Central intelligent kitchen remote collaborative management method based on Internet of Things and edge computing

Abstract

The invention provides a central intelligent kitchen remote collaborative management method based on Internet of Things and edge computing, and relates to the technical field of intelligent kitchen management. The method comprises the steps of collecting task and resource states through each edge intelligent node, constructing a task classification and abstract data set, and generating a cross-kitchen task allocation, energy consumption optimization and food safety control scheme through a cloud management center in combination with a historical database. Each node realizes self-diagnosis and parameter sharing based on self-adaptive parameters, consistency verification is completed by using a lightweight distributed consensus mechanism, and a cloud side generates a visual remote management interface and a cooperative control result set according to the consistency verification, so that intelligent optimization and safe and efficient operation of a kitchen group level are realized. According to the invention, the problem that cross-kitchen intelligent cooperation and global efficient regulation and control are difficult to realize due to multi-node data inconsistency and cooperation delay of an existing central kitchen management system is solved.

Description

Technical Field

[0001] This invention relates to the field of smart kitchen management technology, and in particular to a remote collaborative management method for a central smart kitchen based on the Internet of Things and edge computing. Background Technology

[0002] With the rapid development of restaurant chains and standardized food production, central kitchens have gradually become key nodes in food processing, cold chain storage, and food distribution. Existing central kitchen systems generally employ IoT technology to monitor temperature, humidity, energy consumption, and food status, relying on cloud servers for centralized data processing and decision-making. However, these systems still primarily operate on a "central server—terminal device" model, with edge nodes only handling data collection and uploading, lacking autonomous analysis and local decision-making capabilities. As the system scales and the number of kitchen nodes increases, the cloud's load increases, and response delays become more pronounced, failing to meet the high real-time and stability requirements of the processing site.

[0003] With the rapid development of edge computing, artificial intelligence, and digital twin models, a trend towards local decision-making and hierarchical collaboration at edge nodes is emerging in the field of kitchen management. Theoretically, edge nodes can autonomously complete local control, task allocation, and data optimization on-site in the kitchen, thereby achieving a low-latency response mechanism. However, in existing architectures, edge nodes are mostly isolated structures, lacking cross-kitchen context information sharing and consistency maintenance mechanisms. This leads to frequent data asynchrony, task conflicts, or policy overlap when multiple kitchens simultaneously execute tasks or share resources. This situation not only limits system collaboration efficiency but also affects the globally optimal control capability of the entire central kitchen network.

[0004] Furthermore, existing central kitchen systems typically rely on cloud-based decision-making for unified management of all kitchens. If the network is interrupted or cloud feedback is delayed, some nodes cannot continue operating, resulting in weak system fault tolerance and self-recovery capabilities. Simultaneously, the data distribution and computation processes lack intelligent self-evolution mechanisms, preventing individual kitchen nodes from autonomously learning or dynamically optimizing their task execution strategies. Therefore, current central kitchen management systems generally suffer from problems such as inconsistent data across multiple nodes, collaboration delays, and insufficient autonomy. Summary of the Invention

[0005] In order to overcome the shortcomings of the existing technology, the purpose of this invention is to provide a remote collaborative management method for a central smart kitchen based on the Internet of Things and edge computing. This invention solves the problem that the existing central kitchen management system is difficult to achieve intelligent collaboration and efficient global control across kitchens due to data inconsistency and collaboration delays among multiple nodes.

[0006] To achieve the above objectives, the present invention provides the following solution: A method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing includes: Deploy edge intelligent nodes with local computing and wireless communication capabilities in multiple kitchen scenarios, and use each edge intelligent node to collect environmental parameters, equipment status, energy consumption data and cold chain temperature data in real time during the kitchen operation through connected IoT sensing devices to obtain the raw operation dataset; Extract key operational metrics features from the original operational dataset to obtain a set of operational feature parameters; A digital twin model of the kitchen is constructed based on the set of operational feature parameters; Based on the kitchen digital twin model and the edge intelligent node, the actual task requirements and resource usage status are analyzed in real time to obtain a task classification result set and a task summary dataset. The cloud management center receives the task classification result set and task summary dataset, and performs comprehensive analysis and optimization decision-making on the task classification result set and task summary dataset in conjunction with the historical operation database, generating cross-kitchen task allocation schemes, energy consumption optimization schemes and food safety control schemes, and obtaining the set of optimization control strategies distributed by the cloud. Each edge intelligent node performs self-diagnosis based on the optimized control strategy set and the task operation status information updated in real time by the kitchen digital twin model, and obtains the diagnosis results. If the diagnosis results are performance deviation, task abnormality and equipment imbalance, the model parameters and control commands are automatically adjusted, and the optimized parameters are shared among adjacent edge intelligent nodes to obtain an adaptive parameter dataset. Each edge intelligent node uses the adaptive parameter dataset to perform consistency verification on task status, resource allocation and control logic through a lightweight distributed consensus mechanism, and obtains a multi-node consistent context dataset. Based on the multi-node consistency context dataset, the cloud management center generates a unified visual remote management interface and obtains a visual collaborative control result set.

