A temperature control method for making pastries based on artificial intelligence

Abstract

This invention relates to the field of intelligent vending equipment technology, and discloses an artificial intelligence-based temperature control method for pastry making, comprising: Step 1, collecting real-time disturbance data and historical operating data of an unmanned pastry vending terminal including a heating module; Step 2, based on the historical operating data, predicting the estimated purchase volume within a future time window using an artificial intelligence model; Step 3, setting a basic target temperature, and dynamically calculating a comprehensive temperature compensation amount by combining the real-time disturbance data and the estimated purchase volume. By using an artificial intelligence model to predict the estimated purchase volume and combining it with a pastry quality index to implement a differentiated control strategy for ensuring supply during peak periods and maintaining quality during off-peak periods, this method achieves the technical effect of ensuring the core temperature meets the standard during peak periods and maintaining the taste of pastries during off-peak periods. Compared with the existing technology that uses a fixed heating and heat preservation strategy, this method solves the shortcomings of undercooked pastries during peak periods and poor taste during off-peak periods when customer flow fluctuates.

Description

Technical Field

[0001] This invention relates to the field of intelligent vending equipment technology, specifically to a temperature control method for pastry making based on artificial intelligence. Background Technology

[0002] In recent years, unmanned pastry vending terminals, as a convenient form of food retail, have been widely deployed in subways, office buildings, schools, and other locations, providing consumers with instant-heated pastries such as buns and steamed bread. To ensure food safety and taste, the temperature inside the heating chamber must be strictly controlled.

[0003] However, existing unmanned pastry vending terminals have many shortcomings in temperature control:

[0004] Existing unmanned noodle vending terminals use fixed heating and heat preservation strategies, which cannot intelligently adapt to the dynamic fluctuations of actual customer flow. This results in insufficient temperature recovery of the cavity due to continuous food taking during peak sales periods. If fixed-time heating is continued, food safety hazards may occur due to the center temperature of the noodles not meeting the standards. During off-peak sales periods, maintaining high temperature for a long time will cause the noodles to lose moisture, deteriorate in taste, and decline in quality.

[0005] During daily operation, the frequent opening and closing of the door when customers pick up their food at the vending terminal introduces cold air from the outside, causing significant temperature fluctuations in the heating chamber inside the equipment. Existing temperature control logic is usually based on fixed sensing strategies, which are not easy to sensitively detect and quickly compensate for the instantaneous heat loss caused by opening and closing the door, thus disrupting the stability of the insulation environment.

[0006] Smart food cabinets typically store a variety of pastry products. Different products have different heating rates, required steam volumes, and optimal heat preservation temperatures. However, existing equipment offers a limited range of heating control modes and cannot provide personalized heating and heat preservation parameters for different batches and types of pastries inside the cabinet. Therefore, it cannot guarantee the best flavor and taste for all products sold at the same time.

[0007] The existing vending terminal's operation backend lacks the ability to monitor the actual internal temperature of the equipment in real time, and the operators cannot remotely grasp the temperature curves and quality status of each batch of pastries inside the cavity. As a result, when the equipment is underheated or the product tastes poor due to prolonged heat preservation, the management is not likely to proactively discover and intervene in a timely manner. Instead, they rely on passive consumer complaints or offline manual inspections to troubleshoot the faults, resulting in a lagging operation and maintenance management model.

[0008] To address the aforementioned issues, this invention proposes an artificial intelligence-based temperature control method for pastry production. This method introduces an AI model to predict and estimate purchase quantities, dynamically calculates comprehensive temperature compensation based on real-time disturbance data such as door opening and ambient temperature differences, and implements differentiated heating strategies according to the quality index of each batch of pastry products. This achieves a comprehensive control effect, ensuring temperature compliance during peak periods, maintaining pastry texture during off-peak periods, proactively compensating for environmental disturbances, and actively sending early warnings to the backend system. Summary of the Invention

[0009] To address the shortcomings of existing technologies, this invention provides an artificial intelligence-based temperature control method for pastry making, thereby solving the problems mentioned in the background section.

[0010] To achieve the above objectives, the present invention provides the following technical solution: a temperature control method for pastry making based on artificial intelligence, comprising:

[0011] Step 1: Collect real-time disturbance data and historical operational data of the unmanned pastry vending terminal, which includes a heating module;

[0012] Step 2: Based on the historical operational data, predict the estimated purchase volume within the future time window using an artificial intelligence model;

[0013] Step 3: Set the basic target temperature, and dynamically calculate the comprehensive temperature compensation amount by combining the real-time disturbance data and the estimated purchase quantity.

