Man-machine interaction cooking system based on multi-dimensional data monitoring

Abstract

The invention discloses a man-machine interaction cooking system based on multi-dimensional data monitoring, which relates to the technical field of intelligent cooking and comprises a data monitoring module, an analysis interaction module, a dish preparation module, a cooking module and an interaction module. The dish preparation module gives an instruction to the dish selected by the current user based on the database and indicates the user how to prepare the dish, and after the user selects the instruction, the next operation is carried out based on the interaction module; after the cooking module obtains the next operation instruction, cooking instructions including food material feeding and cover lifting opportunities and stir-frying opportunities and time are carried out, real-time display is carried out through the interaction module, and a user carries out operation based on the displayed instructions. The data monitoring module acquires temperature, color and time data in real time, provides a basis for the analysis interaction module, realizes close cooperation, realizes ordered operation by a user according to instructions and prompts from dish preparation to cooking, avoids errors and time waste, enables the cooking process to be smoother and more efficient, and saves time and energy.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent cooking, and specifically to a human-computer interaction cooking system based on multi-dimensional data monitoring. Background Art

[0002] Currently, cooking poses great difficulties for many people who are not good at cooking. Traditional cooking methods rely on personal experience, and it is very difficult for novices to master skills such as proper ingredient processing, cooking time, and heat control, resulting in dishes that are difficult to achieve the desired effect. Existing cooking aids have single functions and cannot provide comprehensive and intelligent cooking guidance, failing to meet users' needs for convenient, efficient, and precise cooking. For this reason, we propose a human-computer interaction cooking system based on multi-dimensional data monitoring. Summary of the Invention

[0003] To solve the above technical problems, a human-computer interaction cooking system based on multi-dimensional data monitoring is provided, and this technical solution solves the problem of the single function of the above-mentioned cooking aids.

[0004] To achieve the above object, the technical solution adopted by the present invention is: a human-computer interaction cooking system based on multi-dimensional data monitoring, including: a data monitoring module, an analysis and interaction module, a food preparation module, a cooking module, and an interaction module;

[0005] The food preparation module gives instructions on the dishes prepared by the current user based on the database, indicating how the user should prepare the ingredients. After the user selects the instructions, the next operation is carried out based on the interaction module.

[0006] After the cooking module obtains the next operation instruction, it gives cooking instructions, including lifting and covering the pot lid, the timing and time of ingredient placement, and the timing and time of stir-frying, which are displayed in real time through the interaction module, and the user operates based on the displayed instructions.

[0007] The data monitoring module monitors the temperature in the pot, the color data of the dishes, and the time of cooking operations in real time, and uploads the collected data to the analysis and interaction module.

[0008] Based on the obtained data, the analysis and interaction module evaluates and analyzes the timing of cooking operations for the current dishes, and gives prompts for the time of stir-frying, the timing and time of lifting and covering the pot lid, and the timing of ingredient placement. The user operates based on the prompts.

[0009] The interaction module interacts with the user in real time, and the user performs operations based on the interaction information.

[0010] Preferably, inside the cooking system, through big data in advance, the cooking recipes of various dishes are obtained and constructed into a recipe database. The recipe includes the detailed cooking methods of the dishes, including the steps of preparing ingredients, cooking steps, and cooking operation time. The ingredient preparation module operates based on the constructed recipe database. After the user selects a dish on the interaction module, the ingredient preparation module extracts the ingredient preparation data of the dish from the database, converts the data into instructions, and presents them to the user through the display interface of the interaction module. The instructions include the types, quantities, and processing techniques of the required ingredients, arranged in the order of ingredient preparation, guiding the user to operate.

[0011] Preferably, after obtaining the instructions, the cooking module issues cooking instructions, accurate to the order of ingredient placement, the timing of placement, and the specific quantity of each placement. The method for obtaining cooking instructions is to establish a cooking model, predict the instructions based on the real-time collected data, so as to obtain the instructions for the next operation. The cooking model is established through deep learning algorithms to establish cooking models for different dishes. The models are trained based on cooking experiment data and the experience data of professional chefs to improve the models.