[0007] The present invention discloses the following technical effects: This invention provides a remote collaborative management method for a central smart kitchen based on the Internet of Things (IoT) and edge computing. The method includes: deploying edge smart nodes with local computing and wireless communication capabilities in multiple kitchen scenarios; utilizing each edge smart node to collect environmental parameters, equipment status, energy consumption data, and cold chain temperature data in real time during kitchen operation via connected IoT sensing devices to obtain a raw operation dataset; extracting key operational indicator features from the raw operation dataset to obtain an operation feature parameter set; constructing a digital twin model of the kitchen based on the operation feature parameter set; performing real-time analysis of actual task requirements and resource usage status based on the kitchen digital twin model and the edge smart nodes to obtain a task classification result set and a task summary dataset; and receiving the task classification result set and task summary dataset in a cloud management center, and combining them with a historical operation database to refine the task classification result set and task summary dataset. The dataset undergoes comprehensive analysis and optimization decisions to generate cross-kitchen task allocation schemes, energy consumption optimization schemes, and food safety control schemes, resulting in an optimized control strategy set distributed from the cloud. Each edge intelligent node performs self-diagnosis based on the optimized control strategy set and the real-time updated task operation status information of the kitchen digital twin model, obtaining diagnostic results. If the diagnostic results indicate performance deviation, task anomaly, or equipment imbalance, the model parameters and control commands are automatically adjusted, and the optimized parameters are shared among adjacent edge intelligent nodes, resulting in an adaptive parameter dataset. Each edge intelligent node uses the adaptive parameter dataset to perform consistency verification on task status, resource allocation, and control logic through a lightweight distributed consensus mechanism, obtaining a multi-node consistent context dataset. The cloud management center generates a unified visual remote management interface based on the multi-node consistent context dataset, obtaining a visual collaborative control result set. This invention deploys edge intelligent nodes with computing and communication capabilities in various kitchen scenarios to achieve real-time collection of environmental parameters, equipment status, and energy consumption information; combined with the digital twin model, it dynamically maps and predicts the operation status, enabling rapid identification of performance deviations and task anomalies; the cloud management center optimizes cross-kitchen tasks and energy consumption based on multi-node feedback, achieving dual closed-loop control of global decision-making and local adaptation. This method effectively improves the data interoperability and task collaboration among multiple kitchens, reduces energy consumption and latency, enhances the system's real-time response and intelligent optimization capabilities, and significantly improves the technical defects of information lag and resource waste in traditional centralized management methods. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 A flowchart illustrating a remote collaborative management method for a central smart kitchen based on the Internet of Things and edge computing, provided in this embodiment of the invention; Figure 2 A flowchart for obtaining the set of runtime characteristic parameters provided in an embodiment of the present invention. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0012] like Figure 1 As shown, this invention provides a remote collaborative management method for a central smart kitchen based on the Internet of Things and edge computing, comprising: Step 100: Deploy edge intelligent nodes with local computing and wireless communication capabilities in multiple kitchen scenarios, and use each edge intelligent node to collect environmental parameters, equipment status, energy consumption data and cold chain temperature data in real time during the kitchen operation through the connected IoT sensing devices to obtain the raw operation dataset; Step 200: Extract key operational indicator features from the original operational dataset to obtain a set of operational feature parameters; Step 300: Construct a digital twin model of the kitchen based on the set of operational feature parameters; Step 400: Based on the kitchen digital twin model and the edge intelligent node, perform real-time analysis of actual task requirements and resource usage status to obtain a task classification result set and a task summary dataset; Step 500: The cloud management center receives the task classification result set and task summary dataset, and performs comprehensive analysis and optimization decision-making on the task classification result set and task summary dataset in combination with the historical operation database, generating cross-kitchen task allocation scheme, energy consumption optimization scheme and food safety control scheme, and obtaining the set of optimization control strategies distributed by the cloud. Step 600: Each edge intelligent node performs self-diagnosis based on the optimized control strategy set and the task operation status information updated in real time by the kitchen digital twin model, and obtains the diagnosis result. If the diagnosis result is performance deviation, task abnormality and equipment imbalance, the model parameters and control commands are automatically adjusted, and the optimized parameters are shared among adjacent edge intelligent nodes to obtain an adaptive parameter dataset. Step 700: Each edge intelligent node uses the adaptive parameter dataset to perform consistency verification on task status, resource allocation and control logic through a lightweight distributed consensus mechanism to obtain a multi-node consistent context dataset. Step 800: Based on the multi-node consistency context dataset, the cloud management center generates a unified visual remote management interface and obtains a visual collaborative control result set.

[0013] Furthermore, the specific implementation process of step 100 is as follows: This embodiment deploys multiple intelligent nodes with edge computing capabilities in different work areas of the central intelligent kitchen. Each node is equipped with a local processing unit and a wireless communication module. Each edge intelligent node can independently perform data acquisition and preliminary preprocessing functions, enabling real-time monitoring of the kitchen work environment, equipment operating status, and energy consumption. Nodes form a collaborative link through a local wireless network to ensure the effectiveness of data synchronization and temporary caching, thereby achieving rapid data acquisition and status awareness without relying on cloud responses.

[0014] In this embodiment, the input end of the edge intelligent node is connected to various IoT sensing devices to sense key parameters during kitchen operation. Specifically, environmental sensors are used to detect temperature, humidity, air quality, and smoke concentration in real time; the intelligent cooking equipment interface is used to collect information on the equipment's start / stop status, heat level, and cooking time; energy consumption detection devices are used to measure the instantaneous and cumulative consumption of electricity and gas; and cold chain monitoring devices are used to track temperature change curves in cold storage, refrigeration equipment, and cold chain transportation units. These sensing devices are connected to the edge node's data acquisition interface via a standardized communication protocol to achieve synchronous sampling and data timing marking.

[0015] In this embodiment, after data acquisition, the edge intelligent node formats and denoises the obtained raw signal, removes invalid or abnormal data points, and generates a raw operational dataset for subsequent analysis based on time tags and device identifiers. This dataset contains multi-dimensional feature data information under various kitchen operating conditions, serving as the basic input data source for subsequent digital twin modeling and task hierarchical calculations, thereby ensuring the authenticity, reliability, and continuity of the kitchen operating status mapping.

[0016] Furthermore, such as Figure 2 As shown, the specific implementation process of step 200 is as follows: In this embodiment, after collecting the original operational dataset, the data is first preprocessed to ensure the accuracy and consistency of subsequent feature extraction. During preprocessing, the edge intelligent nodes automatically perform data denoising and time alignment operations, eliminate errors caused by inconsistent sampling frequencies through sliding window resampling, and perform anomaly detection and removal on various numerical signals to eliminate sensor drift points. Subsequently, the edge nodes index and group the preprocessed data according to equipment category and task, generating a structured preprocessed dataset to reflect the fundamental changes in the kitchen environment, equipment, and energy consumption information over time and in dynamic states.

[0017] Based on the preprocessed dataset, this embodiment uses feature normalization and standardization algorithms to calculate the distribution characteristics and relative rates of change of different types of data, such as temperature, energy consumption, current, voltage, task duration, and equipment load, thereby obtaining a preliminary feature vector set of multidimensional numerical features. Edge intelligent nodes utilize an embedded multidimensional correlation analysis model to process the preliminary feature vector set, extracting highly correlated indicators that characterize kitchen operating efficiency and equipment stability. This model automatically identifies candidate variables with significant coupling characteristics in multidimensional space by jointly measuring the linear and nonlinear dependencies between features through the mean difference index and covariance regularization term, generating a key feature candidate set to reflect the energy consumption and task behavior characteristics under different kitchen operating conditions.

[0018] This embodiment performs feature importance assessment on the candidate set of key features. Edge nodes automatically calculate the contribution value of each feature based on the correlation distribution of historical samples, and select the task load level, energy efficiency, equipment health, and operation time distribution rate, which have the greatest impact on kitchen operating efficiency, as the final key operating indicators. Based on this set of key operating indicator features, edge nodes further calculate comprehensive operating evaluation parameters to form a set of operating feature parameters that comprehensively reflects the overall state of kitchen operations. This result provides statistical support for the parameter input and operating status evaluation of the subsequent digital twin model, achieving high-precision mapping and dynamic prediction of kitchen operating status.