[0014] Step four: Based on the basic target temperature and the comprehensive temperature compensation amount, generate the final dynamic control target temperature, and control the heating module to perform temperature adjustment.

[0015] Preferably, the real-time disturbance data includes at least the cumulative number of times the door is opened within the time window, and the ambient temperature of the external environment where the device is located;

[0016] The historical operational data, including historical sales curves, weather factors, and terminal deployment location types, is used to train the artificial intelligence model.

[0017] Preferably, the calculation of the comprehensive temperature compensation includes: a door disturbance compensation item for compensating for heat loss caused by door opening; and a demand forecast compensation item for adjusting heat demand in advance based on the estimated purchase amount.

[0018] Preferably, the door disturbance compensation item is calculated based on the cumulative number of door openings and a preset door compensation coefficient;

[0019] The calculation of the comprehensive temperature compensation amount, taking into account the temperature difference between the external ambient temperature and the current temperature of the cavity, is used to compensate for continuous heat loss.

[0020] Preferably, the temperature control method for pastry making includes: collecting the heat preservation time of each batch of pastry products in the cavity; and calculating the quality index of the batch of pastry products based on the heat preservation time and the optimal serving time of the product.

[0021] When it is determined that the quality index of a batch of pastry products has declined, an alert message is proactively sent to the operations backend to prompt adjustments to the replenishment strategy.

[0022] Preferably, the execution of the dynamic control of the target temperature includes:

[0023] When the estimated purchase volume is higher than the preset threshold and the surface quality index of the cavity is generally high, the target temperature is dynamically controlled for high-temperature rapid heating to ensure that the center temperature meets the standard during peak periods.

[0024] When the estimated purchase volume is lower than the preset threshold and the quality index of a specific batch of pastry products is detected to have decreased, the dynamic control target temperature should be appropriately lowered to slow down the loss of moisture from the pastry.

[0025] Preferably, step one further includes:

[0026] Sub-step Temperature sensors are installed on the inner wall of the heating chamber and the outer shell of the unmanned pasta vending terminal, and a sampling frequency is set. Real-time reading of instantaneous temperature data inside the heating chamber Instantaneous temperature data of the external environment Furthermore, the moving average filtering algorithm is used to filter samples during the sampling period. Multiple instantaneous temperature data collected internally are processed using the formula:

[0027] The current temperature of the smoothed internal heating chamber of the device is calculated. At the same time, using the formula: Calculate the ambient temperature of the device. ,

[0028] in, For the sampling sequence index, For the first The instantaneous temperature inside the cavity during the second sampling. For the first The instantaneous external temperature of the sampled cavity;

[0029] Sub-step Based on the Hall sensor or limit switch installed at the door of the unmanned pasta vending terminal, the closing status of the door is monitored in real time. ,set up This indicates that the door is open. This indicates that the door is closed within the current control time window. Inside, using the formula:

[0030] Calculate the cumulative number of times the door is opened within the time window. ,

[0031] in, This represents the total number of status monitoring operations within the time window. This is a jump detection function;

[0032] when and When, the function value is Conversely, the function value is ;

[0033] The cumulative number of times With current temperature and external ambient temperature Together they constitute real-time disturbance data;

[0034] Sub-step The batch identification of each batch of pastry products placed inside the heating chamber is identified by an RFID device or a vision recognition module installed inside the chamber. Record the batch identifier. Insertion timestamp detected entering the cavity Combined with the current system time Using the formula:

[0035] The duration of heat preservation for this batch of pastry products was calculated. Iterate through all batches currently on sale within the cavity to generate a product status dataset containing the insulation time of each batch:

[0036] ,

[0037] in, This represents the total number of batches of products currently deposited inside the cavity;

[0038] Sub-step The cloud server retrieves historical operating data from the unmanned pastry vending terminal, which includes historical sales volume sequences. Weather characteristic values and deployment location type coefficient Using the maximum-minimum normalization algorithm formula:

[0039] The historical operational data is preprocessed to generate standardized feature vectors. ,in, The original data values, It is the minimum value in the historical data sample. The standardized feature vector is the maximum value in the historical data sample. Used as input for subsequent artificial intelligence models, and combined with real-time perturbation data and product status datasets. Together they form a multi-dimensional state dataset.

