Charcoal barbecue temperature control method and system based on machine learning

Abstract

The present invention discloses a method and system for controlling the temperature of a charcoal barbecue based on machine learning, and the method includes data collection, data preprocessing, establishing a temperature feedback prediction model, model parameter optimization and real-time temperature control. The present invention belongs to the field of temperature control, and specifically refers to a method and system for controlling the temperature of a charcoal barbecue based on machine learning. The present invention designs an asymmetric mean square logarithmic error loss function to impose different penalties on the lag of the predicted value and the advance of the predicted value; by designing a monotonic regularized loss function, the stability and reliability of the temperature feedback prediction are improved, and the temperature control is achieved to the best state; the search strategy is dynamically adjusted in the exploration stage, and the error ratio and scaling factor are introduced; a refined search is performed in the development stage; and an offset optimization strategy is used to finally optimize the candidate positions, thereby improving the accuracy of the temperature control.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature control, and in particular to a charcoal barbecue temperature control method and system based on machine learning. Background Art

[0002] The temperature control method based on machine learning is a method that uses machine learning technology to optimize and automate the temperature control process. Compared with the traditional temperature control method, this method improves the accuracy and efficiency of temperature regulation through data-driven intelligent algorithms. However, the general temperature control method cannot adapt to actual needs, resulting in insufficient temperature control accuracy and large temperature control fluctuations leading to poor control effects; the general temperature control method has the problem that model parameter optimization is not suitable for dealing with nonlinear problems in temperature feedback and cannot deal with dynamic changes in temperature control, resulting in poor temperature control effects. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a charcoal barbecue temperature control method and system based on machine learning. In view of the fact that the general temperature control method cannot adapt to the actual needs of temperature control in the charcoal barbecue process, resulting in insufficient temperature control accuracy and poor control effect due to large temperature control volatility, this scheme designs an asymmetric mean square logarithmic error loss function, imposes different penalties on the predicted value lag and the predicted value advance, and can effectively handle the nonlinear changes of temperature feedback, so that the loss function is more sensitive to the temperature feedback prediction error; by designing a monotonic regularized loss function, unnecessary fluctuations in the predicted value are prevented, and the stability of the temperature feedback prediction is improved. Stability and reliability are enhanced to achieve the best temperature control state; the general temperature control method has the problem that model parameter optimization is not suitable for dealing with nonlinear problems in temperature feedback and cannot deal with dynamic changes in temperature control, resulting in poor temperature control effect. This solution dynamically adjusts the search strategy in the exploration stage, introduces error ratio and scaling factor, and helps to make precise adjustments in parameter optimization; conducts refined search in the development stage to ensure the precision of the search; adopts the offset optimization strategy to perform final optimization on the candidate positions, uses weighted average and covariance matrix to ensure the accuracy and robustness of the model parameters, thereby improving the accuracy of the temperature feedback prediction model and thus improving the accuracy of temperature control.

[0004] The technical solution adopted by the present invention is as follows: The charcoal barbecue temperature control method based on machine learning provided by the present invention comprises the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data preprocessing;

[0007] Step S3: Establishing a temperature feedback prediction model;

[0008] Step S4: model parameter optimization;

[0009] Step S5: Real-time temperature control.

[0010] Furthermore, in step S1, the data collection is to collect historical processing temperature data; the historical processing temperature data includes the temperature inside the oven, the ambient temperature, humidity, meat thickness, grilling time and meat quality feedback level; the higher the meat quality feedback level, the higher the evaluation satisfaction, and the meat quality feedback level is used as a data label.

[0011] Furthermore, in step S2, the data preprocessing is to perform data cleaning, data conversion and data set division on the collected data; the data cleaning is to process missing values, duplicate values ​​and outliers; the data conversion is to convert the cleaned data into vectors and perform standardization processing; the data set division is to divide the standardized data into a training set and a test set.