[0019] Specifically, the multidimensional correlation analysis model described in this embodiment is used to calculate the comprehensive correlation index between kitchen operation characteristics to quantify the coupling strength between various operation characteristics and kitchen status and results, thereby achieving a joint evaluation of operational efficiency and equipment status. This model uses statistical methods to calculate the data distribution, covariance, and mean difference of different characteristic dimensions, comprehensively reflecting the linear and nonlinear dependencies between characteristic quantities. In this embodiment, the preprocessed operational data is divided into subdivisions, and each type of operational characteristic (such as equipment temperature change rate, operating time, energy consumption rate, and load fluctuation) is analyzed as an input sample pair to obtain corresponding correlation values. These values ​​are then used to construct a comprehensive operational characteristic mapping matrix, providing input parameters for the subsequent digital twin model.

[0020] In the specific implementation process, this embodiment first establishes a running status sample set and a running efficiency sample set in the edge computing node, corresponding to the kitchen's status detection features and task execution results, respectively. The edge node calculates the statistical covariance value of each pair of sample sets to measure the synchronous change trend between features. To avoid bias caused by different data units, the node calculates the fluctuation amplitude and mean deviation of each type of input signal and introduces a normalization term in the calculation to balance the influence of different features. The normalization term is composed of the volatility index (i.e., standard deviation) of each feature, which can ensure the comparability between numerical orders of magnitude. In addition, to prevent the existence of zero denominators or extreme value imbalances in the data, this embodiment sets a positive smoothing constant, which is relatively small, between 0.0001 and 0.001, and is dynamically adjusted according to the accuracy of the specific sensor.

[0021] To reflect the nonlinear relationship between features as the mean difference changes, this embodiment introduces an exponential decay factor in the calculation. This factor calculates the similarity based on the mean difference of the sample features. When the means of two features are close, the exponential term has a higher value, indicating a strong correlation between the two dimensions. When the mean difference is large, the factor gradually approaches zero to weaken the influence of weakly correlated features. The final correlation result is then normalized and averaged across all feature dimensions to form a comprehensive correlation index, which measures the overall coordination between the multi-dimensional operational features of the kitchen. For example, when equipment power is stable, energy efficiency is high, and task execution time is highly synchronized with equipment temperature fluctuations, the comprehensive index value is close to 1; conversely, when multi-dimensional signal differences are significant and state fluctuations are inconsistent, the index value approaches zero. The calculation results of this model can intuitively reflect the performance synergy of the kitchen during actual operation, providing a quantifiable basis for subsequent state prediction and resource allocation.

[0022] Furthermore, the specific implementation process of step 300 is as follows: After obtaining the set of operational characteristic parameters, this embodiment first constructs the basic data framework for a digital twin model of the kitchen based on the equipment operating status, energy consumption parameters, and task execution information uploaded by each edge intelligent node. This framework consists of a physical layer, a data layer, and a model layer. The physical layer corresponds to the equipment operating status and environmental monitoring data of the real kitchen; the data layer stores the corresponding time-series operational characteristic parameters; and the model layer is used to establish the state mapping of the virtual kitchen. By dynamically synchronizing and updating various parameters through edge intelligent nodes, the virtual kitchen can maintain consistency with the real kitchen in terms of time and performance. This structure ensures that the model can reflect changes in the on-site state with low latency during operation, providing data support for subsequent prediction and control.

[0023] This embodiment defines the input variables and internal mapping relationships of a kitchen digital twin model based on four key indicators: task load level, energy efficiency, equipment health, and operation time distribution rate, using a set of operational characteristic parameters. During model building, a multi-node data fusion algorithm is used to weight and combine data from different devices and tasks, and feature smoothing and time-series differencing techniques are employed to improve the model's dynamic response accuracy. The model internally uses a nonlinear adaptive function structure to describe the interaction between task execution efficiency, energy consumption changes, and environmental disturbances, enabling the virtual kitchen to simulate actual operational characteristics under different operating scenarios. Edge nodes automatically adjust their weight parameters according to different equipment categories and operational intensity, achieving adaptive stability of the model under multiple operating conditions.

[0024] In this embodiment, during the model training and calibration phase, model parameters are updated by comparing real-time operational data with historical reference data. Edge intelligent nodes detect discrepancies between the virtual state values ​​output by the digital twin model and the actual collected operational states. When the discrepancy exceeds a set threshold, the corresponding task execution coefficients and energy balance coefficients in the model are automatically corrected to ensure the accuracy of the bidirectional mapping. After multiple rounds of updates, the model can continuously reflect the task load fluctuations, energy efficiency changes, and equipment health trends of the kitchen under different operational states, achieving a predictable and adjustable digital twin mapping effect that corresponds to the real world. This provides high-precision dynamic support for subsequent resource optimization and intelligent control.

[0025] Specifically, the kitchen digital twin model described in this embodiment constructs a virtual model that dynamically reflects the actual operating status of the kitchen by fusing and calculating multi-source operating parameters collected from edge nodes. Specifically, in the model's calculation process, this embodiment comprehensively considers factors such as the task execution status of each edge node within different time sampling periods, equipment operational stability, energy consumption performance, and ambient temperature changes. The model uses a nonlinear function structure to weight and combine these features to reflect the comprehensive operating status value of the kitchen under the influence of task load and environmental disturbances during actual operation. The model's output is used to predict the kitchen's operating trend and forms a closed-loop mapping between cloud management and edge control, achieving a synergistic effect between state prediction and energy efficiency optimization.

[0026] In this embodiment, the model's input data comes from multiple edge intelligent nodes distributed across various kitchen scenarios. Each edge node periodically collects equipment operational stability parameters, task completion coefficients, unit task energy consumption, and ambient temperature values, forming time-series data. Equipment operational stability parameters characterize equipment fluctuations over a period of time and can be obtained by monitoring equipment vibration amplitude, current stability, or the number of malfunctions. The task completion coefficient represents the current task progress, ranging from zero to one; a higher value indicates more complete task completion. Unit task energy consumption reflects the average energy consumption within a single work cycle and can be obtained from real-time power integration. Ambient temperature parameters are collected in real-time by temperature sensors to reflect the impact of the external kitchen thermal environment on the operation of internal equipment. The time sampling interval is defined as the time difference between two consecutive data samples and is used to calculate the rate of temperature change and the dynamic response trend of the task.

