Inventory threshold value and purchase quantity calculation processing method and system based on AI prediction

Abstract

The invention relates to an inventory threshold and purchase quantity calculation processing method and system based on AI prediction, and relates to the field of supply chain management, and the method comprises the steps: collecting historical data, real-time inventory data and external parameter data; preprocessing the historical data to form a training data set and an input data set; using the training data set to train the long-short-term memory network model to obtain a prediction model; inputting the input data set into a prediction model for analysis and prediction to determine a future demand trend; determining a material inventory threshold value according to the future demand trend and preset industry characteristic parameters; obtaining a basic purchase quantity according to the future demand trend, the real-time inventory data and the external parameter data; a material purchase amount is generated based on the external parameter data, the material inventory threshold, and the base purchase amount. The method and the system have the effects of intellectualization and accuracy of purchase decision.

Description

A method and system for calculating and processing inventory thresholds and purchase quantities based on AI prediction Technical Field

[0001] This invention relates to the field of supply chain management, and in particular to a method and system for calculating and processing inventory thresholds and purchase quantities based on AI prediction. Background Technology

[0002] Inventory threshold refers to the critical level of material inventory set to trigger replenishment or warnings, while purchase quantity calculation refers to the decision-making process for the quantity of materials ordered to meet future demand.

[0003] In existing technologies, inventory management and procurement decisions in the catering industry mainly rely on human experience to set fixed inventory thresholds. When inventory is insufficient, managers estimate future demand based on historical sales averages or trends, and then calculate the procurement quantity.

[0004] However, fixed thresholds cannot adapt to dynamic demands such as seasonal changes and fluctuations in passenger flow during holidays, easily leading to inventory shortages during peak seasons or inventory backlogs during off-seasons. Furthermore, manual estimation of purchase quantities struggles to accurately and in real-time incorporate key parameters such as material loss rates, potentially resulting in over-purchasing leading to waste or under-purchasing causing supply disruptions. This reduces the intelligence and accuracy of procurement decisions and requires improvement. Summary of the Invention

[0005] To improve the intelligence and accuracy of procurement decisions, this invention provides a method and system for calculating and processing inventory thresholds and procurement quantities based on AI prediction.

[0006] In a first aspect, the present invention provides an AI-based method for calculating and processing inventory thresholds and purchase quantities, employing the following technical solution: An AI-based method for calculating and processing inventory thresholds and purchase quantities includes: collecting historical data, real-time inventory data, and external parameter data; preprocessing the historical data to form a training dataset and an input dataset; training a long short-term memory network model using the training dataset to obtain a prediction model; inputting the input dataset into the prediction model for analysis and prediction to determine future demand trends; determining material inventory thresholds based on future demand trends and preset industry characteristic parameters; obtaining a basic purchase quantity based on future demand trends, real-time inventory data, and external parameter data; and generating the material purchase quantity based on the external parameter data, the material inventory threshold, and the basic purchase quantity.

[0007] By adopting the above technical solution, a long short-term memory network model with time-series prediction capabilities is trained using historical data and external parameters to obtain future demand trends. Then, based on this trend and industry characteristic parameters, material inventory thresholds are dynamically calculated, enabling inventory management to adapt to seasonal changes and holiday fluctuations. Finally, by integrating future demand trends, real-time inventory status, and external parameters, material procurement quantities are generated, realizing the transformation of procurement decisions from relying on manual estimation to intelligent calculation based on multi-source data, thereby effectively improving the intelligence and accuracy of procurement decisions.

[0008] Optionally, the model optimization method may also be included: collecting material consumption data within a preset detection period; calculating the prediction deviation value based on the material consumption data and future demand trends; generating optimized model parameters based on the prediction deviation value when the prediction deviation value exceeds a preset deviation threshold; and updating the prediction model based on the optimized model parameters.

[0009] By adopting the above technical solution, actual material consumption data is collected and compared with the predicted trend of the AI ​​model to calculate the deviation. When the deviation exceeds the allowable range, optimization parameters are generated and applied based on the deviation characteristics to update the prediction model, thereby ensuring the accuracy and reliability of long-term prediction.