[0012] Furthermore, in step S3, the establishment of the temperature feedback prediction model specifically includes the following steps:

[0013] Step S31: Model architecture design; the established temperature feedback prediction model includes: an input layer, which is used to receive the data obtained after preprocessing; an embedding layer, which embeds the input features; an LSTM layer, in which the LSTM layer is in a stacked form, the first LSTM layer is used to process the input sequence and output a hidden state sequence, and the second LSTM layer is used to further refine the data pattern; a fully connected layer, which is used to pass the output of the LSTM layer to the output layer; the output layer is used to output the meat quality feedback level predicted by the temperature feedback prediction model;

[0014] Step S32: design an asymmetric mean square logarithmic error loss function; introduce an asymmetric penalty mechanism based on the mean square logarithmic error loss function, and give different penalty weights when the predicted value is delayed and the predicted value is advanced; the formula used is as follows:

[0015] ;

[0016] Where, ME is the asymmetric mean square logarithmic error loss function; T1 is the total number of samples; and are the actual label and predicted label at time step t1 respectively;

[0017] Step S33: Design a monotonic regularization loss function; use the regularization mechanism to ensure the monotonicity of the output prediction and prevent excessive fluctuations; the formula used is as follows:

[0018] ;

[0019] ;

[0020] Where L1 is the monotonic regularization loss function; MAX(·) is the maximum value operation; is the difference between the predicted values ​​at adjacent time steps; , , and are the predicted labels for the 1st, 2nd, 3rd and t1-1 time steps respectively;

[0021] Step S34: Design a total loss function; the formula used is as follows:

[0022] ;

[0023] Where L is the total loss function; α and β are loss weights;

[0024] Step S35: model determination; training the temperature feedback prediction model based on the training set; when the loss of the temperature feedback prediction model on the training set converges, the training of the temperature feedback prediction model is completed; an accuracy threshold is set in advance, and when the prediction accuracy of the trained temperature feedback prediction model on the test set is higher than the accuracy threshold, the temperature feedback prediction model is established; otherwise, the data set is re-divided to optimize the model parameters, and then the temperature feedback prediction model is retrained.

[0025] Furthermore, in step S4, the model parameter optimization specifically includes the following steps:

[0026] Step S41: Initialization stage; based on the loss weight, initial weight and bias of the temperature feedback prediction model, the search space is resumed; the position of the search individual in the population is randomly initialized; the prediction accuracy of the temperature feedback prediction model trained based on the individual position on the test set is used as the individual fitness value;

[0027] Step S42: Exploration phase; providing a dynamically adjusted search process to explore globally; the formula used is as follows:

[0028] ;

[0029] ;

[0030] ;

[0031] ;

[0032] In the formula, and They are the candidate positions of the i-th individual at the t+1-th iteration and the t-th iteration in the j-th dimension; is the optimal solution at the tth iteration of the jth dimension; is the error ratio; r1, r2 and r3 are random numbers between 0 and 1, independent of each other; and is the candidate position of the jth dimension of different random individuals; T is the maximum number of iterations; It is an evolutionary factor; is the scaling factor; is the smoothing term; n is the total number of dimensions; and They are the upper and lower limits of the search in the j dimension respectively; is the individual fitness value; is the final position of the i-th individual in the j-th dimension at the t-th iteration;

[0033] Step S43: Development phase; further optimize the model parameters through a more refined search; the formula used is as follows:

[0034] ;

[0035] In the formula, r4 is a random number between 0 and 1;

[0036] Step S44: offset optimization strategy; through the overall offset optimization strategy, the candidate positions are finally optimized through weighted average and offset optimization; the formula used is as follows:

[0037] ;

[0038] ;

[0039] ;

[0040] ;

[0041] In the formula, is the final position of the i-th individual in the j-th dimension at the t+1-th iteration; is the weighted average position of the dominant population; yj is a random variable generated from a normal distribution with mean 0 and covariance matrix Cov(i); NP is the number of candidate positions; is the weight coefficient in the dominant population;

[0042] Step S45: search judgment; a fitness threshold is set in advance; when there is an individual fitness value higher than the fitness threshold, a temperature feedback prediction model is established based on the individual position; if the maximum number of iterations is reached, return to step S41; otherwise, return to step S42.