[0027] To ensure the model's adaptability and stability, this embodiment sets an automatically adjusted dynamic scaling factor for each feature input to balance the relative influence of task features and environmental features at different operating stages. When the kitchen workload is high and tasks change rapidly, the model automatically increases the weight of task features; when the ambient temperature fluctuates significantly, the model increases the weight of environmental features. Small constant parameters are used to avoid division by zero or over-amplification in calculations and are typically set to very small fixed values. The dynamic decay factor adjusts the model's response to historical biases; when network load or energy consumption fluctuates rapidly, this factor limits the model's oversensitivity to short-term anomalies. The cloud-based reference correlation mean is calculated from historical operating data and reflects the average feature coupling relationship of the kitchen over long-term operation. The model's global scaling constant adjusts the proportional relationship between the virtual space and the actual kitchen's operating state, ensuring that the twin model's output numerically matches the actual measurement range, thus enabling the virtual kitchen's performance to accurately correspond to the actual kitchen's working state. For example, when equipment overloads and causes energy consumption to rise along with temperature fluctuations, the model output value will increase accordingly, indicating that the kitchen's operating state deviates from the optimal energy efficiency range; when the equipment stabilizes, the model automatically decays and corrects, and the output value returns to the normal range, thus achieving real-time dynamic mapping that is consistent with reality.

[0028] More specifically, this embodiment assumes that the central kitchen is deployed with 5 edge intelligent nodes, hence the node count is 5. When calculating the virtual operating state output, the model synthesizes the input parameters of the 5 nodes according to weights to form a comprehensive twin output. The equipment operating stability parameter of each node is calculated by monitoring equipment vibration, current fluctuations, and fault frequency, with a value ranging from 0 to 1. The closer the value is to 1, the more stable the equipment. For example, the equipment operating stability of the first node can be 0.92, the second node 0.87, and the remaining nodes 0.89, 0.95, and 0.83 respectively.

[0029] The task completion coefficient reflects the task execution progress within a corresponding time period, with a value ranging from 0 to 1. A higher value indicates a higher degree of task completion. For example, the task completion coefficient for the first node could be 0.78, and for the second node, it could be 0.85. Unit task energy consumption represents the average energy consumption of an edge node when completing a standard cooking task or cold chain operation, and can be measured in kilowatt-hours or joules. Here, the examples use values ​​of 1.6, 2.1, 1.8, 1.9, and 2.0 kilowatt-hours, respectively. The ambient temperature parameter reflects the current temperature sampling results of the node, measured in degrees Celsius. The current sampling temperatures are 26°C, 24°C, 25°C, 24°C, and 26°C. The time sampling interval is defined as the time difference between two consecutive samples; in this embodiment, it is set to 30 seconds to capture the dynamic response characteristics of temperature and energy consumption.

[0030] In the model, the dynamic scaling factors for task features and environment features can be varied between 0.5 and 1, and 0.3 and 0.7, respectively, to balance the influence of the two types of features. Taking the current working condition as an example, the task scaling factor can be set to 0.85, and the environment scaling factor to 0.5. To avoid division by zero in the calculation, a small constant of 1 × 10⁻⁶ is set. -5 The mean reference correlation of the same kitchen in the cloud-based historical window is calculated based on the past 72 hours of operation history and is 0.78. The dynamic decay factor is set to 0.05, corresponding to the model's moderate sensitivity in the historical deviation recovery process. The global scaling constant is set to 1.05, serving as a proportional correction parameter between the virtual space and the actual operating state.

[0031] Based on the above specific numerical settings, when the equipment operates stably, energy consumption remains within the normal range, and the task load increases slightly, the overall twin state value output by the model is approximately between 0.8 and 0.9, indicating that the virtual kitchen is operating at high efficiency. If the energy consumption of a certain edge node increases significantly or the temperature changes drastically, the model output value will quickly drop below 0.6, reflecting the trend of the kitchen's operating state deviating from the energy efficiency equilibrium range, thereby achieving real-time synchronization and intelligent correlation between the virtual kitchen state and the actual operating process.

[0032] Furthermore, the specific implementation process of step 400 is as follows: In this embodiment, after obtaining the virtual operating status of the kitchen digital twin model, each edge intelligent node first collects task request information and equipment resource usage status in real time for the current kitchen scenario. Task request information includes job type, task urgency, expected execution time, and related equipment requirements; equipment resource usage status includes the number of currently online devices, runtime occupancy rate, available time periods, and energy consumption load level. After uploading the above information to the local data cache unit, the edge nodes compare it with the virtual operating status generated by the kitchen digital twin model. By calculating the degree of difference between task requirements and virtual equipment load, a task-resource matching dataset is generated. This process can reflect the degree of competition for resources and potential conflicts among various tasks in the kitchen in real time, providing basic input for the next step of task priority calculation.

[0033] This embodiment, based on the aforementioned task-resource matching dataset, employs a load balancing evaluation algorithm and a resource constraint analysis model to calculate the execution priority and corresponding resource occupancy ratio of each task. The load balancing evaluation algorithm dynamically adjusts the task execution order by examining the impact of each task under the current device load, ensuring that high-priority tasks receive optimal resource allocation under limited resource conditions. The resource constraint analysis model utilizes parameters such as device availability, energy reserve, and runtime window to analyze the execution cost and efficiency gains of each type of task, outputting quantified priority indicators. This embodiment introduces a dynamic weight adjustment mechanism during priority calculation, automatically reducing the priority of energy-intensive tasks when device energy consumption approaches its limit to prevent overload operation. The priority evaluation result set obtained through model calculation includes the comprehensive priority value and resource occupancy ratio of each task, used to guide subsequent classification and scheduling.

[0034] This embodiment comprehensively analyzes the parallelism of multiple tasks, energy efficiency, and equipment load in a kitchen scenario based on the task priority evaluation result set, generating a task classification result set. This result set categorizes all tasks in the kitchen into three types: high-priority tasks, standard-priority tasks, and low-priority tasks, and generates corresponding scheduling strategies for different task levels. Based on the task classification result set, edge intelligent nodes perform summary statistics on the data of completed and currently executing tasks, including runtime distribution, equipment call frequency, and energy consumption characteristics. By compressing and extracting features from the above data, a task summary dataset is formed, which is used to quickly identify the task execution characteristics and energy efficiency levels in the kitchen at different time periods, providing real-time data support for subsequent task scheduling optimization and energy consumption prediction.