[0012] Preferably, the steps for establishing the cooking model are as follows: There are M training samples, and each sample contains an input feature vector x i and the corresponding target cooking instruction vector y i , where i = 1, 2,..., M. The input feature vector x i contains various real-time data during the cooking process, such as time t i , temperature T i , and the ingredient state feature vector s i is expressed as The target cooking instruction vector y i represents the cooking instruction that should be executed at this moment, the frequency of placing a certain ingredient and stir-frying. Combine all the input feature vectors into an input matrix X ∈ R M×n , where n is the dimension of the input features; combine all the target cooking instruction vectors into a target matrix Y ∈ R M×m , where m is the dimension of the target instructions; use a multi-layer perceptron as the cooking model, including an input layer, L hidden layers, and an output layer. For the l-th layer, l = 1,..., L + 1, l = 1 is the input layer, l = L + 1 is the output layer. Suppose there are h l neurons in this layer. The input and output of the j-th neuron in the l-th layer are calculated as follows:

[0013] Input calculation:

[0014] Among them, is the weight between the j-th neuron in the l-th layer and the k-th neuron in the l - 1-th layer. is the bias of the j-th neuron in the l-th layer, is the output of the k-th neuron in the (l-1)-th layer;

[0015] Output calculation:

[0016] f is the activation function, which is the ReLU function: f(x) = max(0, x), and in matrix form it is:

[0017] z (l) = W (l) a (l-1) + b (l)

[0018] a (l) = f(z (l) ) where, is the weight matrix of the l-th layer, is the bias vector of the l-th layer, a( 0 ) = x is the input feature vector;

[0019] The constructed cooking model is obtained by performing forward propagation, backward propagation, and parameter update.

[0020] Preferably, after the model training is completed, for the newly real-time collected data xnew, the predicted cooking instruction is calculated through forward propagation

[0021] Initialization: a( 0 ) = x new , for l from 1 to L+1 layers, perform successively:

[0022] z (l) = W (l) a (l-1) + b (l)

[0023] a (l) = f(z (l) )

[0024] Final prediction: is the next cooking operation instruction predicted based on the real-time collected data. By continuously iterating the above forward propagation, backward propagation, and parameter update processes, the model gradually learns the patterns and rules in the cooking data, so as to accurately predict the cooking instruction according to the input real-time data.

[0025] Preferably, the data monitoring module is used to continuously and real-time monitor the data during the cooking process. In terms of temperature monitoring, a temperature sensor is adopted and installed at the specified position required by the human-computer interaction cooking system. The temperature sensor can directly sense the heat transfer situation at the contact part between the pot body and the ingredients, measure the real-time temperature of the pot body and the temperature of the ingredients in real time;

[0026] For the monitoring of the dish color data, the data monitoring module uses image recognition technology. Based on the state of the dish in the pot captured by the camera, the camera captures the image of the dish in real time and transmits the image to the image analysis processor. The image analysis processor extracts and analyzes the color information in the image based on the deep learning algorithm, identifies the color change stage of the dish during the cooking process, and judges the cooking progress of the dish by comparing with the preset color database;

[0027] The monitoring of the cooking operation time is realized through the built-in time counter. When the user starts the cooking operation, the time counter is automatically started to record the cooking operation duration, and the cooking operation time must be completed within the time specified by the system at that time.

[0028] After completing the data collection, the data monitoring module uploads the data to the analysis and interaction module for analysis.

[0029] Preferably, the analysis and interaction module first preprocesses the data transmitted by the data monitoring module, cleans up the missing values, outliers and normalizes them. Then it extracts the key features from the temperature, color and time data, matches the preset cooking model in the system, inputs the features to predict the cooking stage and operation time. During cooking, the module compares the actual data with the predicted data in real time. If there is a deviation, it dynamically adjusts the instructions. After determining the operation time, it generates prompts in the form of text, voice and visualization. After the user's operation, it feeds back the results, and the module continues to evaluate and analyze accordingly to guide the user to complete the cooking process.