[0040] Preferably, step two further includes:

[0041] Sub-step Obtain the historical operational data, which includes standardized feature vectors. Corresponding historical actual purchase volume tags A time series prediction model or regression model is constructed as the artificial intelligence model, using the standardized feature vector. Compared with the historical actual purchase volume tag The artificial intelligence model is trained offline, and the training process aims to minimize the root mean square error. For the objective, the root mean square error Calculated using the formula:

[0042] ,

[0043] in, The total number of training samples, For the first The actual purchase volume of the sample For artificial intelligence models to the first The predicted purchase volume of the sample is used to obtain the finally trained prediction model. ;

[0044] Sub-step Upon entering the aforementioned future time window Prior to this, data was collected within the aforementioned future time window. The corresponding prediction input features, wherein the prediction input features contain at least the current time period information. Imminent weather characteristic values Coefficient of the terminal deployment location type The predicted input features are processed using a normalization algorithm into the standardized feature vector. Current feature vectors with the same dimensions ;

[0045] Sub-step The current feature vector Input to the trained prediction model Through the prediction model The inference calculation output is for the future time window. scalar prediction And set a minimum safety stock level. With maximum production capacity inventory Boundary condition constraint functions are used:

[0046] For the scalar predicted value The calibration is performed to ultimately generate the estimated purchase volume within the future time window. The estimated purchase quantity It will be used for calculating the overall temperature compensation.

[0047] Preferably, step three further includes:

[0048] Sub-step Retrieve the preset base target temperature And obtain the cumulative number of times the event was activated from the real-time disturbance data. The external ambient temperature of the equipment and the current temperature of the heating chamber inside the equipment. Using the physical disturbance compensation formula:

[0049] Calculate the temperature compensation component caused by the physical environment. ;

[0050] in, The heat loss coefficient of the door is used to characterize the temperature drop that needs to be compensated for during a single door opening. The environmental temperature difference conduction coefficient is used to characterize the temperature value that needs to be compensated for due to the heat conduction caused by the temperature difference between the inside and outside.

[0051] The temperature compensation component Used to quantify heat loss caused by physical environmental factors;

[0052] Sub-step Obtain the estimated purchase quantity It also provides real-time statistics on the actual sales volume within the current control period. Using the demand forecasting compensation formula:

[0053] Calculate the demand-driven temperature compensation component ;

[0054] in, This is the passenger flow response gain coefficient, used to adjust the sensitivity to passenger flow fluctuations. The maximum load capacity of the unmanned pastry vending terminal is used to normalize the sales difference; the temperature compensation component is... Used to store heat in advance before peak passenger flow arrives;

[0055] Sub-step The temperature compensation component With the temperature compensation component Perform a weighted summation using the comprehensive calculation formula:

[0056] The comprehensive temperature compensation amount is obtained. And using the target generation formula: Calculate the target temperature for dynamic control Furthermore, a safety boundary determination logic is introduced; if the calculated dynamic control target temperature is... Exceeding the preset maximum safe temperature Then, the dynamically controlled target temperature is forced to be... Equal to the maximum safe temperature The final output is the dynamically controlled target temperature. To the heating module controller.

[0057] Preferably, step four further includes:

[0058] Sub-step The heat preservation time of each batch of pastry products collected inside the cavity was obtained. And retrieve the preset optimal tasting time for this type of product. The mass decay formula is used for the maximum heat preservation tolerance time tmax(i):

[0059] Calculate the quality index of this batch of pastry products. ;

[0060] in, For batch indexing, the quality index The range of values ​​is arrive Generate a quality index dataset by iterating through all batches: And calculate the quality index dataset. Average quality index in With minimum quality index ;

[0061] Sub-step Obtain the estimated purchase quantity And set peak passenger flow thresholds Quality judgment threshold Simultaneously, the average quality index is obtained. With the minimum quality index Peak condition determination logic is adopted:

[0062] Determine whether peak supply guarantee strategy is triggered. ;

[0063] Using the trough condition judgment logic:

[0064] Determine whether the low-end shelf-life strategy has been triggered. When the trough condition determination logic is mentioned When true, the low-peak quality assurance strategy is triggered. At the same time, the system automatically generates and sends a warning message to the operations backend to prompt adjustments to the replenishment strategy;