[0035] Specifically, after obtaining the task-resource matching dataset, this embodiment first performs standardization to eliminate dimensional differences in resource requirements, execution latency, and load characteristics among different task types. The standardization steps include mapping the resource requirement parameters of each task to maximum and minimum values, normalizing the available device resources across intervals, and correcting the historical average execution latency using standard deviation, ensuring that the characteristic parameters of different tasks can be calculated within the same scale space. The standardized data is then used to construct a task resource constraint matrix, describing the coupling relationship between task requirements, available device resources, and execution latency. This matrix forms a dynamic constraint structure numerically, providing a computational basis for subsequent load calculation and task priority determination, and ensuring that the model maintains data consistency and comparability in the face of heterogeneous devices and multi-task parallelism.

[0036] This embodiment, based on the aforementioned task resource constraint matrix, employs a multi-objective load balancing evaluation algorithm to calculate the load occupancy rate of each edge intelligent node and the task's sensitivity to resources. In this process, each edge intelligent node calculates its load value based on its task input rate, device availability, and real-time energy consumption level. By comparing the node's task response rate within the current time window with its historical average window, the node's dynamic balance coefficient is obtained to measure its load stability. Simultaneously, a sensitivity calculation is performed on the matching degree between tasks and nodes; when a task has a high dependence on specific resources or is sensitive to time latency, its sensitivity value increases accordingly. By combining the load occupancy rate and task sensitivity of each node, a node load evaluation result set is formed. This result set can characterize the pressure each task exerts on the overall resource pool and its resource affinity to different nodes, providing a decision-making basis for subsequent calculations.

[0037] This embodiment, after obtaining the node load assessment result set, further utilizes a resource constraint analysis model to solve for the task scheduling priority sequence and resource optimization allocation ratio. The analysis model constructs a joint solution equation with the objectives of maximizing task execution efficiency and optimizing resource utilization, calculating the optimal execution order and resource allocation coefficient of tasks within the schedulable time window. High-priority tasks occupy higher weights in the solution results, allowing for priority allocation of available equipment resources during peak load periods. Subsequently, the obtained task priority calculation result set undergoes comprehensive weight normalization and stability correction; normalization eliminates priority deviations caused by different task sizes, while stability correction offsets priority jumps caused by short-term load fluctuations. After the above processing, a task priority assessment result set is obtained, which can reflect the multi-task execution order and resource allocation ratio in a kitchen scenario in real time, providing an accurate basis for dynamic task scheduling and energy consumption optimization.

[0038] Specifically, the resource constraint analysis model described in this embodiment achieves reasonable resource allocation and execution priority determination among multiple tasks in the kitchen by establishing a multi-objective optimization solution process. The model's primary objective is to minimize task execution latency, while simultaneously constraining resource utilization deviation and twin state deviation, making the model's output resource allocation strategy more consistent with the actual operating rules of a kitchen scenario. The model solution process includes three main parts: first, calculating the execution latency of each task under different resource configurations to reflect the relationship between task completion speed and allocation strategy; second, evaluating the deviation between the real-time resource utilization of the task and historical reference values ​​to ensure the balance and stability of resource allocation; and third, comparing the dynamic state values ​​of the task calculated in the digital twin model with the historical reference states stored in the cloud, using state similarity measurement to determine whether the task's current operation deviates from the normal range. Combining these indicators, the model ultimately outputs the optimal set of task allocation and resource utilization ratios, thereby determining the optimal scheduling scheme for different tasks under limited computing resources.

[0039] In this embodiment, the set of resource allocation strategy variables to be solved describes the resource allocation ratio of each task under parallel operation conditions and is the main solution objective of the model. The total number of tasks is the total number of tasks to be scheduled in the system during the current time period, which is detected by the edge intelligent nodes. The task execution delay function is used to represent the completion time of the task under the selected allocation strategy, which can be obtained by calculating the execution history and the current remaining load of the system. The available computing resources represent the number of processing units, CPU usage ratio, or memory capacity that can be allocated to the task in the current node or resource pool. The real-time resource utilization rate is the actual utilization rate of the task during execution, which can be obtained by the monitoring module based on time series statistics; the historical average resource utilization benchmark for similar tasks is the average utilization rate under similar operating conditions, which is used to dynamically compare whether the current utilization efficiency of the task is normal. The dynamic status value is calculated in real time by the kitchen digital twin model during task execution and is used to reflect the operational stability and energy efficiency of the task; the reference twin status is the feature value of the task in a good operating state recorded by the cloud management center, which is used to determine the degree of deviation of the actual operating state.

[0040] In this embodiment, the internal adjustment coefficients are used to balance the impact of the different constraint terms on the target optimization result. The first adjustment coefficient controls the weight of the resource utilization deviation term, and its value is typically between 0.1 and 1.0; the second adjustment coefficient controls the influence of the twin state deviation term, and its value ranges from 0.05 to 0.5. When the task resource usage fluctuates significantly, the model automatically increases the first adjustment coefficient to strengthen the resource balance constraint; when the twin state deviates significantly, the model increases the second adjustment coefficient to strengthen the state correlation correction. To avoid division by zero or excessive amplification of errors, a small constant is set in the model, with a typical value of 1×10. -5This is used to maintain numerical stability during denominator calculations. For example, within the same time window, if the real-time resource utilization rate of a task is 0.75, the historical average utilization baseline is 0.70, and the corresponding twin state relative deviation is 0.08, then the resource allocation ratio obtained by the model after solving is approximately 0.78, indicating that the task can obtain a slightly higher-than-average resource allocation under overall load balancing conditions. Through this method, the model can dynamically and adaptively achieve high-precision determination of multi-task resource optimization allocation and scheduling priorities, thereby ensuring the efficient and stable operation of the kitchen digital twin system.

[0041] Furthermore, the specific implementation process of step 500 is as follows: In this embodiment, after data aggregation at the edge intelligent nodes is completed, the cloud management center first receives the task hierarchy result set and task summary dataset uploaded by each node, and then correlates and matches them with historical records in the cloud historical operation database to establish multi-source data association relationships across kitchens. In this step, the cloud management center compares the uploaded data with historical records one-to-one based on task identifiers, timestamps, and device numbers, thereby forming a multi-kitchen historical operation comparison dataset. This dataset contains comparison results of operational characteristics of different kitchens under similar task types, resource allocation ratios, ambient temperatures, and energy consumption levels. By establishing a unified data alignment mechanism, this embodiment ensures the comparability and temporal continuity of the comparison data between different kitchens, providing a high-precision data foundation for subsequent feature extraction and pattern recognition.