[0043] Furthermore, in step S5, the real-time temperature control is based on the established temperature feedback prediction model, and the temperature inside the oven, ambient temperature, humidity, meat thickness and barbecue time are collected in real time; after preprocessing, they are input into the temperature feedback prediction model, and the temperature inside the oven corresponding to the highest meat quality feedback level predicted by the model is used as the final controlled temperature output.

[0044] The charcoal barbecue temperature control system based on machine learning provided by the present invention includes a data acquisition module, a data preprocessing module, a temperature feedback prediction model establishment module, a model parameter optimization module and a real-time temperature control module;

[0045] The data acquisition module collects historical processing temperature data and sends the data to the data preprocessing module;

[0046] The data preprocessing module performs data cleaning, data conversion and data set division on the collected data; and sends the data to the temperature feedback prediction model establishment module;

[0047] The temperature feedback prediction model establishment module designs an asymmetric mean square logarithmic error loss function to impose different penalties on the predicted value lag and the predicted value advance; designs a monotonic regularization loss function to prevent unnecessary fluctuations in the predicted value; obtains the total loss function, and then completes the establishment of the temperature feedback prediction model; and sends the data to the model parameter optimization module;

[0048] The model parameter optimization module dynamically adjusts the search strategy in the exploration phase, introduces error ratios and scaling factors; performs refined search in the development phase; uses an offset optimization strategy to perform final optimization on candidate positions; and then completes the optimization of the temperature feedback prediction model; and sends data to the real-time temperature control module;

[0049] The real-time temperature control module realizes real-time temperature control based on the established temperature feedback prediction model.

[0050] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0051] (1) In view of the fact that general temperature control methods cannot adapt to actual needs, resulting in insufficient temperature control accuracy and poor control effect due to large temperature control fluctuations, this scheme designs an asymmetric mean square logarithmic error loss function to impose different penalties on the predicted value lag and predicted value advance, which can effectively handle the nonlinear changes of temperature feedback and make the loss function more sensitive to the temperature feedback prediction error. By designing a monotonic regularized loss function, unnecessary fluctuations in the predicted value are prevented, the stability and reliability of the temperature feedback prediction are improved, and the temperature control is achieved to achieve the best state.

[0052] (2) In view of the fact that the general temperature control method has the problem that the model parameter optimization is not suitable for dealing with the nonlinear problem in temperature feedback and cannot handle the dynamic changes in temperature control, resulting in poor temperature control effect, this scheme dynamically adjusts the search strategy in the exploration stage, introduces error ratio and scaling factor, and helps to make precise adjustments in parameter optimization; conducts refined search in the development stage to ensure the precision of the search; adopts the offset optimization strategy to perform final optimization on the candidate positions, uses weighted average and covariance matrix to ensure the accuracy and robustness of the model parameters, thereby improving the accuracy of the temperature feedback prediction model and thus improving the accuracy of temperature control. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A schematic diagram of a process flow of a charcoal barbecue temperature control method based on machine learning provided by the present invention;

[0054] Figure 2 A schematic diagram of a charcoal barbecue temperature control system based on machine learning provided by the present invention;

[0055] Figure 3 is a schematic flow chart of step S3;

[0056] Figure 4 It is a schematic diagram of the process of step S4.

