A kitchen dynamic scheduling method and system based on real-time order data

Abstract

The application discloses a back kitchen dynamic scheduling method and system based on real-time order data, and relates to the technical field of intelligent catering. The method comprises the following steps: acquiring real-time order data; collecting back kitchen state data, including stove occupation state, chef task queue and skill label, and dish standard preparation time; constructing a multi-objective dynamic optimization model, taking the real-time order data and the back kitchen state data as inputs, taking the minimization of order average waiting time and the maximization of equipment utilization as optimization objectives, and generating an optimal scheduling scheme; outputting scheduling instructions, including cooking task sequence adjustment instructions and manpower reallocation instructions; acquiring the actual completion time of each dish through real-time feedback, and dynamically correcting model parameters. The system comprises an order data acquisition module, a back kitchen state sensing module, a dynamic scheduling decision module, an instruction output module and a feedback correction module. The application realizes real-time dynamic optimization scheduling of back kitchen tasks, significantly reduces order waiting time during meal peak periods, improves back kitchen operation efficiency, and can be widely applied to intelligent canteens, chain catering and take-out kitchens and the like.

Description

Technical Field

[0001] This invention relates to the field of smart catering technology, and more specifically, to a method and system for dynamic scheduling of kitchen operations based on real-time order data. It is particularly suitable for scenarios such as smart canteens, chain catering enterprises, and takeaway kitchens that require real-time optimization and scheduling of kitchen production tasks. Background Technology

[0002] As the digital transformation of the catering industry accelerates, Back-of-House (BOH) systems are evolving from basic inventory recording tools into intelligent operational hubs. During peak dining hours, a large number of orders flood the kitchen simultaneously. Traditional order scheduling methods rely heavily on chefs' experience and judgment, which can easily lead to chaotic food preparation decisions and order backlogs, severely impacting customer experience. Statistics show that restaurants using traditional manual order scheduling often experience average wait times exceeding 30 minutes during peak hours, and the rationality of food preparation order is difficult to guarantee.

[0003] In the existing technology, there are some solutions for kitchen scheduling. For example, patent document CN110930040A discloses an intelligent kitchen system based on dynamic priority, mainly for handling the special event of "returned dishes," using a greedy algorithm to find the same dishes as the returned dishes among the dishes waiting to be prepared and then redistributing them. However, this solution focuses on handling abnormal events and fails to solve the problem of global dynamic scheduling under normal order flow. Another example is patent application CN121582030A, which discloses a method for arranging restaurant dishes in the kitchen, establishing a dining database based on customer reservations two days in advance to coordinate dish scheduling. However, this solution is based on a reservation system and is difficult to cope with sudden changes in real-time orders and the need for dynamic adjustments during peak periods. Furthermore, while some existing capacity management systems achieve visualized order monitoring, their main functions remain at the level of order display and simple sequence adjustment, lacking intelligent decision-making capabilities based on multi-objective optimization.

[0004] Kitchen scheduling is essentially a complex multi-objective decision-making problem involving various factors such as dish type, preparation time, equipment occupancy, chef skills, and order priority. Existing technologies generally suffer from the following shortcomings: (1) a lack of dynamic response capability to real-time order data, with scheduling schemes often being static or based solely on simple rules; (2) a failure to effectively integrate kitchen status data (such as stove occupancy and chef workload) for collaborative scheduling; and (3) a lack of feedback loop mechanism, making it impossible to dynamically optimize the scheduling model based on actual meal preparation time. Therefore, there is an urgent need for a kitchen scheduling method that can dynamically optimize the cooking sequence and manpower arrangement based on real-time order data, comprehensively considering multiple constraints. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for dynamic kitchen scheduling based on real-time order data, so as to solve the technical problems of slow kitchen scheduling response, poor decision-making, and inability to dynamically adapt to changes in orders in the prior art, thereby effectively reducing waiting time during peak dining hours and improving kitchen operation efficiency.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a dynamic kitchen scheduling method based on real-time order data, comprising the following steps: acquiring real-time order data, the real-time order data including order identifier, dish information, order time, and dining time attributes; collecting kitchen status data, the kitchen status data including the occupancy status of multiple stoves, the current task queue and skill tags of each chef's position, and the standard preparation time of each dish; constructing a multi-objective dynamic optimization model, using the real-time order data and the kitchen status data as input, with the optimization objectives of minimizing the average order waiting time and maximizing equipment utilization, to generate an optimal scheduling scheme; outputting scheduling instructions, the scheduling instructions including cooking task sequence adjustment instructions and manpower reallocation instructions; real-time feedback and closed-loop correction, obtaining the actual completion time of each dish through sensors or manual confirmation, comparing the actual completion time with the predicted completion time, and dynamically correcting the parameters of the multi-objective dynamic optimization model.