[0065] Sub-step Obtain the dynamic control target temperature With the aforementioned base target temperature ;

[0066] If the peak supply guarantee strategy is determined If true, set the final execution temperature. ;

[0067] Low-peak shelf-life strategy If true, set the final execution temperature. ,

[0068] in, Minimum temperature for food safety; in the peak supply strategy With the aforementioned low-peak shelf-life strategy If none of the above are true, set the final execution temperature. Equal to the base target temperature ;

[0069] The heating module employs a proportional-integral-derivative method. The control algorithm is based on the final execution temperature. With the current temperature of the cavity Deviation: Calculate control output quantity The control output quantity Calculated using the formula:

[0070] ,

[0071] in, This is the proportionality coefficient. The integral coefficient is... The differential coefficient is the control output quantity. Used to adjust the heating power of the heating module.

[0072] This invention provides a temperature control method for pastry making based on artificial intelligence. It has the following beneficial effects:

[0073] 1. This invention uses an artificial intelligence model to predict and estimate purchase volume, and combines it with the pastry quality index to implement a differentiated control strategy for ensuring supply during peak periods and maintaining quality during off-peak periods. This achieves the technical effect of ensuring the center temperature meets the standard during peak periods and ensuring the taste of pastries during off-peak periods. Compared with the existing technology that uses a fixed heating and heat preservation strategy, this invention solves the shortcomings of the pastries being easily undercooked during peak periods and having poor taste during off-peak periods when customer flow fluctuates.

[0074] 2. This invention uses real-time monitoring of the number of times the door is opened and the ambient temperature difference to dynamically calculate the comprehensive temperature compensation amount, thereby achieving the technical effect of actively compensating for instantaneous heat loss and maintaining a constant cavity temperature. Compared with the existing technology that uses a fixed sensing strategy for control, this invention solves the shortcomings of frequent door opening and closing causing large temperature fluctuations and damage to the constantness of the insulation environment.

[0075] 3. This invention tracks the heat preservation time of each batch of pastries in the cavity and calculates the quality index independently, achieving the technical effect of personalized temperature control management for different types of pastries stored together. Compared with the limited types of heating control modes provided in the existing technology, it solves the shortcomings of not being able to take into account the heating needs of different products and not being able to guarantee the best flavor and taste of all products.

[0076] 4. This invention proactively sends warning messages to the operations backend when the pastry quality index is judged to be declining, thereby achieving the technical effect of changing passive management to proactive intervention and real-time monitoring of equipment and product status. Compared with the existing technology that relies on passive consumer complaints or offline manual inspections, this invention solves the shortcomings of operators' management being lagging behind and not being able to proactively detect and intervene in faults in a timely manner. Attached Figure Description

[0077] Figure 1 This is a flowchart illustrating the overall process of a temperature control method for pastry making based on artificial intelligence, according to the present invention.

[0078] Figure 2 This is a detailed flowchart of the dynamic calculation of the comprehensive temperature compensation in step three of this invention;

[0079] Figure 3 This is a detailed flowchart of the implementation of the differentiated control strategy in step four of this invention. Detailed Implementation

[0080] To enable those skilled in the art to understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0081] The present invention will now be described in detail with reference to the accompanying drawings:

[0082] Example:

[0083] Please see the appendix Figure 1 - Appendix Figure 3 This invention provides an artificial intelligence-based temperature control method for pastry making, comprising:

[0084] Step 1: Collect real-time disturbance data and historical operational data of the unmanned pastry vending terminal, which includes a heating module;

[0085] Step two: Based on historical operational data, use artificial intelligence models to predict the estimated purchase volume within future time windows;

[0086] Step 3: Set the basic target temperature, and dynamically calculate the comprehensive temperature compensation amount by combining real-time disturbance data and estimated purchase volume;

[0087] Step four: Based on the basic target temperature and the comprehensive temperature compensation, generate the final dynamic control target temperature, and control the heating module to perform temperature adjustment.