[0057] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0059] In the description of the present invention, it is necessary to understand that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred system or element must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0060] Example 1, see Figure 1 The present invention provides a method for controlling the temperature of a barbecue based on machine learning, the method comprising the following steps:

[0061] Step S1: data collection, collecting historical processing temperature data;

[0062] Step S2: Data preprocessing, data cleaning, data conversion and data set division of the collected data;

[0063] Step S3: Establish a temperature feedback prediction model, and impose different penalties on the prediction value lag and prediction value advance by designing an asymmetric mean square logarithmic error loss function; design a monotonic regularization loss function; obtain the total loss function, and then complete the establishment of the temperature feedback prediction model;

[0064] Step S4: Model parameter optimization, dynamically adjust the search strategy in the exploration phase, introduce error ratio and scaling factor; perform refined search in the development phase; use offset optimization strategy to perform final optimization on candidate positions; and then complete the optimization of the temperature feedback prediction model;

[0065] Step S5: Real-time temperature control, which is based on the established temperature feedback prediction model to achieve real-time temperature control.

[0066] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the historical processing temperature data includes the temperature inside the oven, the ambient temperature, the humidity, the thickness of the meat, the roasting time and the meat quality feedback level. The higher the meat quality feedback level, the higher the evaluation satisfaction, and the meat quality feedback level is used as a data label.

[0067] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, data cleaning is to process missing values, duplicate values ​​and outliers; data conversion is to convert the cleaned data into vectors and perform standardization; data set partitioning is to divide the standardized data into training sets and test sets.

[0068] Example 4, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S3, establishing a temperature feedback prediction model specifically includes the following steps:

[0069] Step S31: Model architecture design; the established temperature feedback prediction model includes: an input layer, which is used to receive the data obtained after preprocessing; an embedding layer, which embeds the input features; an LSTM layer, in which the LSTM layer is in a stacked form, the first LSTM layer is used to process the input sequence and output a hidden state sequence, and the second LSTM layer is used to further refine the data pattern; a fully connected layer, which is used to pass the output of the LSTM layer to the output layer; the output layer is used to output the meat quality feedback level predicted by the temperature feedback prediction model;

[0070] Step S32: design an asymmetric mean square logarithmic error loss function; introduce an asymmetric penalty mechanism based on the mean square logarithmic error loss function, and give different penalty weights when the predicted value is delayed and the predicted value is advanced; the formula used is as follows:

[0071] ;

[0072] Where, ME is the asymmetric mean square logarithmic error loss function; T1 is the total number of samples; and are the actual label and predicted label at time step t1 respectively;

[0073] Step S33: Design a monotonic regularization loss function; use the regularization mechanism to ensure the monotonicity of the output prediction and prevent excessive fluctuations; the formula used is as follows:

[0074] ;

[0075] ;

[0076] Where L1 is the monotonic regularization loss function; MAX(·) is the maximum value operation; is the difference between the predicted values ​​at adjacent time steps; , , and are the predicted labels for the 1st, 2nd, 3rd and t1-1 time steps respectively;

[0077] Step S34: Design a total loss function; the formula used is as follows:

[0078] ;

[0079] Where L is the total loss function; α and β are loss weights;

[0080] Step S35: model determination; training the temperature feedback prediction model based on the training set; when the loss of the temperature feedback prediction model on the training set converges, the training of the temperature feedback prediction model is completed; an accuracy threshold is set in advance, and when the prediction accuracy of the trained temperature feedback prediction model on the test set is higher than the accuracy threshold, the temperature feedback prediction model is established; otherwise, the data set is re-divided to optimize the model parameters, and then the temperature feedback prediction model is retrained.

[0081] By performing the above operations, the general temperature control method cannot adapt to actual needs, resulting in insufficient temperature control accuracy, and the large temperature control volatility leads to poor control effect. This scheme designs an asymmetric mean square logarithmic error loss function to impose different penalties on the predicted value lag and the predicted value advance, which can effectively handle the nonlinear changes of temperature feedback and make the loss function more sensitive to the temperature feedback prediction error. By designing a monotonic regularized loss function, unnecessary fluctuations in the predicted value are prevented, the stability and reliability of the temperature feedback prediction are improved, and the temperature control is achieved in the best state.