[0010] Optionally, real-time control methods may also be included: collecting data on pedestrian density at intersections, parking lot vacancy rates, and queue lengths at competing restaurants; combining these data to calculate a passenger flow impact index; when the passenger flow impact index exceeds a preset passenger flow threshold, calculating the excess value based on the passenger flow impact index and the passenger flow threshold; determining the inventory increase ratio based on the excess value; and increasing the material inventory threshold based on the inventory increase ratio.

[0011] By adopting the above technical solution, the passenger flow impact index is calculated using the pedestrian density at intersections, parking lot vacancy rates, and queue lengths at competing restaurants. When the index exceeds a preset threshold, the inventory threshold of the corresponding materials is calculated and adjusted upwards based on the extent of the excess. This achieves dynamic adjustment of inventory levels based on real-time changes in the external environment, enhancing the responsiveness of the inventory management system to sudden passenger flow opportunities.

[0012] Optionally, pressure response methods are also included: when the passenger flow impact index exceeds a preset pressure threshold, real-time consumption rate of materials, current order data, and reserved order data are collected; an estimated consumption rate is calculated based on the current order data and reserved order data; the real-time consumption rate and the estimated consumption rate are superimposed to obtain a comprehensive consumption rate; the material sustainability time is obtained based on the comprehensive consumption rate and real-time inventory data; when the material sustainability time is not greater than a preset second warning time, a replenishment urgency coefficient is determined based on the material sustainability time and a preset safety redundancy time; an expedited replenishment instruction is generated based on the replenishment urgency coefficient; when the material sustainability time falls within a preset inventory synchronization range, an inventory synchronization prompt is reported.

[0013] By adopting the above technical solution, the inventory sustainability is quantified by integrating real-time consumption and known order data, and a tiered response mechanism is triggered based on the different warning intervals of the duration, thereby improving the resilience and stability of the inventory system during peak demand periods.

[0014] Optionally, an internal adjustment method is also included: when the material's sustainability time is no greater than the second warning time, the remaining material inventory is determined based on real-time inventory data, and the capacity load data of each workstation in the back kitchen is collected; the inventory-related dishes are determined based on the remaining material inventory and the preset daily menu; the replaceable dishes are determined based on the inventory-related dishes; the replaceable dishes and capacity load data are combined to select executable replaceable dishes; an internal adjustment plan is generated based on the executable replaceable dishes; the net material demand gap is calculated based on the internal adjustment plan, and the expedited replenishment order is modified based on the net material demand gap.

[0015] By adopting the above technical solutions, internal resource scanning is initiated when inventory is in urgent need. Combined with capacity load screening, executable replacement dishes are selected and adjustment plans are generated. In this way, existing materials and capacity are used to internally absorb part of the demand, achieving precise reduction of urgent replenishment needs and improving self-adjustment capability and resource utilization efficiency under supply chain pressure.

[0016] Optionally, it also includes a method for generating internal adjustment plans: determining the matching degree of dishes based on the list of replaceable dishes and the dishes of the day; determining the load increment based on the capacity load data and replaceable dishes; combining the matching degree of dishes and the load increment to comprehensively score and rank each replaceable dish to obtain the ranking result; and generating an internal adjustment plan based on the ranking result and the remaining inventory of materials.

[0017] By adopting the above technical solutions, quantifying the impact of dish matching degree and production capacity load, and conducting multi-objective comprehensive scoring and ranking of alternative options, the optimal internal adjustment plan that balances flavor continuity and kitchen execution efficiency is automatically generated when inventory is tight, thus improving the scientific nature and operability of the plan.

[0018] Optionally, it also includes a heat perception and pre-adjustment method: collecting local catering topic data; extracting volume growth rate, sentiment trend data, and local user participation data from the local catering topic data; combining the volume growth rate, sentiment trend data, and local user participation data to calculate the topic relevance; determining the pre-adjustment coefficient based on the topic relevance; reducing the fusion weight of intersection personnel density in the calculation of the passenger flow impact index based on the pre-adjustment coefficient, and recalculating the passenger flow impact index based on the reduced fusion weight; updating the material inventory threshold based on the recalculated passenger flow impact index.

[0019] By adopting the above technical solution, the popularity of catering topics on social media is monitored and quantified, and the sensitivity parameters of the customer flow perception system are adjusted according to the popularity. This allows for the pre-adjustment of inventory thresholds before the popularity of topics translates into actual customer flow, thus realizing a shift from passive response to proactive prediction in demand management and enhancing the system's preparedness for unexpected events.