[0030] Preferably, the specific analysis process in the analysis and interaction module is as follows: First, complete the filling of the missing values; correct the abnormal data beyond the normal range according to the set reasonable range to complete the data cleaning. Then, according to their respective characteristics, uniformly convert the cleaned data to the numerical range from 0 to 1 to achieve data normalization.

[0031] Preferably, extract the key features from the data. For the temperature data, calculate the rate of change of the temperature at adjacent time points and record the highest temperature and the lowest temperature during the whole process. For the color data, calculate the average value of the color and the degree of dispersion of the color, and extract the main color features. For the time data, calculate the duration of each cooking operation interval and the total duration of continuous cooking operations, and integrate them together. Input the extracted features into this model, and the model will predict which stage of cooking and the operation time are currently in.

[0032] Preferably, the interaction module includes device screen interaction and voice interaction. The intelligent device screen displays text steps, operation diagrams, and teaching videos, while voice interaction allows users to query cooking steps and remaining time through voice commands. The interaction content includes the ingredient preparation and cooking stages. During ingredient preparation, an ingredient list, usage amounts, selection, and processing techniques are displayed. During cooking, the progress is updated in real-time based on the cooking model and real-time data. After receiving the information, the user operates according to the prompts. After completion, through manual feedback, the interaction module analyzes the feedback. If the operation meets the standards, the next step of guidance is pushed; if there are deviations, corrections and prompts are given in a timely manner to ensure the smooth completion of cooking until the dish is cooked.

[0033] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0034] The present invention provides detailed instructions based on the database, enabling users to clearly know the required ingredients and processing techniques. The analysis interaction module accurately evaluates based on data and timely prompts the frying time and moment, the timing of lifting and covering the pot lid, and the timing of adding ingredients, effectively solving the key problems in the cooking process and making cooking no longer difficult. Users can not only intuitively obtain information through the screen but also receive instructions through voice interaction when their hands are busy, achieving natural interaction with the system. The data monitoring module collects temperature, color, and time data in real-time, providing a basis for the analysis interaction module. They cooperate closely. From ingredient preparation to cooking, users operate in an orderly manner according to the instructions and prompts, avoiding mistakes and time waste, making the cooking process smoother and more efficient, saving time and energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a framework diagram of the human-computer interaction cooking system of the present inventor. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations.

[0037] Referring to Figure 1 As shown, the human-computer interaction cooking system based on multi-dimensional data monitoring includes: a data monitoring module, an analysis interaction module, an ingredient preparation module, a cooking module, and an interaction module;

[0038] The ingredient preparation module gives instructions based on the database for the currently selected dish by the user, indicating how the user should prepare the ingredients. After the user selects the instructions, the next operation is carried out based on the interaction module.

[0039] After the cooking module obtains the next operation instruction, it gives cooking instructions, including the frying time and moment, the timing of adding ingredients, and the timing of lifting and covering the pot lid, which are displayed in real-time through the interaction module. The user operates based on the displayed instructions.

[0040] The data monitoring module monitors the temperature in the pot, the color data of the dish, and the time of the cooking operation in real time, and uploads the collected data to the analysis and interaction module;

[0041] Based on the acquired data, the analysis and interaction module evaluates and analyzes the timing of the current dish's cooking operation, and gives prompts for the timing and time of stir-frying, the timing of lifting and covering the pot lid, and the timing of adding ingredients. The user operates based on the prompts;

[0042] The interaction module interacts with the user in real time, and the user performs operations based on the interactive information.