[0042] This embodiment, based on the aforementioned multi-kitchen historical operation comparison dataset, employs clustering analysis and pattern recognition algorithms to extract task synergy, energy consumption characteristics, and risk factors. Clustering analysis identifies task sets with similar execution patterns and energy consumption patterns to discover task groups that can be executed in parallel or share resources. Pattern recognition algorithms extract task synergy coefficients, energy consumption sensitivity coefficients, and risk weighting factors by analyzing task execution trajectories, energy consumption fluctuation curves, and the frequency of risk event triggers. The task synergy coefficient measures the degree of synergy between tasks in different kitchens in terms of time overlap and resource sharing; the energy consumption sensitivity coefficient characterizes the response speed of energy consumption indicators to changes in task load; and the risk weighting factor represents the potential impact of tasks on food safety or equipment safety during execution. The cross-kitchen operation feature mapping set obtained through model iteration clearly reflects the interdependence and operational patterns between tasks in a multi-dimensional space, providing accurate feature input for subsequent optimization solutions.

[0043] This embodiment comprehensively solves for task allocation efficiency, energy consumption minimization, and cold chain food safety constraints based on the cross-kitchen operation feature mapping set, generating a cloud-based optimization decision result set. This embodiment uses a multi-objective joint optimization algorithm to jointly calculate the task execution order, equipment resource allocation ratio, and cold chain temperature control strategy, ensuring that cross-kitchen task balance and optimal energy consumption are achieved while meeting food safety constraints. When the model solves and finds that some kitchens have excessively high resource utilization or a sudden increase in energy consumption, the cloud management center automatically adjusts the task allocation weights for those kitchens and issues early cooling or time-sharing scheduling instructions to the cold chain temperature control nodes. Finally, based on the cloud-based optimization decision result set, a cross-kitchen task allocation scheme, energy consumption optimization scheme, and food safety control scheme are generated, and the generated unified decision instructions are sent to each edge intelligent node, forming a set of directly executable optimization control strategies to achieve coordinated operation and energy efficiency optimization among the kitchen group.

[0044] Furthermore, the specific implementation process of step 600 is as follows: In this embodiment, after the optimized control strategy set is distributed from the cloud, each edge intelligent node first receives the strategy set and then synchronously compares it with the real-time updated task operation status information in the kitchen digital twin model. The synchronous comparison process includes two parts: first, matching the task allocation instructions, energy consumption scheduling parameters, and safety control thresholds distributed from the cloud with the task execution progress, energy consumption data, and equipment operation parameters in the current twin model; second, establishing a strategy-state matching dataset to record the difference between the strategy target value and the actual feedback value. This embodiment ensures the consistency between strategy data and operation status through timestamp comparison and data hash indexing mechanisms. When there is a deviation between the strategy instructions and the task status, the deviation vector in the matching dataset is dynamically marked for subsequent performance diagnosis and adaptive correction logic calls.

[0045] This embodiment, based on the aforementioned strategy-state matching dataset, utilizes a performance evaluation algorithm to perform real-time diagnosis of the task execution process, detecting performance deviations, energy consumption anomalies, and equipment imbalances. The performance evaluation algorithm quantifies task operational health by calculating indicators such as task completion time deviation rate, equipment energy consumption exceedance rate, and inter-node load imbalance. When task execution deviates from the target range or energy consumption indicators exceed a certain threshold of historical averages, the model automatically generates a task diagnostic analysis result set. This result set includes task performance deviation categories (e.g., response latency, task interruption), energy consumption anomaly categories (e.g., continuous high energy consumption, unstable energy consumption fluctuations), and equipment load imbalance levels (e.g., excessively high single-node resource utilization or excessive synchronization latency). By analyzing the result set, this embodiment can quickly identify key issues affecting system efficiency and stability during kitchen operation.

[0046] In this embodiment, after determining that the task diagnostic analysis result set represents performance deviation, task anomaly, or equipment imbalance, each edge intelligent node will automatically adjust the corresponding parameters and control commands in the kitchen digital twin model according to the deviation type. The adjustment process includes three parts: First, the task execution time coefficient and resource allocation weight in the twin model are corrected according to the type of performance deviation; second, when energy consumption is abnormal, the node automatically adjusts the dynamic scaling coefficient and energy efficiency balance threshold of the energy consumption prediction module; third, for equipment imbalance, the node optimizes the equipment allocation strategy through internal scheduling logic, so that resources are redistributed to nodes with lower loads. After completing the parameter correction, each edge intelligent node will generate a corrected model parameter set and share it among adjacent nodes. To ensure parameter synchronization consistency, this embodiment sets a parameter consistency confirmation mechanism, compares the timestamps and version numbers of the corrected data between nodes, and generates an adaptive parameter dataset only when all nodes confirm that synchronization is correct. Through this mechanism, this embodiment realizes the adaptive optimization and collaborative dynamic update of the kitchen digital twin system in a distributed environment, thereby maintaining the stability and robustness of the overall operating performance.

[0047] Furthermore, the specific implementation process of step 700 is as follows: In this embodiment, after generating the adaptive parameter dataset, each edge intelligent node receives the dataset and loads the latest model parameters and task control instructions into its local operating environment to establish a node-local parameter mapping table. This mapping table records the correspondence between various parameters and control logic within the current node, including task execution priority parameters, resource allocation ratio parameters, energy consumption threshold parameters, and device scheduling control instructions. This embodiment uses a hash index to establish unique identifiers for different parameters to ensure the accuracy of parameter matching during loading. After loading, each node compares its parameter mapping table with the previous version, marking any added, updated, or deleted parameter items to quickly identify the source of change during subsequent consensus calculation, ensuring data structure consistency and parameter content integrity across multiple nodes.

[0048] This embodiment uses a lightweight distributed consensus mechanism based on the node local parameter mapping table to calculate the parameter differences between each node in terms of task status and resource allocation, generating a node state difference dataset. The lightweight distributed consensus mechanism is a fast consensus protocol specifically designed for edge computing scenarios. Through multiple rounds of parameter verification and state hash comparison between nodes, it identifies key indicators such as task execution latency differences, resource utilization differences, and energy consumption balance deviations. During the difference calculation process, each node broadcasts its own state summary data packet to its neighboring nodes and receives summary data synchronously reported by other nodes to form a multi-node difference matrix. This embodiment calculates the parameter offset vector and covariance ratio between nodes to accurately locate the parameter items and task logic units causing inconsistencies, ultimately generating a node state difference dataset to guide subsequent dynamic balance correction steps.