[0088] Step one further includes:

[0089] Sub-step Temperature sensors are installed on the inner wall of the heating chamber and the outer shell of the unmanned pasta vending terminal, and a sampling frequency is set. Real-time reading of instantaneous temperature data inside the heating chamber Instantaneous temperature data of the external environment Furthermore, the moving average filtering algorithm is used to filter samples during the sampling period. Multiple instantaneous temperature data collected internally are processed using the formula:

[0090] The current temperature of the smoothed internal heating chamber of the device is calculated. At the same time, using the formula: Calculate the ambient temperature of the device. ,

[0091] in, For the sampling sequence index, For the first The instantaneous temperature inside the cavity during the second sampling. For the first The instantaneous external temperature of the sampled cavity;

[0092] Sub-step Based on Hall effect sensors or limit switches installed at the door of the unmanned pasta vending terminal, the closing status of the door is monitored in real time. ,set up This indicates that the door is open. This indicates that the door is closed within the current control time window. Inside, using the formula:

[0093] Calculate the cumulative number of times the door is opened within the time window. ,

[0094] in, This represents the total number of status monitoring operations within the time window. This is a jump detection function;

[0095] when and When, the function value is Conversely, the function value is ;

[0096] Total number of times opened With current temperature and external ambient temperature Together they constitute real-time disturbance data;

[0097] Sub-step The batch identification of each batch of pastry products placed inside the heating chamber is identified by an RFID device or a vision recognition module installed inside the chamber. Record the batch identifier. Insertion timestamp detected entering the cavity Combined with the current system time Using the formula:

[0098] The duration of heat preservation for this batch of pastry products was calculated. Iterate through all batches currently on sale within the cavity to generate a product status dataset containing the insulation time of each batch:

[0099] ,

[0100] in, This represents the total number of batches of products currently deposited inside the cavity;

[0101] Sub-step The cloud server retrieves historical operational data from the unmanned pastry vending terminals, which includes historical sales volume sequences. Weather characteristic values and deployment location type coefficient Using the maximum-minimum normalization algorithm formula:

[0102] Historical operational data is preprocessed to generate standardized feature vectors. ,in, The original data values, It is the minimum value in the historical data sample. The standardized feature vector is the maximum value in the historical data sample. Used as input for subsequent artificial intelligence models, and combined with real-time perturbation data and product status datasets. Together they form a multi-dimensional state dataset.

[0103] Step two further includes:

[0104] Sub-step Obtain historical operational data, which includes standardized feature vectors. Corresponding historical actual purchase volume tags Construct time series prediction or regression models as artificial intelligence models, using standardized feature vectors. Labels based on historical actual purchase volume The artificial intelligence model is trained offline, with the training process aiming to minimize the root mean square error. For the objective, root mean square error Calculated using the formula:

[0105] ,

[0106] in, The total number of training samples, For the first The actual purchase volume of the sample For artificial intelligence models to the first The predicted purchase volume of the sample is used to obtain the finally trained prediction model. ;

[0107] Sub-step Entering the future time window Current and future time windows The corresponding predicted input features must contain at least the information of the current time period. Imminent weather characteristic values Coefficient of Terminal Deployment Location Type The normalization algorithm is used to process the predicted input features into standardized feature vectors. Current feature vectors with the same dimensions ;

[0108] Sub-step , will the current feature vector Input to the trained prediction model Through predictive models The inference calculation output is for future time windows. scalar prediction And set a minimum safety stock level. With maximum production capacity inventory Boundary condition constraint functions are used:

[0109] scalar prediction value After calibration, the estimated purchase volume for the future time window is finally generated. Estimated purchase volume It will be used for calculating the overall temperature compensation.

[0110] Step three further includes:

[0111] Sub-step Retrieve the preset base target temperature And obtain the cumulative number of times it is activated from the real-time disturbance data. The external ambient temperature of the equipment and the current temperature of the heating chamber inside the equipment. Using the physical disturbance compensation formula:

[0112] Calculate the temperature compensation component caused by the physical environment. ;

[0113] in, The heat loss coefficient of the door is used to characterize the temperature drop that needs to be compensated for during a single door opening. The environmental temperature difference conduction coefficient is used to characterize the temperature value that needs to be compensated for due to the heat conduction caused by the temperature difference between the inside and outside.