[0043] The ingredient preparation module of this application provides detailed instructions based on the database, enabling the user to clearly know the required ingredients and the processing methods, and even those without any cooking experience can easily get started. The analysis and interaction module accurately evaluates based on the data, and timely prompts the timing and time of stir-frying, the timing of lifting and covering the pot lid, and the timing of adding ingredients, effectively solving the key problems in the cooking process, making cooking no longer difficult. With real-time interaction, the cooking experience is improved. The interaction module supports graphical interfaces and voice interaction to meet different scenario requirements. The user can not only intuitively obtain information through the screen, but also receive instructions through voice interaction when their hands are busy, realizing natural interaction with the system and making the cooking process more relaxed and pleasant. The data monitoring module collects temperature, color, and time data in real time, providing a basis for the analysis and interaction module, which dynamically adjusts the instructions accordingly to ensure that the cooking process is always in the best state, maximally guaranteeing the color, aroma, taste, and appearance of the dish, and improving the cooking success rate. Each module cooperates closely. From ingredient preparation to cooking, the user operates in an orderly manner according to the instructions and prompts, avoiding mistakes and time waste, making the cooking process smoother and more efficient, saving time and energy.

[0044] Inside the cooking system, through big data in advance, the production recipes of various dishes are obtained and built into a recipe database. The recipes include the detailed cooking methods of the dishes, including ingredient preparation steps, cooking steps, and cooking time. The ingredient preparation module operates based on the built recipe database. After the user selects a dish on the interaction module, the ingredient preparation module extracts the ingredient preparation data of the dish from the database, converts the data into instructions, and presents them to the user through the display interface of the interaction module. The instructions include the types, amounts, and processing techniques of the required ingredients, arranged in the order of ingredient preparation, guiding the user to operate.

[0045] The recipe database constructed by this application through big data covers all kinds of cooking methods. After the user selects the dish, the preparation module can accurately extract the data and convert it into instructions. From the type of ingredients, dosage to the processing techniques, they are all arranged and displayed in order. Even cooking novices can prepare the ingredients in an orderly manner according to the steps, reduce the probability of errors, and ensure the smooth progress of subsequent cooking. The rich recipe database contains dishes of various cuisines and flavors. Whether it is home-cooked dishes or special dishes, it can meet the dietary preferences and needs of different users. Whether it is pursuing healthy light meals or wanting to try complex delicacies, users can easily find the corresponding recipes; users no longer need to spend a lot of time looking for recipes and sorting out the preparation steps on the Internet or in books. The cooking system directly provides one-stop service. The system automatically plans the preparation order to avoid users being in a hurry because of thinking about the steps and dosage, making the preparation process more efficient and convenient.

[0046] After the cooking module obtains the instructions, it issues the cooking instructions, accurately indicating the order, timing and specific amount of ingredients added each time. The cooking instruction acquisition method establishes a cooking model and predicts the instructions based on the real-time collected data, so as to obtain the instructions for the next operation. The cooking model is established through a deep learning algorithm to establish cooking models for different dishes. The model is trained based on cooking experimental data and the experience data of professional chefs to improve the model.

[0047] With the help of deep learning algorithms, this application uses a model trained based on a large number of cooking experiments and professional chefs' experience data to accurately predict the cooking instructions of different dishes. Whether it is the order, timing or specific amount of ingredients, it can be accurately given according to the unique needs and real-time data of each dish, meeting the user's personalized cooking requirements for different dishes and making the cooking process more scientific and accurate. The data collected in real time can enable the cooking model to timely perceive various changes in the cooking process, such as temperature fluctuations and changes in the state of ingredients. The model dynamically adjusts the cooking instructions based on these changes, and can give appropriate operational guidance to effectively respond to various emergencies and ensure smooth cooking.

[0048] The steps of building a cooking model are as follows: Suppose there are M training samples, each of which contains an input feature vector x i and the corresponding target cooking instruction vector y i , where i = 1, 2, ..., M, input feature vector x i Contains various real-time data during the cooking process, such as time t i , Temperature T i , food state feature vector s i Expressed as Target cooking instruction vector y i Indicates the cooking instructions that should be executed at this moment, the frequency of adding certain ingredients and stir-frying, and combines all input feature vectors into an input matrix X∈RM×n , where n is the dimension of the input features; all target cooking instruction vectors are combined into a target matrix Y ∈ R M×m , where m is the dimension of the target instructions; a multi-layer perceptron is used as the cooking model, which includes an input layer, L hidden layers, and an output layer. For the l-th layer, l = 1, …, L + 1, l = 1 is the input layer, and l = L + 1 is the output layer. Let there be h l neurons in this layer, and the input and output of the j-th neuron in the l-th layer are calculated as follows:

[0049] Input calculation:

[0050] where, is the weight between the j-th neuron in the l-th layer and the k-th neuron in the l - 1-th layer, is the bias of the j-th neuron in the l-th layer, is the output of the k-th neuron in the l - 1-th layer;

[0051] Output calculation:

[0052] f is the activation function, which is the ReLU function: f(x) = max(0, x), and in matrix form it is expressed as:

[0053] z (l) = W (l) a (l-1) + b (l)

[0054] a (l) = f(z (l) ) where, is the weight matrix of the l-th layer, is the bias vector of the l-th layer, a( 0 ) = x is the input feature vector;

[0055] The constructed cooking model is obtained by performing forward propagation, backpropagation, and parameter update.

[0056] The multi-layer perceptron of this application performs non-linear transformation on the input features through multiple hidden layers, and can automatically learn the complex patterns and rules in the cooking process. Based on a large number of training samples containing real-time data, the model can not only master the instructions in common cooking scenarios, but also make reasonable inferences for new and not completely identical cooking situations, realizing generalization applications and coping with various different cooking conditions and ingredient combinations; using the real-time data of time, temperature, and ingredient status as the input feature vector comprehensively covers the key factors affecting the cooking results.

[0057] After the model training is completed, for the new real-time collected data xnew, the predicted cooking instructions are obtained through forward propagation calculation

[0058] Initialization: a( 0 )=x new , for layers l=1 to L+1, proceed in sequence:

[0059] z (l) =W (l) a (l-1) +b (l)

[0060] a (l) =f(z (l) )

[0061] Final prediction: The next cooking operation instructions are predicted based on real-time collected data. Through continuous iteration of the above-mentioned forward propagation, backpropagation and parameter update processes, the model gradually learns the patterns and rules in the cooking data, so that it can accurately predict cooking instructions based on the input real-time data.

[0062] Through forward propagation, the model of this application can calculate the next precise cooking instructions based on new real-time collected data, such as the temperature during cooking and the state of the ingredients, according to the learned patterns and rules, including the timing of adding ingredients and lifting the lid, and the frequency of stir-frying, so that users can cook more scientifically. The real-time collected data reflects the dynamic changes in the cooking process. The model makes predictions based on these continuously updated data and can adapt to the complex situations of different ingredient characteristics to ensure that reasonable cooking suggestions can be given in various scenarios.

[0063] The data monitoring module is used to continuously monitor the data in the cooking process in real time. In terms of temperature monitoring, a temperature sensor is used and installed at the designated position required by the human-computer interactive cooking system. The temperature sensor can directly sense the heat transfer between the pot and the food, measure the real-time temperature of the pot and the temperature of the food in real time;

[0064] For the monitoring of dish color data, the data monitoring module uses image recognition technology to shoot the state of the dish in the pot based on the camera. The camera captures the image of the dish in real time and transmits the image to the image analysis processor. The image analysis processor extracts and analyzes the color information in the image based on the deep learning algorithm, identifies the color change stage of the dish during the cooking process, and judges the cooking progress of the dish by comparing it with the preset color database;

[0065] The monitoring of cooking operation time is achieved through a built-in time counter. When the user starts a cooking operation, the time counter is automatically started to record the duration of the cooking operation, and the cooking operation time must be completed within the time specified by the system at that time.

[0066] After completing data collection, the data monitoring module uploads the data to the analysis and interaction module for analysis.

[0067] In this application, the temperature sensor can accurately measure the temperature of the directly heated part of the food material, grasp the overall temperature distribution in the pot, comprehensively reflect the heat change during the cooking process, provide a scientific basis for cooking temperature control, prevent the food material from being undercooked or burnt due to improper temperature, monitor the color change of the dish by means of image recognition technology and deep learning algorithm, and can accurately identify the cooking stage by comparing with the preset color database. For example, it can judge whether the steak has reached the required doneness, effectively avoid cooking mistakes caused by lack of experience, and improve the cooking quality.