[0007] Preferably, the construction of the multi-objective dynamic optimization model further includes: constructing a scheduling decision model using a reinforcement learning algorithm, defining a state space, an action space, and a reward function; wherein, the state space includes the current order backlog, the load status of each stove, and the workload of each chef; the action space includes assigning specific orders to specific stoves or chefs and adjusting the execution order of existing task sequences; the reward function is set based on the reduction rate of the average waiting time of orders and the reduction rate of equipment idle rate.

[0008] Preferably, the generation of the optimal scheduling scheme includes: extracting the preparation time attribute and equipment dependency attribute of each dish based on the dish information; calculating the expected idle time of each stove and the expected task completion time of each chef based on the kitchen status data; solving the Pareto optimal solution set through a multi-objective genetic algorithm, and selecting the scheduling scheme with the highest overall satisfaction from the Pareto optimal solution set.

[0009] Preferably, the manpower reallocation instruction includes: when the current task queue length of the first chef position exceeds a first threshold, automatically identifying a second chef position with the same skill tag; if the current task queue length of the second chef position is lower than a second threshold, generating an instruction to transfer some tasks of the first chef position to the second chef position.

[0010] Preferably, the real-time feedback and closed-loop correction further include: calculating the deviation between the actual completion time of each dish and the completion time predicted by the multi-objective dynamic optimization model; if the deviation value exceeds a preset range, triggering an online learning and update mechanism for model parameters, and updating the model weights using incremental learning.

[0011] This invention also provides a kitchen dynamic scheduling system based on real-time order data, comprising: an order data acquisition module for acquiring real-time order data, the real-time order data including order identifier, dish information, order time, and dining time attributes; a kitchen status perception module for collecting kitchen status data, the kitchen status data including the occupancy status of multiple stoves, the current task queue and skill tags of each chef's position, and the standard preparation time of each dish; a dynamic scheduling decision module connected to the order data acquisition module and the kitchen status perception module, for constructing a multi-objective dynamic optimization model, using the real-time order data and the kitchen status data as input, with the optimization objectives of minimizing the average order waiting time and maximizing equipment utilization, to generate an optimal scheduling scheme; an instruction output module connected to the dynamic scheduling decision module, for outputting scheduling instructions, the scheduling instructions including cooking task sequence adjustment instructions and manpower reallocation instructions; and a feedback correction module for obtaining the actual completion time of each dish through sensors or manual confirmation, comparing the actual completion time with the predicted completion time, and dynamically correcting the parameters of the multi-objective dynamic optimization model.

[0012] Preferably, the dynamic scheduling decision module includes a reinforcement learning decision unit, which is used to: define a state space, an action space, and a reward function; wherein, the state space includes the current order backlog, the load status of each stove, and the workload of each chef; the action space includes assigning specific orders to specific stoves or chefs and adjusting the execution order of existing task sequences; the reward function is set based on the reduction rate of the average order waiting time and the reduction rate of equipment idle rate.

[0013] Preferably, the dynamic scheduling decision module includes a multi-objective optimization unit, which is used to: extract the preparation time attribute and equipment dependency attribute of each dish based on the dish information; calculate the expected idle time of each stove and the expected task completion time of each chef based on the kitchen status data; solve the Pareto optimal solution set through a multi-objective genetic algorithm, and select the scheduling scheme with the highest overall satisfaction from the Pareto optimal solution set.

[0014] Preferably, the instruction output module includes: a task sequence adjustment unit, used to generate cooking task sequence adjustment instructions and send them to the display screens of each stove terminal; and a human resource scheduling unit, used to automatically identify a second chef position with the same skill tag when the current task queue length of the first chef position exceeds a first threshold, and to generate an instruction to transfer some tasks of the first chef position to the second chef position if the current task queue length of the second chef position is lower than a second threshold.