[0020] Optionally, a heat cycle procurement method is also included: when the topic relevance exceeds a preset high relevance threshold, the demand growth coefficient is determined based on the topic relevance; the additional material procurement quantity is calculated by combining the demand growth coefficient, real-time inventory data, and material inventory threshold; an incremental procurement order is generated based on the additional material procurement quantity; and the incremental procurement order is superimposed with the material procurement quantity to generate a total procurement order.

[0021] By adopting the above technical solution, when the topic relevance exceeds the relevance threshold, the demand growth coefficient is obtained by understanding the topic relevance. This coefficient is then combined with real-time inventory data and material inventory thresholds to determine the additional material purchase quantity, thereby generating an incremental purchase order. Finally, this order is superimposed on the regular material purchase quantity to generate a total purchase order, thus ensuring that the procurement decision is both forward-looking and systematic.

[0022] Optionally, a method for monitoring the decline in popularity is also included: after generating an incremental purchase order, subsequent catering topic data is collected; the topic relevance is updated based on the subsequent catering topic data; when the updated topic relevance is lower than the high relevance threshold and shows a continuous downward trend, downward trend data is collected; the demand decay coefficient is determined based on the downward trend data; the reduction purchase quantity is calculated based on the demand decay coefficient and the additional purchase quantity of materials; a purchase reduction order is generated based on the reduction purchase quantity, and the total purchase order is updated based on the purchase reduction order.

[0023] By adopting the above technical solutions, continuously tracking the declining trend of topic popularity, and calculating the amount of procurement to be reduced based on quantitative data, the incremental procurement plan in the early stage can be reduced when the popularity fades, effectively reducing the backlog of materials caused by the fading of the hot topic.

[0024] Secondly, this application provides an AI-based inventory threshold and purchase quantity calculation and processing system, which adopts the following technical solution: An AI-based inventory threshold and purchase quantity calculation and processing system includes: a data acquisition module for acquiring historical data, real-time inventory data, and external parameter data; a memory for storing a program that implements an AI-based inventory threshold and purchase quantity calculation and processing method; and a processor for loading and executing the program stored in the memory.

[0025] In summary, this application includes at least one of the following beneficial technical effects: 1. By using historical data and external parameters to train a long short-term memory network model with time-series prediction capabilities, future demand trends can be obtained; then, based on this trend and industry characteristic parameters, material inventory thresholds are dynamically calculated, enabling inventory management to adapt to seasonal changes and holiday fluctuations; finally, by integrating future demand trends, real-time inventory status, and external parameters, material procurement quantities are generated, realizing the transformation of procurement decisions from relying on manual estimation to intelligent calculation based on multi-source data, thereby effectively improving the intelligence and accuracy of procurement decisions; 2. Utilizing intersection personnel density, The system calculates the customer flow impact index based on parking lot vacancy rate and queue length at competing restaurants. When the index exceeds a preset threshold, it calculates and adjusts the corresponding material inventory threshold based on the extent of the exceedance. This achieves dynamic adjustment of inventory levels based on real-time changes in the external environment, enhancing the responsiveness of the inventory management system to sudden customer flow opportunities. 3. By monitoring and quantifying the popularity of restaurant topics on social media and adjusting the sensitivity parameters of the customer flow perception system based on the popularity, the system pre-adjusts the inventory threshold before the topic popularity translates into actual customer flow. This achieves a shift from passive response to proactive prediction in demand management, enhancing the system's preparedness for unexpected events. Attached Figure Description

[0026] Figure 1 is a flowchart of a method for calculating and processing inventory thresholds and purchase quantities based on AI prediction. Detailed Implementation

[0027] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0028] Referring to Figure 1, this application discloses an AI-based prediction method for calculating and processing inventory thresholds and purchase quantities, including the following steps: S10: Collecting historical data, real-time inventory data, and external parameter data.

[0029] Historical data refers to structured operational records extracted from the restaurant management system over a continuous period of time in the past. In this embodiment, it may include the following data collected from January 1, 2022 to December 31, 2024: the total number of diners per day, the daily consumption weight of pork belly (fresh food), and the daily consumption weight of rice. The restaurant management system records and updates the above data daily.

[0030] Real-time inventory data refers to the actual physical inventory quantity of all materials at a given moment, obtained through an interface with inventory management software. Inventory management software includes real-time inventory data.