[0068] The analysis and interaction module first preprocesses the data transmitted by the data monitoring module, cleans up the missing values, outliers and normalizes them. Then it extracts key features from the temperature, color and time data, matches the preset cooking model in the system, inputs the features to predict the cooking stage and operation time. During cooking, the module compares the actual data with the predicted data in real time. If there is a deviation, it dynamically adjusts the instructions. After determining the operation time, it generates prompts in the form of text, voice and visualization. After the user's operation, it feeds back the result, and the module continues to evaluate and analyze accordingly to guide the user to complete the cooking process.

[0069] This application preprocesses the data, cleans up the missing values and outliers and normalizes them, ensuring the accuracy and consistency of the input data. This makes the subsequent feature extraction and model prediction more reliable, provides a solid data foundation for the stable operation of the entire cooking system, avoids wrong instructions caused by data problems, improves the credibility of cooking guidance, and can accurately predict the cooking stage and operation time based on real-time data by extracting key features and matching the preset cooking model. Whether it is the timing of putting in the food materials, the timing and time of stir-frying, or the timing of lifting or covering the pot lid, it can give scientific and reasonable suggestions to help the user master the best cooking rhythm and improve the quality and taste of the dish.

[0070] The specific analysis process in the analysis and interaction module is as follows: First, the missing values are filled in completely; the abnormal data beyond the normal range is corrected according to the set reasonable range to complete data cleaning. Then, the cleaned data is uniformly converted to the numerical range from 0 to 1 according to their respective characteristics to achieve data normalization.

[0071] By supplementing the missing values, this application can avoid analysis biases or errors caused by missing data. During the cooking process, for example, the missing of key data such as temperature and time may affect the accurate judgment of the cooking stage. After supplementing the missing values, the data can be made more complete, thus providing a more reliable basis for subsequent analysis. The correction of abnormal data can exclude unreasonable data caused by measurement errors, equipment failures or other abnormal factors, ensuring that the data reflects the real cooking situation and improving the accuracy of the data. Data normalization converts data with different characteristics and dimensions into a numerical range from 0 to 1, making various data have the same scale. In cooking data, the original dimensions and value ranges of temperature, color values, and time data vary greatly. After normalization, these data can be compared and analyzed on the same scale, which is convenient for the model to better learn the relationships and patterns between the data. This helps to improve the training effect and prediction accuracy of the model, making the analysis results more reliable and persuasive.

[0072] Extract key features from the data. For temperature data, calculate the rate of change of temperature between adjacent time points and record the highest and lowest temperatures during the whole process. For color data, calculate the average value of the color and the degree of color dispersion, and extract the main color features. For time data, calculate the duration of each cooking operation interval and the total duration of continuous cooking operations. Integrate them together and input the extracted features into this model, and the model will predict which stage of cooking and the operation time are currently in.

[0073] By extracting the rate of change, the highest temperature and the lowest temperature of the temperature data, this application can accurately grasp the change of heat during the cooking process. For example, when frying steak, the rate of change of temperature can reflect the size of the heat, and the highest and lowest temperatures can help judge whether the steak is fried to the appropriate degree of doneness. The average value, degree of dispersion and main color features of the color data can intuitively reflect the color change of the dish. For example, the color of braised pork will gradually darken during the cooking process. These features can be used as important bases for judging the cooking progress. The duration of the cooking operation interval and the total duration of continuous cooking operations in the time data can quantify the key operation of stir-frying in the cooking operation, which helps to understand the evenness of the heat received by the ingredients, thus accurately reflecting the actual state of cooking.

[0074] The interaction module includes device screen interaction and voice interaction. The intelligent device screen displays text steps, operation diagrams, and instructional videos; voice interaction allows users to query cooking steps and remaining time through voice commands. The interaction content includes the ingredient preparation and cooking stages. During ingredient preparation, the ingredient list, quantities, selection, and processing techniques are displayed; during cooking, the progress is updated in real time based on the cooking model and real-time data. After receiving the information, the user operates according to the prompts. After completion, through manual feedback, the interaction module analyzes the feedback. If the operation meets the standards, the next step of guidance is pushed; if there are deviations, prompt corrections are made in a timely manner to ensure the smooth completion of cooking until the dish is cooked.