[0049] This embodiment executes a node synchronization calibration process based on a node state difference dataset to dynamically balance and correct task scheduling logic and control strategies with deviations. The process first triggers a node self-calibration mechanism based on the abnormal node positions marked in the difference matrix, adjusting the relevant task state values ​​and resource allocation parameters proportionally. Second, the task scheduling instructions are recalibrated through the control logic monitoring module within each node, ensuring that affected tasks return to the target balance range in the next cycle. Finally, after all nodes have completed local corrections, a network-wide consensus confirmation process is executed, comparing the version number and hash digest of each node's correction results to verify parameter consistency. When all nodes confirm state synchronization and no conflicts, a multi-node consistency context dataset is generated. This dataset records the unified state information of each node's task state, resource allocation, and control logic after consensus adjustment, serving as the basis for subsequent distributed scheduling and coordinated control, thereby achieving unified decision-making and overall stable collaborative operation of the digital twin kitchen network.

[0050] Furthermore, the specific implementation process of step 800 is as follows: In this embodiment, after each edge intelligent node completes multi-node consistency calibration and generates a multi-node consistency context dataset, the cloud management center constructs a unified visual remote management interface based on this dataset to achieve centralized monitoring and collaborative control across kitchens. The visual interface extracts core parameters such as task status, resource allocation ratio, energy consumption level, and equipment operating status of each node in real time through a data parsing module and displays them dynamically in a graphical manner. This embodiment introduces a multi-layer data mapping mechanism in the interface design, where different kitchen nodes are identified as independent area blocks. Each area displays task load curves, equipment operating health, and energy efficiency trends in real time, enabling managers to intuitively grasp the operating status of the entire kitchen group.

[0051] During the visualization rendering process, the cloud management center hierarchically processes key indicators in the multi-node consistency context dataset: the first layer is overall operational status monitoring, used to display overall energy efficiency, task completion rate, and safety status; the second layer is a single-node status view, used to display detailed task execution details, parameter adjustment records, and device status changes within a specific edge node; the third layer is an alarm and optimization suggestion module, which automatically generates diagnostic prompts and optimization instructions when the system detects abnormal energy consumption, device load imbalance, or unexpected task delays. Through the above visualization mechanism, this embodiment achieves a fully graphical presentation of the entire process from data synchronization to status prediction.

[0052] Based on the dynamic operational feedback generated in the visual interface, and combined with the collaborative control logic in the multi-node consistency context dataset, the cloud management center outputs a visual collaborative control result set. This result set includes remote scheduling instructions, energy consumption control strategy update information, and security protection adjustment parameters, which can be directly distributed to each edge node for execution. Administrators can manually intervene in or automatically approve strategies through the interface, and the system will adjust model parameters and operational strategies in real time according to user instructions. In this way, this embodiment achieves visualized intelligent collaborative control at the kitchen group level, enabling the complex multi-node digital twin system to have dynamic management capabilities that are monitorable, diagnosable, and interventionable, ensuring that the overall kitchen operation is efficient, energy-saving, and safe and controllable.

[0053] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0054] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing, characterized in that, include: Deploy edge intelligent nodes with local computing and wireless communication capabilities in multiple kitchen scenarios, and use each edge intelligent node to collect environmental parameters, equipment status, energy consumption data and cold chain temperature data in real time during the kitchen operation through connected IoT sensing devices to obtain the raw operation dataset; Extract key operational metrics features from the original operational dataset to obtain a set of operational feature parameters; A digital twin model of the kitchen is constructed based on the set of operational feature parameters; Based on the kitchen digital twin model and the edge intelligent node, the actual task requirements and resource usage status are analyzed in real time to obtain a task classification result set and a task summary dataset. The cloud management center receives the task classification result set and task summary dataset, and performs comprehensive analysis and optimization decision-making on the task classification result set and task summary dataset in conjunction with the historical operation database, generating cross-kitchen task allocation schemes, energy consumption optimization schemes and food safety control schemes, and obtaining the set of optimization control strategies distributed by the cloud. Each edge intelligent node performs self-diagnosis based on the optimized control strategy set and the task operation status information updated in real time by the kitchen digital twin model, and obtains the diagnosis results. If the diagnosis results are performance deviation, task abnormality and equipment imbalance, the model parameters and control commands are automatically adjusted, and the optimized parameters are shared among adjacent edge intelligent nodes to obtain an adaptive parameter dataset. Each edge intelligent node uses the adaptive parameter dataset to perform consistency verification on task status, resource allocation and control logic through a lightweight distributed consensus mechanism, and obtains a multi-node consistent context dataset. Based on the multi-node consistency context dataset, the cloud management center generates a unified visual remote management interface and obtains a visual collaborative control result set.

2. The method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing as described in claim 1, characterized in that, The IoT sensing devices include: Environmental sensors, smart cooking equipment interfaces, energy consumption detection devices, and cold chain monitoring devices.

3. The method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing according to claim 1, characterized in that, The key operational indicators include the following characteristics: Task load level, energy efficiency, equipment health, and operation time distribution.

4. The method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing as described in claim 1, characterized in that, Extract key operational metrics features from the original operational dataset to obtain a set of operational feature parameters, including: The original running dataset is preprocessed to obtain a preprocessed dataset; The preprocessed dataset is processed using feature normalization and standardization algorithms to calculate the distribution characteristics and relative rates of change of various data types, resulting in a preliminary feature vector set. The preliminary feature vector set is extracted using a multidimensional correlation analysis model to extract descriptive indicators characterizing kitchen operating efficiency and equipment stability, thus obtaining a key feature candidate set. The task load level, energy efficiency, equipment health and operation time distribution rate in the candidate set of key features are selected by using the feature importance evaluation algorithm to obtain the key operation indicator feature set. Based on the key operational indicator feature set, comprehensive operational evaluation parameters are calculated to obtain the operational feature parameter set; The expression for the multidimensional correlation analysis model is: ； in, The number of feature dimensions involved in the calculation; For the first An observation sample set of each operating state feature vector; For the first A sample set of operational efficiency response feature vectors; The covariance between the two features; and Features and Standard deviation; and Features and The mean; To prevent division by zero and small positive constants in smooth calculations; It is a comprehensive correlation index used to characterize the multidimensional mapping strength between the kitchen status and the operating results of each operating dimension.