[0031] External parameter data refers to environmental and planning information that may affect future demand. Examples include holidays and work schedules. External parameter data is obtained by retrieving preset holiday and work schedules from the restaurant management system.

[0032] S11: Preprocess historical data to form training and input datasets.

[0033] The training dataset refers to the data set used to train the Long Short-Term Memory (LSTM) network prediction model. Before construction, the data underwent preprocessing, such as using the 3σ principle to remove outliers in the number of diners due to equipment malfunctions (e.g., a record of 0 diners on a certain day), and linear interpolation to complete missing rice consumption data. The training dataset was obtained by preprocessing historical data.

[0034] The input dataset refers to the set of data used for the current prediction. In this embodiment, the preprocessed historical data for the entire year of 2024 (January 1 to December 31, 2024) is used as the input dataset.

[0035] S12: Train the Long Short-Term Memory network model using the training dataset to obtain a prediction model.

[0036] The prediction model refers to a pre-trained LSTM neural network model with fixed parameters. In this embodiment, the LSTM model is trained using the aforementioned training dataset. The training process is configured as follows: using the Adam optimizer with mean squared error (MSE) as the loss function, the model converges through iterative calculation, ultimately obtaining a fully trained model capable of predicting future demand based on input data.

[0037] S13: Input the input dataset into the prediction model for analysis and prediction to determine future demand trends.

[0038] Future demand trends refer to the quantitative prediction sequence output by the prediction model for a specific future period. In this embodiment, the input dataset is input into the trained prediction model to obtain the demand forecast for February 2025 (especially the period after the Spring Festival). The specific output is: the predicted average daily number of diners is 300, the average daily consumption of pork belly is 30kg, and the average daily consumption of rice is 20kg.

[0039] S14: Determine the material inventory threshold based on future demand trends and preset industry characteristic parameters.

[0040] Industry-specific parameters refer to rules preset based on management practices in the catering industry. In this embodiment, the following rules are preset: the minimum inventory threshold for fresh produce is adjusted by no less than 50%, and the minimum inventory threshold for dry goods is adjusted by no less than 20%.

[0041] Material inventory threshold refers to the critical inventory level that is dynamically calculated for each type of material.

[0042] Material inventory thresholds are obtained by querying a pre-defined inventory adjustment lookup table. This table records the material inventory thresholds corresponding to different combinations of future demand trends and industry characteristic parameters. The inventory adjustment lookup table is pre-constructed by those skilled in the art for different material categories and their corresponding industry characteristic parameters, and will not be elaborated upon here.

[0043] S15: Determine the basic purchase quantity based on future demand trends, real-time inventory data, and external parameter data.

[0044] The basic procurement quantity refers to the initial procurement quantity calculated based on current inventory to meet future projected demand. In this embodiment, for the projected total demand (300 kg of pork belly and 200 kg of rice) from February 7th to February 16th, 2025 (a total of 10 days), combined with the real-time inventory of S10 (15 kg of pork belly and 50 kg of rice) and the preset loss rate (5% for pork belly and 2% for rice), the basic procurement quantity is calculated using the formula: Basic Procurement Quantity = (Projected Demand - Current Inventory) × (1 + Loss Rate). The basic procurement quantity for pork belly is 299.25 kg, and the basic procurement quantity for rice is 153 kg.

[0045] S16: Generate material purchase quantity based on external parameter data, material inventory threshold, and basic purchase quantity.

[0046] Material procurement quantity refers to the final quantity of actual purchase orders generated.

[0047] The material procurement quantity can be obtained through a pre-set material procurement table. This table records the final material procurement quantity corresponding to different combinations of basic procurement quantity, material inventory threshold, and external parameter data. For example, if the basic procurement quantity shows that 299.25 kg of pork belly needs to be purchased, the current material inventory threshold is 75 kg, and the external parameter data shows that the next day is a holiday, then the corresponding result obtained from querying this table is: adding an 8% safety buffer to the basic procurement quantity, the final material procurement quantity is determined to be approximately 323 kg. If the basic procurement quantity, material inventory threshold, and other conditions are different, other corresponding procurement quantity results can be obtained by querying. This comparison table is pre-constructed by those skilled in the art based on procurement strategies and historical experience, and will not be elaborated here.