[0075] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements fall within the scope of the present invention claimed.

Claims

1. A human-machine interactive cooking system based on multi-dimensional data monitoring, characterized in that: include: Data monitoring module, analysis and interaction module, food preparation module, cooking module and interaction module; The food preparation module gives instructions for the dishes currently selected by the user based on the database, instructing the user on how to prepare the dishes. After the user selects the instructions, the next step is performed based on the interactive module; After the cooking module obtains the next operation instruction, it performs the cooking instruction, including the timing of adding ingredients and lifting the lid, and the timing and time of stir-frying, which are displayed in real time through the interactive module, and the user operates based on the displayed instructions; The data monitoring module monitors the temperature in the pot, the color data of the dish, and the time and timing of the cooking operation in real time, and uploads the collected data to the analysis and interaction module; The analysis and interaction module evaluates and analyzes the timing of the cooking operation of the current dish based on the acquired data, and prompts the user when to stir-fry, lift the lid and add ingredients after the cooking time of the dish is reached, and the user operates based on the prompts; The interactive module interacts with the user in real time, and the user performs operations based on the interactive information.

2. The human-machine interactive cooking system based on multi-dimensional data monitoring according to claim 1, characterized in that: The cooking system uses big data to obtain recipes for various dishes in advance and builds a recipe database. The recipes include detailed methods of making the dishes, including preparation steps, cooking steps and cooking time. The preparation module operates based on the constructed recipe database. After the user selects a dish on the interactive module, the preparation module extracts the preparation data of the dish from the database, converts the data into instructions, and presents them to the user through the interactive module display interface. The instructions include the types, quantities and processing techniques of the required ingredients, which are arranged in sequence according to the order of preparation to guide the user to perform the operation.

3. The human-computer interactive cooking system based on multi-dimensional data monitoring according to claim 1, characterized in that: After the cooking module obtains the instructions, it issues the cooking instructions, accurately indicating the order, timing and specific amount of ingredients added each time. The cooking instruction acquisition method establishes a cooking model and predicts the instructions based on the real-time collected data, so as to obtain the instructions for the next operation. The cooking model is established through a deep learning algorithm to establish cooking models for different dishes. The model is trained based on cooking experimental data and the experience data of professional chefs to improve the model.

4. The human-machine interactive cooking system based on multi-dimensional data monitoring according to claim 3 is characterized in that: The steps of building a cooking model are as follows: Suppose there are M training samples, each of which contains an input feature vector x i and the corresponding target cooking instruction vector y i , where i = 1, 2, ..., M, input feature vector x i Contains various real-time data during the cooking process, such as time t i , Temperature T i , food state feature vector s i Expressed as Target cooking instruction vector y i Indicates the cooking instructions that should be executed at this moment, the frequency of adding certain ingredients and stir-frying, and combines all input feature vectors into an input matrix X∈R M×n , where n is the dimension of the input features; All target cooking instruction vectors are combined into the target matrix Y∈R M×m , where m is the dimension of the target instruction; a multilayer perceptron is used as the cooking model, which includes an input layer, L hidden layers and an output layer. For the lth layer, l=1,…,L+1,l=1 is the input layer, l=L+1 is the output layer, and the layer has h l neurons, the input of the jth neuron in the lth layer and output The calculation steps are: Enter the calculation: in, is the weight between the jth neuron in layer l and the kth neuron in layer l-1, is the bias of the jth neuron in the lth layer, is the output of the kth neuron in the l-1th layer; Output calculation: f is the activation function, which is the ReLU function: f(x) = max(0,x), expressed in matrix form as: z (l) =W (l) a (l-1) +b (l) a (l) =f(z (l) )in, is the weight matrix of the lth layer, is the bias vector of the lth layer, a (0) =x is the input feature vector; The cooking model is constructed by performing forward propagation, back propagation and parameter updating.