[0114] Temperature compensation component Used to quantify heat loss caused by physical environmental factors;

[0115] Sub-step Get the estimated purchase volume It also provides real-time statistics on the actual sales volume within the current control period. Using the demand forecasting compensation formula:

[0116] Calculate the demand-driven temperature compensation component ;

[0117] in, This is the passenger flow response gain coefficient, used to adjust the sensitivity to passenger flow fluctuations. The maximum load capacity of the unmanned pastry vending terminal is used to normalize the sales difference and includes temperature compensation. Used to store heat in advance before peak passenger flow arrives;

[0118] Sub-step Temperature compensation component Temperature compensation component Perform a weighted summation using the comprehensive calculation formula:

[0119] The comprehensive temperature compensation amount is obtained. And using the target generation formula: Calculate the target temperature for dynamic control Furthermore, a safety boundary determination logic is introduced; if the calculated dynamic control target temperature is... Exceeding the preset maximum safe temperature Then, the target temperature will be dynamically controlled. Equal to the maximum safe temperature The final output is a dynamically controlled target temperature. To the heating module controller.

[0120] Step four further includes:

[0121] Sub-step The heat preservation time of each batch of pastry products collected inside the cavity was obtained. And retrieve the preset optimal tasting time for this type of product. The mass decay formula is used for the maximum heat preservation tolerance time tmax(i):

[0122] Calculate the quality index of this batch of pastry products. ;

[0123] in, For batch indexing, quality index The range of values ​​is arrive Generate a quality index dataset by iterating through all batches: And calculate the quality index dataset Average quality index in With minimum quality index ;

[0124] Sub-step Get the estimated purchase volume And set peak passenger flow thresholds Quality judgment threshold Simultaneously obtain the average quality index. With minimum quality index Peak condition determination logic is adopted:

[0125] Determine whether peak supply guarantee strategy is triggered. ;

[0126] Using the trough condition judgment logic:

[0127] Determine whether the low-end shelf-life strategy has been triggered. ; Logic for determining trough conditions When true, the low-end quality assurance strategy is triggered. At the same time, the system automatically generates and sends a warning message to the operations backend to prompt adjustments to the replenishment strategy;

[0128] Sub-step To obtain the target temperature for dynamic control Compared with the baseline target temperature ;

[0129] If the peak supply guarantee strategy is determined If true, set the final execution temperature. ;

[0130] Low-peak shelf-life strategy If true, set the final execution temperature. ,

[0131] in, Minimum temperature for food safety; peak supply strategy Low-end quality preservation strategy If none of the above are true, set the final execution temperature. equal to the base target temperature ;

[0132] The heating module uses a proportional-integral-derivative method. The control algorithm is based on the final execution temperature. With the current temperature of the cavity Deviation: Calculate control output quantity Control the output amount Calculated using the formula:

[0133] ,

[0134] in, This is the proportionality coefficient. The integral coefficient is... The differential coefficient controls the output quantity. Used to adjust the heating power of the heating module.

[0135] By comprehensively collecting real-time temperature of the heating cavity, ambient temperature, number of times the door is opened, batch heat preservation time of pastries, and historical operational data, input data is provided for subsequent passenger flow forecasting, compensation calculation, and quality grading.

[0136] By using historical data to train artificial intelligence models, intelligent predictions of future passenger flow demand can be achieved, enabling the system to transform from a passive response to proactive pre-adjustment, providing a basis for ensuring supply during peak periods and maintaining quality during off-peak periods.

[0137] By setting a basic target temperature and combining physical disturbance compensation with demand forecast compensation to dynamically calculate the comprehensive temperature compensation amount, quantitative compensation for disturbances caused by factors such as door opening, internal and external temperature difference, and passenger flow fluctuations is achieved, generating a scientific and dynamic control target.

[0138] By calculating the quality index of pastries and combining it with the estimated purchase volume, we can implement differentiated peak supply and off-peak quality assurance strategies, and proactively issue warnings when the quality declines. Ultimately, this achieves closed-loop control in ensuring food safety, improving the taste of pastries, and optimizing operational efficiency.