5. A method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing as described in claim 4, characterized in that, The expression for the digital twin model of the kitchen is: ； in, For at any time The corresponding digital twin model of the kitchen is output in a comprehensive manner to represent the operating status of the virtual kitchen; The number of edge intelligent nodes; For the first The device operating stability parameters collected by each node; For the first Task completion coefficient for each node; For the first The energy consumption of a single task recorded by each node; For the first The ambient temperature parameters sampled at each node; The time sampling interval; and These are automatically calculated dynamic scaling factors used to balance the impact of task and environmental characteristics; A small constant to prevent division by zero or over-amplification of errors; This represents the mean of the reference correlation for the same kitchen within the cloud-based historical window. This is a dynamic decay factor that controls the model's sensitivity to historical biases. This is the global scaling constant of the model, used to calibrate the scale between the twin space and the actual state.

6. The method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing as described in claim 1, characterized in that, The process involves real-time analysis of actual task requirements and resource usage status based on the kitchen digital twin model and the edge intelligent nodes, resulting in a task classification result set and a task summary dataset, including: Each edge intelligent node collects task request information and equipment resource usage status in the current kitchen scene, and compares the task request information and equipment resource usage status with the virtual operation status in the kitchen digital twin model to obtain a task-resource matching dataset; Based on the task-resource matching dataset, the execution priority and resource usage ratio of each task are calculated using a load balancing evaluation algorithm and a resource constraint analysis model to obtain a task priority evaluation result set. Based on the task priority evaluation result set, a comprehensive analysis is performed on the multi-task parallelism, energy efficiency and equipment operating load in the kitchen scenario to obtain a task classification result set. Based on the task classification result set, the running time, device call frequency and energy consumption statistics of each task are summarized to obtain a task summary dataset.

7. A method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing as described in claim 6, characterized in that, Based on the task-resource matching dataset, the execution priority and resource usage ratio of each task are calculated using a load balancing evaluation algorithm and a resource constraint analysis model to obtain a task priority evaluation result set, including: The task-resource matching dataset is standardized to establish the constraint relationship between task demand, resource availability, and execution latency, thus obtaining the task resource constraint matrix. Based on the task resource constraint matrix, a multi-objective load balancing evaluation algorithm is used to calculate the load occupancy rate of each edge intelligent node and the sensitivity of the task to resources, and a node load evaluation result set is obtained. Based on the node load assessment result set, the task scheduling priority sequence and resource optimization allocation ratio are solved using the resource constraint analysis model to obtain the task priority calculation result set; The task priority calculation result set is subjected to comprehensive weight normalization and stability correction to obtain the task priority evaluation result set. The expression for the resource constraint analysis model is: ； in, This is the set of optimal solutions for the resource constraint analysis model, used to output the ratio of task allocation to resource utilization. The set of resource allocation strategy variables to be solved; Total number of tasks; For the first The execution delay function of each task under the current allocation strategy; For the first The amount of available computing resources corresponding to each task; For the first Real-time resource utilization of each task; This serves as a historical average resource utilization benchmark for similar tasks. These are the dynamic state values ​​calculated by the kitchen digital twin model during task execution; The task reference twin state is stored in the cloud management center; and These are the internal adjustment coefficients for automatically balancing the influence of different constraints.

8. A method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing as described in claim 1, characterized in that, The cloud management center receives the task classification result set and task summary dataset, and performs comprehensive analysis and optimization decisions on the task classification result set and task summary dataset in conjunction with the historical operation database. This generates cross-kitchen task allocation schemes, energy consumption optimization schemes, and food safety control schemes, resulting in a set of optimization control strategies distributed from the cloud, including: The cloud management center receives the task classification result set and task summary dataset uploaded by each edge intelligent node, and performs correlation matching between the task classification result set, task summary dataset and corresponding records in the historical operation database to obtain a multi-kitchen historical operation comparison dataset. Based on the historical operation comparison dataset of the multi-kitchen, cluster analysis and pattern recognition algorithms are used to extract task synergy, energy consumption characteristics and risk factors to obtain a cross-kitchen operation feature mapping set. Based on the cross-kitchen operation feature mapping set, the task allocation efficiency, energy consumption minimization and cold chain food safety constraints are comprehensively solved to obtain the cloud-based optimization decision result set. Based on the cloud-based optimization decision result set, a cross-kitchen task allocation scheme, energy consumption optimization scheme, and food safety control scheme are generated; Based on cross-kitchen task allocation schemes, energy consumption optimization schemes, and food safety control schemes, unified decision instructions are generated and issued for execution, resulting in a set of optimized control strategies distributed from the cloud.

9. A method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing as described in claim 1, characterized in that, Each edge intelligent node performs self-diagnosis based on the optimized control strategy set and the real-time updated task operation status information of the kitchen digital twin model, obtaining diagnostic results. If the diagnostic results indicate performance deviation, task anomaly, or equipment imbalance, the model parameters and control commands are automatically adjusted, and the optimized parameters are shared among adjacent edge intelligent nodes to obtain an adaptive parameter dataset, including: Each edge intelligent node receives the set of optimized control strategies from the cloud and compares the set of optimized control strategies with the real-time updated task running status information in the kitchen digital twin model to obtain a strategy-state matching dataset. Based on the policy-state matching dataset, a performance evaluation algorithm is used to detect task execution deviations, energy consumption anomalies, and equipment operation imbalances, resulting in a task diagnosis and analysis result set. When the task diagnosis analysis result set characterizes performance deviation, task abnormality or equipment imbalance, each edge intelligent node automatically adjusts the relevant parameters and control commands in the kitchen digital twin model according to the deviation type to obtain the corrected model parameter set. The modified model parameter set is shared among adjacent edge smart nodes, and after parameter consistency is confirmed, an adaptive parameter dataset is obtained.

10. A method for remote collaborative management of a central smart kitchen based on the Internet of Things and edge computing as described in claim 1, characterized in that, Each edge intelligent node utilizes the adaptive parameter dataset to perform consistency verification on task status, resource allocation, and control logic through a lightweight distributed consensus mechanism, resulting in a multi-node consistent context dataset, including: Each edge intelligent node receives the adaptive parameter dataset and loads the latest model parameters and task control instructions in its local operating environment to obtain a node local parameter mapping table; Based on the node local parameter mapping table, a lightweight distributed consensus mechanism is used to calculate the parameter differences between the task status and resource allocation of each node, and a node status difference dataset is obtained. Based on the node state difference dataset, a node synchronization calibration process is executed to dynamically balance and correct the task scheduling logic and control strategy with deviations, and a task and control logic calibration result set is obtained. The task and control logic calibration result set is subjected to consensus confirmation and eventual consistency verification across multiple nodes to obtain a multi-node consistency context dataset.

Images