[0015] Preferably, the feedback correction module includes: a deviation calculation unit, used to calculate the deviation between the actual completion time of each dish and the completion time predicted by the multi-objective dynamic optimization model; and an online learning unit, used to trigger an online learning update mechanism for model parameters when the deviation value exceeds a preset range, and to update the model weights of the dynamic scheduling decision module using incremental learning.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention acquires order data and kitchen status data in real time, constructs a multi-objective dynamic optimization model, realizes dynamic optimization scheduling of cooking order and manpower arrangement, and can significantly reduce the average waiting time of orders during peak dining hours compared with the traditional order scheduling method that relies on chef experience; (2) The present invention adopts intelligent optimization methods such as reinforcement learning or multi-objective genetic algorithm, which can solve the optimal scheduling scheme under multiple constraints, and overcomes the limitations of the prior art which is based on simple rules or single objective optimization; (3) The present invention introduces a real-time feedback and closed-loop correction mechanism, and dynamically corrects the model parameters through actual completion time, so that the system has adaptive learning ability, can continuously optimize the scheduling effect, and adapt to changes in the kitchen operation environment; (4) The present invention realizes cross-position task collaboration through manpower redistribution instructions, effectively balances the load of each chef position, avoids excessive backlog of orders in individual positions, and improves the overall meal preparation efficiency; (5) The present invention can be widely applied to smart canteens, chain restaurants, takeaway kitchens and other scenarios, and has good versatility and promotion value. Attached Figure Description Figure 1: System structure block diagram. This is a structural block diagram of the kitchen dynamic scheduling system based on real-time order data provided in an embodiment of the present invention. Description of the drawing: Figure 1 illustrates the system architecture of the present invention, including five core modules and their interconnections. Figure 2 The flowchart is a schematic diagram of the kitchen dynamic scheduling method based on real-time order data provided in an embodiment of the present invention. (See attached figures for details.) Figure 2 The five core steps of the present invention and their sequential relationship, as well as the closed-loop feedback path, are illustrated. Figure 3The schematic diagram of the multi-objective dynamic optimization model is a schematic diagram of the construction and optimization process of the multi-objective dynamic optimization model in this embodiment of the invention. Figure 3 illustrates the input, processing, optimization objectives, and output process of the multi-objective dynamic optimization model. Figure 4 The flowchart for generating manpower reallocation instructions is a schematic diagram of the process for generating manpower reallocation instructions in an embodiment of the present invention. (See attached figures for details.) Figure 4 The logic flow of dynamic manpower scheduling based on job load is demonstrated. Figure 5 The flowchart illustrating the real-time feedback and closed-loop correction mechanism in this embodiment of the invention is shown below. Figure 5 illustrates the closed-loop correction process for deviation calculation and online learning.

Claims

1. A method for dynamic kitchen scheduling based on real-time order data, characterized in that, Includes the following steps: Obtain real-time order data, which includes order identifier, menu information, order time, and dining time period attributes; Collect kitchen status data, which includes the occupancy status of multiple stoves, the current task queue and skill tags of each chef position, and the standard preparation time of each dish. A multi-objective dynamic optimization model is constructed, taking the real-time order data and the kitchen status data as inputs, with the optimization objectives of minimizing the average order waiting time and maximizing equipment utilization, to generate the optimal scheduling scheme; Output scheduling instructions, which include cooking task sequence adjustment instructions and manpower reallocation instructions; Real-time feedback and closed-loop correction: The actual completion time of each dish is obtained through sensors or manual confirmation. The actual completion time is compared with the predicted completion time, and the parameters of the multi-objective dynamic optimization model are dynamically corrected.

2. The kitchen dynamic scheduling method based on real-time order data according to claim 1, characterized in that, The construction of the multi-objective dynamic optimization model further includes: A scheduling decision model is constructed using a reinforcement learning algorithm, defining a state space, an action space, and a reward function. The state space includes the current order backlog, the load status of each stove, and the workload of each chef. The action space includes assigning specific orders to specific stoves or chefs and adjusting the execution order of existing task sequences. The reward function is set based on the reduction rate of average order waiting time and the reduction rate of equipment idle time.

