multiple first key dimension feedback information and multiple second key dimension feedback information is obtained to obtain multiple overlapping key dimension feedback information. Through the first cooking curve optimization model corresponding to the first multidimensional feedback information, the evaluation feedback information of other dimensions besides the first dimension evaluation feedback information, the multiple overlapping key dimension feedback information and the similarity weights corresponding to the multiple overlapping key dimension feedback information are optimized to obtain the optimized cooking curve corresponding to the first cooking curve.

8. The method as described in claim 7, characterized in that, Obtain multiple first key dimension feedback information from the first multidimensional feedback information and multiple second key dimension feedback information from the second multidimensional feedback information, including: Based on the characteristics of the first ingredient, determine multiple key dimension feedback information of the first ingredient in the first multidimensional feedback information; obtain multiple first historical multidimensional feedback information of the user's feedback on the first baked product in historical cooking data; based on the characteristics of the second ingredient, determine multiple key dimension feedback information of the second ingredient in the second multidimensional feedback information; obtain multiple second historical multidimensional feedback information of the user's feedback on the second baked product in historical cooking data; The intersection of multiple first-factor key dimension feedback information and multiple first-historical multi-dimensional feedback information is obtained to obtain multiple first-factor key dimension feedback information in the first multi-dimensional feedback information; the intersection of multiple second-factor key dimension feedback information and multiple second-historical multi-dimensional feedback information is obtained to obtain multiple second-factor key dimension feedback information in the second multi-dimensional feedback information.

9. A personalized cooking curve generation system based on deep learning, characterized in that, Includes a unit for performing the method according to any one of claims 1 to 8.

Images