[0082] Example 5, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S4, the model parameter optimization specifically includes the following steps:

[0083] Step S41: Initialization stage; based on the loss weight, initial weight and bias of the temperature feedback prediction model, the search space is resumed; the position of the search individual in the population is randomly initialized; the prediction accuracy of the temperature feedback prediction model trained based on the individual position on the test set is used as the individual fitness value;

[0084] Step S42: Exploration phase; providing a dynamically adjusted search process to explore globally; the formula used is as follows:

[0085] ;

[0086] ;

[0087] ;

[0088] ;

[0089] In the formula, and They are the candidate positions of the i-th individual at the t+1-th iteration and the t-th iteration in the j-th dimension; is the optimal solution at the tth iteration of the jth dimension; is the error ratio; r1, r2 and r3 are random numbers between 0 and 1, independent of each other; and is the candidate position of the jth dimension of different random individuals; T is the maximum number of iterations; It is an evolutionary factor; is the scaling factor; is the smoothing term; n is the total number of dimensions; and They are the upper and lower limits of the search in the j dimension respectively; is the individual fitness value; is the final position of the i-th individual in the j-th dimension at the t-th iteration;

[0090] Step S43: Development phase; further optimize the model parameters through a more refined search; the formula used is as follows:

[0091] ;

[0092] In the formula, r4 is a random number between 0 and 1;

[0093] Step S44: offset optimization strategy; through the overall offset optimization strategy, the candidate positions are finally optimized through weighted average and offset optimization; the formula used is as follows:

[0094] ;

[0095] ;

[0096] ;

[0097] ;

[0098] In the formula, is the final position of the i-th individual in the j-th dimension at the t+1-th iteration; is the weighted average position of the dominant population; yj is a random variable generated from a normal distribution with mean 0 and covariance matrix Cov(i); NP is the number of candidate positions; is the weight coefficient in the dominant population;

[0099] Step S45: search judgment; a fitness threshold is set in advance; when there is an individual fitness value higher than the fitness threshold, a temperature feedback prediction model is established based on the individual position; if the maximum number of iterations is reached, return to step S41; otherwise, return to step S42.

[0100] By performing the above operations, in view of the fact that the general temperature control method has the problem that the model parameter optimization is not suitable for dealing with the nonlinear problem in temperature feedback and cannot deal with the dynamic changes in temperature control, resulting in poor temperature control effect, this solution dynamically adjusts the search strategy in the exploration stage, introduces error ratio and scaling factor, and helps to make precise adjustments in parameter optimization; conducts refined search in the development stage to ensure the precision of the search; adopts the offset optimization strategy to perform final optimization on the candidate positions, uses the weighted average and covariance matrix to ensure the accuracy and robustness of the model parameters, thereby improving the accuracy of the temperature feedback prediction model, and thus improving the accuracy of temperature control.

[0101] Example 6, see Figure 1This embodiment is based on the above embodiment. In step S5, real-time temperature control is based on the established temperature feedback prediction model, and real-time collection of oven temperature, ambient temperature, humidity, meat thickness and barbecue time is carried out; after preprocessing, the temperature is input into the temperature feedback prediction model, and the oven temperature corresponding to the highest meat quality feedback level predicted by the model is used as the final controlled temperature output.

[0102] Embodiment 7, see Figure 2 , this embodiment is based on the above embodiment, and the charcoal barbecue temperature control system based on machine learning provided by the present invention includes a data acquisition module, a data preprocessing module, a temperature feedback prediction model establishment module, a model parameter optimization module and a real-time temperature control module;

[0103] The data acquisition module collects historical processing temperature data and sends the data to the data preprocessing module;

[0104] The data preprocessing module performs data cleaning, data conversion and data set division on the collected data; and sends the data to the temperature feedback prediction model establishment module;