[0139] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A temperature control method for pastry making based on artificial intelligence, characterized in that, include: Step 1: Collect real-time disturbance data and historical operational data of the unmanned pastry vending terminal, which includes a heating module; Step two: Based on the historical operational data, use an artificial intelligence model to predict the estimated purchase volume within the future time window. ; Step 3: Set the basic target temperature, and dynamically calculate the comprehensive temperature compensation amount by combining the real-time disturbance data and the estimated purchase quantity. Step three further includes: Sub-step Retrieve the preset base target temperature And obtain the cumulative number of times it is activated from the real-time disturbance data. The external ambient temperature of the equipment and the current temperature of the heating chamber inside the equipment. Using the physical disturbance compensation formula: Calculate the temperature compensation component caused by the physical environment. ; in, The heat loss coefficient of the door is used to characterize the temperature drop that needs to be compensated for during a single door opening. The environmental temperature difference conduction coefficient is used to characterize the temperature value that needs to be compensated for due to the heat conduction caused by the temperature difference between the inside and outside. The temperature compensation component Used to quantify heat loss caused by physical environmental factors; Sub-step Obtain the estimated purchase quantity It also provides real-time statistics on the actual sales volume within the current control period. Using the demand forecasting compensation formula: Calculate the demand-driven temperature compensation component ; in, This is the passenger flow response gain coefficient, used to adjust the sensitivity to passenger flow fluctuations. The maximum load capacity of the unmanned pastry vending terminal is used to normalize the sales difference; the temperature compensation component is... Used to store heat in advance before peak passenger flow arrives; Sub-step The temperature compensation component With the temperature compensation component Perform a weighted summation using the comprehensive calculation formula: The comprehensive temperature compensation amount is obtained. And using the target generation formula: Calculate the target temperature for dynamic control Furthermore, a safety boundary determination logic is introduced; if the calculated dynamic control target temperature is... Exceeding the preset maximum safe temperature Then, the dynamically controlled target temperature is forced to be... Equal to the maximum safe temperature The final output is the dynamically controlled target temperature. To the heating module controller; Step 4: Control the heating module to perform temperature regulation.

2. The temperature control method for pastry making based on artificial intelligence according to claim 1, characterized in that, The real-time disturbance data includes at least the cumulative number of times the door was opened within the time window, and the ambient temperature of the external environment where the device is located. The historical operational data, including historical sales curves, weather factors, and terminal deployment location types, is used to train the artificial intelligence model.

3. The temperature control method for pastry making based on artificial intelligence according to claim 1, characterized in that, The calculation of the comprehensive temperature compensation includes: a door disturbance compensation item for compensating for heat loss caused by door opening; and a demand forecast compensation item for adjusting heat in advance based on the estimated purchase amount.

4. The temperature control method for pastry making based on artificial intelligence according to claim 3, characterized in that, The door disturbance compensation item is calculated based on the cumulative number of door openings and the preset door compensation coefficient. The calculation of the comprehensive temperature compensation amount combines the temperature difference between the external ambient temperature and the current temperature of the cavity to compensate for continuous heat loss.

5. The temperature control method for pastry making based on artificial intelligence according to claim 1, characterized in that, The temperature control method for pastry making includes: collecting the heat preservation time of each batch of pastry products in the cavity; and calculating the quality index of the batch of pastry products based on the heat preservation time and the optimal serving time of the product. When it is determined that the quality index of a batch of pastry products has declined, an alert message is proactively sent to the operations backend to prompt adjustments to the replenishment strategy.

6. The temperature control method for pastry making based on artificial intelligence according to claim 4, characterized in that, The execution of the dynamic control of the target temperature includes: When the estimated purchase volume is higher than the preset threshold and the surface quality index of the cavity is generally high, the target temperature is dynamically controlled for high-temperature rapid heating to ensure that the center temperature meets the standard during peak periods. When the estimated purchase volume is lower than the preset threshold and the quality index of a specific batch of pastry products is detected to have decreased, the dynamic control target temperature should be appropriately lowered to slow down the loss of moisture from the pastry.