3. The kitchen dynamic scheduling method based on real-time order data according to claim 1, characterized in that, The generation of the optimal scheduling scheme includes: Based on the dish information, the preparation time attribute and equipment dependency attribute of each dish are extracted; based on the kitchen status data, the expected idle time of each stove and the expected task completion time of each chef are calculated; a Pareto optimal solution set is solved by a multi-objective genetic algorithm, and the scheduling scheme with the highest overall satisfaction is selected from the Pareto optimal solution set.

4. The back-kitchen dynamic scheduling method based on real-time order data according to claim 1, characterized in that, The human resource reallocation instruction includes: When the current task queue length of the first chef position exceeds the first threshold, the second chef position with the same skill tag is automatically identified; if the current task queue length of the second chef position is lower than the second threshold, an instruction is generated to transfer some tasks of the first chef position to the second chef position.

5. The kitchen dynamic scheduling method based on real-time order data according to claim 1, characterized in that, The real-time feedback and closed-loop correction further include: The deviation between the actual completion time of each dish and the completion time predicted by the multi-objective dynamic optimization model is calculated; if the deviation value exceeds the preset range, the online learning and updating mechanism of model parameters is triggered, and the model weights are updated using incremental learning.

6. A dynamic kitchen scheduling system based on real-time order data, characterized in that, include: The order data acquisition module is used to acquire real-time order data, which includes order identifier, menu information, order time, and dining time period attributes. The kitchen status sensing module is used to collect kitchen status data, which includes the occupancy status of multiple stoves, the current task queue and skill tags of each chef position, and the standard preparation time of each dish. The dynamic scheduling decision module is connected to the order data acquisition module and the kitchen status perception module. It is used to construct a multi-objective dynamic optimization model. The model takes the real-time order data and kitchen status data as inputs and aims to minimize the average order waiting time and maximize equipment utilization to generate the optimal scheduling scheme. The instruction output module, connected to the dynamic scheduling decision module, is used to output scheduling instructions, including cooking task sequence adjustment instructions and manpower reallocation instructions. The feedback correction module is used to obtain the actual completion time of each dish through sensors or manual confirmation, compare the actual completion time with the predicted completion time, and dynamically correct the parameters of the multi-objective dynamic optimization model.

7. The kitchen dynamic scheduling system based on real-time order data according to claim 6, characterized in that, The dynamic scheduling decision module includes a reinforcement learning decision unit, which is used for: Define a state space, an action space, and a reward function; wherein, the state space includes the current order backlog, the load status of each stove, and the workload of each chef; the action space includes assigning specific orders to specific stoves or chefs and adjusting the execution order of existing task sequences; the reward function is set based on the reduction rate of average order waiting time and the reduction rate of equipment idle rate.

8. The kitchen dynamic scheduling system based on real-time order data according to claim 6, characterized in that, The dynamic scheduling decision module includes a multi-objective optimization unit, which is used for: Based on the dish information, the preparation time attribute and equipment dependency attribute of each dish are extracted; based on the kitchen status data, the expected idle time of each stove and the expected task completion time of each chef are calculated; a Pareto optimal solution set is solved by a multi-objective genetic algorithm, and the scheduling scheme with the highest overall satisfaction is selected from the Pareto optimal solution set.

9. The kitchen dynamic scheduling system based on real-time order data according to claim 6, characterized in that, The instruction output module includes: The task sequence adjustment unit is used to generate cooking task sequence adjustment instructions and send them to the display screens of each stove terminal. The human resource scheduling unit is used to automatically identify a second chef position with the same skill tag when the current task queue length of the first chef position exceeds a first threshold. If the current task queue length of the second chef position is lower than a second threshold, an instruction is generated to transfer some tasks of the first chef position to the second chef position.

10. The kitchen dynamic scheduling system based on real-time order data according to claim 6, characterized in that, The feedback correction module includes: The deviation calculation unit is used to calculate the deviation between the actual completion time of each dish and the completion time predicted by the multi-objective dynamic optimization model. The online learning unit is used to trigger the online learning and update mechanism of model parameters when the deviation value exceeds the preset range, and to update the model weights of the dynamic scheduling decision module using incremental learning.

Images