[0105] The temperature feedback prediction model establishment module designs an asymmetric mean square logarithmic error loss function to impose different penalties on the predicted value lag and the predicted value advance; designs a monotonic regularization loss function to prevent unnecessary fluctuations in the predicted value; obtains the total loss function, and then completes the establishment of the temperature feedback prediction model; and sends the data to the model parameter optimization module;

[0106] The model parameter optimization module dynamically adjusts the search strategy in the exploration phase, introduces error ratios and scaling factors; performs refined search in the development phase; uses an offset optimization strategy to perform final optimization on candidate positions; and then completes the optimization of the temperature feedback prediction model; and sends data to the real-time temperature control module;

[0107] The real-time temperature control module realizes real-time temperature control based on the established temperature feedback prediction model.

[0108] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0109] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the invention.

[0110] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A charcoal barbecue temperature control method based on machine learning, characterized in that: The method comprises the following steps: Step S1: data collection, collecting historical processing temperature data; Step S2: Data preprocessing, data cleaning, data conversion and data set division of the collected data; Step S3: Establish a temperature feedback prediction model, and impose different penalties on the prediction value lag and prediction value advance by designing an asymmetric mean square logarithmic error loss function; design a monotonic regularization loss function; obtain the total loss function, and then complete the establishment of the temperature feedback prediction model; Step S4: Model parameter optimization, dynamically adjust the search strategy in the exploration phase, introduce error ratio and scaling factor; perform refined search in the development phase; use offset optimization strategy to perform final optimization on candidate positions; and then complete the optimization of the temperature feedback prediction model; Step S5: Real-time temperature control, realizing real-time temperature control based on the established temperature feedback prediction model; In step S3, the establishment of the temperature feedback prediction model specifically includes the following steps: Step S31: Model architecture design; the established temperature feedback prediction model includes: an input layer, which is used to receive the data obtained after preprocessing; an embedding layer, which embeds the input features; an LSTM layer, in which the LSTM layer is in a stacked form, the first LSTM layer is used to process the input sequence and output a hidden state sequence, and the second LSTM layer is used to further refine the data pattern; a fully connected layer, which is used to pass the output of the LSTM layer to the output layer; the output layer is used to output the meat quality feedback level predicted by the temperature feedback prediction model; Step S32: design an asymmetric mean square logarithmic error loss function; introduce an asymmetric penalty mechanism based on the mean square logarithmic error loss function, and give different penalty weights when the predicted value is delayed and the predicted value is advanced; the formula used is as follows: ； Where, ME is the asymmetric mean square logarithmic error loss function; T1 is the total number of samples; and are the actual label and predicted label at time step t1 respectively; Step S33: Design a monotonic regularization loss function; use the regularization mechanism to ensure the monotonicity of the output prediction and prevent excessive fluctuations; the formula used is as follows: ； ； Where L1 is the monotonic regularization loss function; MAX(·) is the maximum value operation; is the difference between the predicted values ​​at adjacent time steps; , , and are the predicted labels for the 1st, 2nd, 3rd and t1-1 time steps respectively; Step S34: Design a total loss function; the formula used is as follows: ； Where L is the total loss function; α and β are loss weights; Step S35: model determination; training the temperature feedback prediction model based on the training set; when the loss of the temperature feedback prediction model on the training set converges, the training of the temperature feedback prediction model is completed; an accuracy threshold is set in advance, and when the prediction accuracy of the trained temperature feedback prediction model on the test set is higher than the accuracy threshold, the temperature feedback prediction model is established; otherwise, the data set is re-divided to optimize the model parameters, and then the temperature feedback prediction model is retrained.