5. The human-machine interactive cooking system based on multi-dimensional data monitoring according to claim 3 is characterized in that: After the model training is completed, for the new real-time collected data xnew, the predicted cooking instructions are obtained through forward propagation calculation Initialization: a (0) =x new , for layers l=1 to L+1, proceed in sequence: z (l) =W (l) a (l-1) +b (l) a (l) =f(z (l) ) Final prediction: The next cooking operation instructions are predicted based on real-time collected data. Through continuous iteration of the above-mentioned forward propagation, backpropagation and parameter update processes, the model gradually learns the patterns and rules in the cooking data, so that it can accurately predict cooking instructions based on the input real-time data.

6. The human-machine interactive cooking system based on multi-dimensional data monitoring according to claim 1, characterized in that: The data monitoring module is used to continuously monitor the data in the cooking process in real time. In terms of temperature monitoring, a temperature sensor is used and installed at the designated position required by the human-computer interactive cooking system. The temperature sensor can directly sense the heat transfer between the pot and the food, measure the real-time temperature of the pot and the temperature of the food in real time; For the monitoring of dish color data, the data monitoring module uses image recognition technology to shoot the state of the dish in the pot based on the camera. The camera captures the image of the dish in real time and transmits the image to the image analysis processor. The image analysis processor extracts and analyzes the color information in the image based on the deep learning algorithm, identifies the color change stage of the dish during the cooking process, and judges the cooking progress of the dish by comparing it with the preset color database; Cooking operation time monitoring is achieved through the built-in time counter. When the user starts the cooking operation, the time counter is automatically started to record the cooking operation duration. Each cooking operation process must be completed within the time specified by the system at that time. After completing data collection, the data monitoring module uploads the data to the analysis and interaction module for analysis.

7. The human-machine interactive cooking system based on multi-dimensional data monitoring according to claim 1, characterized in that: The analysis and interaction module first pre-processes the data transmitted by the data monitoring module, cleans up missing values ​​and outliers and normalizes them, then extracts key features from the temperature, color, and time data, matches the system's preset cooking model, and inputs the features to predict the cooking stage and operation time. During cooking, the module compares the actual and predicted data in real time, and dynamically adjusts the instructions if there is a deviation. After determining the operation time, prompts are generated in the form of text, voice, and visualization. The user feedbacks the results after the operation, and the module continues to evaluate and analyze to guide the user to complete the cooking process.

8. The human-machine interactive cooking system based on multi-dimensional data monitoring according to claim 7, characterized in that: The specific analysis process in the analysis interaction module is as follows: first, complete the missing values; correct the abnormal data that exceeds the normal range according to the set reasonable range to complete the data cleaning; and convert the cleaned data into a numerical range from 0 to 1 according to their respective characteristics to achieve data normalization.

9. The human-machine interactive cooking system based on multi-dimensional data monitoring according to claim 8, characterized in that: Extract key features from the data. For temperature data, calculate the speed of temperature change at adjacent time points and record the highest and lowest temperatures in the entire process. For color data, calculate the average color value and the degree of color discreteness and extract the main color features. For time data, calculate the duration of each cooking operation interval and the total duration of continuous cooking operations. Integrate these together and input the extracted features into the model. The model will then predict the current cooking stage and the operation time.

10. The human-machine interactive cooking system based on multi-dimensional data monitoring according to claim 1, characterized in that: The interactive module includes device screen interaction and voice interaction. The smart device screen displays text steps, operation diagrams and teaching videos, while voice interaction allows users to query cooking steps and remaining time through voice commands. The interactive content includes preparation and cooking stages. When preparing, the list of ingredients, dosage, selection and processing techniques are displayed; when cooking, the progress is updated in real time based on the cooking model and real-time data. After receiving the information, the user follows the prompts and provides manual feedback after completion. The interactive module analyzes the feedback and pushes the next step of guidance if the operation meets the standards. If there is any deviation, the prompt is corrected in time to ensure that the cooking is completed smoothly until the dish is cooked.

Images