7. The temperature control method for pastry making based on artificial intelligence according to claim 1, characterized in that, Step one further includes: Sub-step Temperature sensors are installed on the inner wall of the heating chamber and the outer shell of the unmanned pasta vending terminal, and a sampling frequency is set. Real-time reading of instantaneous temperature data inside the heating chamber Instantaneous temperature data of the external environment Furthermore, the moving average filtering algorithm is used to filter samples during the sampling period. Multiple instantaneous temperature data collected internally are processed using the formula: The current temperature of the smoothed internal heating chamber of the device is calculated. At the same time, using the formula: Calculate the ambient temperature of the device. , in, For the sampling sequence index, For the first The instantaneous temperature inside the cavity during the second sampling. For the first The instantaneous external temperature of the sampled cavity; Sub-step Based on the Hall sensor or limit switch installed at the door of the unmanned pasta vending terminal, the closing status of the door is monitored in real time. ,set up This indicates that the door is open. This indicates that the door is closed within the current control time window. Inside, using the formula: Calculate the cumulative number of times the door is opened within the time window. , in, This represents the total number of status monitoring operations within the time window. This is a jump detection function; when and When, the function value is Conversely, the function value is ; The cumulative number of times it was opened With current temperature and external ambient temperature Together they constitute real-time disturbance data; Sub-step The batch identification of each batch of pastry products placed inside the heating chamber is identified by an RFID device or a vision recognition module installed inside the chamber. Record the batch identifier. Insertion timestamp detected upon entering the cavity Combined with the current system time Using the formula: The duration of heat preservation for this batch of pastry products was calculated. Iterate through all batches currently on sale within the cavity to generate a product status dataset containing the insulation time for each batch: ， in, This represents the total number of batches of products currently deposited inside the cavity; Sub-step The cloud server retrieves historical operating data from the unmanned pastry vending terminal, which includes historical sales volume sequences. Weather characteristic values and deployment location type coefficient Using the maximum-minimum normalization algorithm formula: The historical operational data is preprocessed to generate standardized feature vectors. ,in, The original data values, It is the minimum value in the historical data sample. The standardized feature vector is the maximum value in the historical data sample. Used as input for subsequent artificial intelligence models, and combined with real-time perturbation data and product status datasets. Together they form a multi-dimensional state dataset.

8. The temperature control method for pastry making based on artificial intelligence according to claim 1, characterized in that, Step two further includes: Sub-step Obtain the historical operational data, which includes standardized feature vectors. Corresponding historical actual purchase volume tags A time series prediction model or regression model is constructed as the artificial intelligence model, using the standardized feature vector. Compared with the historical actual purchase volume tag The artificial intelligence model is trained offline, with the training process aimed at minimizing the root mean square error. For the objective, the root mean square error Calculated using the formula: ， in, The total number of training samples, For the first The actual purchase volume of the sample For artificial intelligence models to the first The predicted purchase volume of the sample is used to obtain the finally trained prediction model. ; Sub-step Upon entering the aforementioned future time window Prior to this, data was collected within the aforementioned future time window. The corresponding prediction input features, wherein the prediction input features contain at least the current time period information. Imminent weather characteristic values Coefficient of Terminal Deployment Location Type The predicted input features are processed using a normalization algorithm into the standardized feature vector. Current feature vectors with the same dimensions ; Sub-step The current feature vector Input to the trained prediction model Through the prediction model The inference calculation output is for the future time window. scalar prediction And set a minimum safety stock level. With maximum production capacity inventory Boundary condition constraint functions are used: For the scalar predicted value The calibration is performed to ultimately generate the estimated purchase volume within the future time window. The estimated purchase quantity It will be used for calculating the overall temperature compensation.

9. The temperature control method for pastry making based on artificial intelligence according to claim 1, characterized in that, Step four further includes: Sub-step The heat preservation time of each batch of pastry products collected inside the cavity was obtained. And retrieve the preset optimal tasting time for this product category. With maximum heat retention time The quality decay formula is used: Calculate the quality index of this batch of pastry products. ; in, For batch indexing, the quality index The range of values ​​is arrive Generate a quality index dataset by iterating through all batches: And calculate the quality index dataset. Average quality index With minimum quality index ; Sub-step Obtain the estimated purchase quantity And set peak passenger flow thresholds Quality judgment threshold Simultaneously, the average quality index is obtained. With the minimum quality index Peak condition determination logic is adopted: Determine whether peak supply guarantee strategy is triggered. ; Using the trough condition judgment logic: Determine whether the low-peak shelf-life strategy has been triggered. When the trough condition determination logic is mentioned When true, the low-peak quality assurance strategy is triggered. At the same time, the system automatically generates and sends a warning message to the operations backend to prompt adjustments to the replenishment strategy; Sub-step Obtain the dynamic control target temperature With the aforementioned base target temperature ; If the peak supply guarantee strategy is determined If true, set the final execution temperature. ; Low-peak shelf-life strategy If true, set the final execution temperature. , in, Minimum temperature for food safety; in the peak supply strategy With the aforementioned low-peak shelf-life strategy If none of the above are true, set the final execution temperature. Equal to the base target temperature ; The heating module adopts a proportional-integral-derivative method. The control algorithm is based on the final execution temperature. With the current temperature of the cavity Deviation: Calculate control output quantity The control output quantity Calculated using the formula: ， in, This is the proportionality coefficient. The integral coefficient is... The differential coefficient is the control output quantity. Used to adjust the heating power of the heating module.

Images