2. The method for controlling temperature of barbecue based on machine learning according to claim 1, characterized in that: In step S4, the model parameter optimization specifically includes the following steps: Step S41: Initialization stage; based on the loss weight, initial weight and bias of the temperature feedback prediction model, the search space is resumed; the position of the search individual in the population is randomly initialized; the prediction accuracy of the temperature feedback prediction model trained based on the individual position on the test set is used as the individual fitness value; Step S42: Exploration phase; providing a dynamically adjusted search process to explore globally; the formula used is as follows: ； ； ； ； In the formula, and They are the candidate positions of the i-th individual at the t+1-th iteration and the t-th iteration in the j-th dimension; is the optimal solution at the tth iteration of the jth dimension; is the error ratio; r1, r2 and r3 are random numbers between 0 and 1, independent of each other; and is the candidate position of the jth dimension of different random individuals; T is the maximum number of iterations; It is an evolutionary factor; is the scaling factor; is the smoothing term; n is the total number of dimensions; and They are the upper and lower limits of the search in the j dimension respectively; is the individual fitness value; is the final position of the i-th individual in the j-th dimension at the t-th iteration; Step S43: Development phase; further optimize the model parameters through a more refined search; the formula used is as follows: ； In the formula, r4 is a random number between 0 and 1; Step S44: offset optimization strategy; through the overall offset optimization strategy, the candidate positions are finally optimized through weighted average and offset optimization; the formula used is as follows: ； ； ； ； In the formula, is the final position of the i-th individual in the j-th dimension at the t+1-th iteration; is the weighted average position of the dominant population; yj is a random variable generated from a normal distribution with mean 0 and covariance matrix Cov(i); NP is the number of candidate positions; is the weight coefficient in the dominant population; Step S45: search judgment; a fitness threshold is set in advance; when there is an individual fitness value higher than the fitness threshold, a temperature feedback prediction model is established based on the individual position; if the maximum number of iterations is reached, return to step S41; otherwise, return to step S42.

3. The method for controlling temperature of barbecue based on machine learning according to claim 1, characterized in that: In step S1, the historical processing temperature data includes the temperature inside the oven, the ambient temperature, humidity, meat thickness, roasting time and meat quality feedback level; the higher the meat quality feedback level, the higher the evaluation satisfaction, and the meat quality feedback level is used as a data label.

4. The method for controlling temperature of barbecue based on machine learning according to claim 1, characterized in that: In step S5, the real-time temperature control is based on the established temperature feedback prediction model, and real-time acquisition of the temperature inside the oven, the ambient temperature, the humidity, the thickness of the meat and the roasting time; After preprocessing, it is input into the temperature feedback prediction model, and the oven temperature corresponding to the highest meat quality feedback level predicted by the model is used as the final controlled temperature output.

5. The method for controlling temperature of barbecue based on machine learning according to claim 1, characterized in that: In step S2, the data cleaning is to process missing values, duplicate values ​​and outliers; the data conversion is to convert the cleaned data into vectors and perform standardization; the data set partitioning is to divide the standardized data into training sets and test sets.

6. A charcoal barbecue temperature control system based on machine learning, used to implement the charcoal barbecue temperature control method based on machine learning as described in any one of claims 1 to 5, characterized in that: It includes data acquisition module, data preprocessing module, temperature feedback prediction model building module, model parameter optimization module and real-time temperature control module; The data acquisition module collects historical processing temperature data and sends the data to the data preprocessing module; The data preprocessing module performs data cleaning, data conversion and data set division on the collected data; and sends the data to the temperature feedback prediction model establishment module; The temperature feedback prediction model establishment module designs an asymmetric mean square logarithmic error loss function to impose different penalties on the predicted value lag and the predicted value advance; designs a monotonic regularization loss function to prevent unnecessary fluctuations in the predicted value; obtains the total loss function, and then completes the establishment of the temperature feedback prediction model; and sends the data to the model parameter optimization module; The model parameter optimization module dynamically adjusts the search strategy during the exploration phase, introducing error ratios and scaling factors; Conduct refined search in the development phase; use offset optimization strategy to optimize candidate locations; and then complete the optimization of the temperature feedback prediction model; and send data to the real-time temperature control module; The real-time temperature control module realizes real-time temperature control based on the established temperature feedback prediction model.

Images