[0048] It also includes model optimization methods: S20: Collect material consumption data within a preset detection period.

[0049] The testing period refers to a fixed time period set for evaluating the accuracy of the prediction model. In this embodiment, the period is set to one month.

[0050] Material consumption data refers to the total actual consumption of various materials in the restaurant during the monitoring period. This data is obtained by connecting to inventory management software. The inventory management software records the consumption of various materials.

[0051] S21: The prediction deviation value is calculated based on material consumption data and future demand trends.

[0052] Prediction deviation is a metric used to quantify the accuracy of a forecasting model. It is calculated by comparing the actual total material consumption within a monitoring period with the predicted total material demand for that monitoring period derived from future demand trends. The formula is: Prediction Deviation = |Actual Total Consumption - Predicted Total Demand| / Predicted Total Demand × 100%.

[0053] S22: When the prediction deviation value exceeds the preset deviation threshold, generate optimized model parameters based on the prediction deviation value.

[0054] The deviation threshold is a critical value used to determine whether the prediction accuracy is acceptable. In this embodiment, this threshold is set to 10%. If the prediction deviation exceeds 10%, the model prediction accuracy is deemed insufficient, and optimization needs to be initiated.

[0055] Optimizing model parameters refers to configurable items used to adjust the behavior of a prediction model.

[0056] Model parameters are optimized by querying a pre-defined parameter adjustment table. This table records the specific model parameter adjustments corresponding to different prediction deviation values ​​and the error characteristics they reflect. The parameter adjustment table is constructed in advance by those skilled in the art based on historical tuning experience and model characteristics, and will not be elaborated upon here.

[0057] When the prediction deviation exceeds the deviation threshold, the model parameters need to be optimized first for subsequent steps.

[0058] S23: Update the prediction model based on optimized model parameters.

[0059] The generated optimized model parameters are injected into the training process of the prediction model, thereby updating the prediction model and making its subsequent predictions more consistent with the actual consumption patterns.

[0060] It also includes real-time control methods: S30: collecting data on pedestrian density at intersections, parking lot vacancy rates, and queue lengths at competing restaurants.

[0061] Intersection crowd density refers to the quantitative value obtained by calling the heat map application programming interface (API) of an electronic map service provider to obtain the crowd gathering heat value of the main intersection area around the restaurant, and then normalizing it.

[0062] The parking lot vacancy rate refers to the ratio of the number of currently vacant parking spaces to the total number of parking spaces in a target parking lot, obtained by accessing the parking management system data interface of the cooperating commercial complex.

[0063] Competitive restaurant queue length refers to the value obtained by calling the public data interface of a restaurant queuing platform to obtain the real-time number of waiting tables or the estimated waiting time of similar restaurants within a set geofence, and then standardizing the result. The specific geofence range is set in advance by those skilled in the art and will not be elaborated here.

[0064] S31: The passenger flow impact index is calculated by combining the pedestrian density at the intersection, the parking lot vacancy rate, and the queue length of competing restaurants.

[0065] The Passenger Flow Impact Index is a comprehensive quantitative indicator that measures the impact of the real-time external environment on a restaurant's potential customer flow. It is calculated using a pre-defined weighted fusion algorithm based on three data points: intersection pedestrian density, parking lot vacancy rate, and competitor restaurant queue length. The specific formula is: Passenger Flow Impact Index = w1 * Intersection Pedestrian Density + w2 * (1 - Parking Lot Vacancy Rate) + w3 * Competitor Restaurant Queue Length, where w1, w2, and w3 are weighting coefficients pre-defined by those skilled in the art and will not be elaborated upon here.

[0066] S32: When the passenger flow impact index exceeds the preset passenger flow threshold, the excess value is calculated based on the passenger flow impact index and the passenger flow threshold.

[0067] Passenger flow threshold refers to the threshold used to determine whether excessive passenger flow has occurred. This threshold is set in advance by those skilled in the art and will not be elaborated upon here.

[0068] The excess value refers to the relative proportion by which the passenger flow impact index exceeds the passenger flow threshold. It is calculated using the formula: Excess Value = (Passenger Flow Impact Index - Passenger Flow Threshold) / Passenger Flow Threshold.

[0069] When the passenger flow impact index exceeds the passenger flow threshold, it indicates that the current passenger flow is too large. The excess value needs to be calculated first for subsequent steps.

[0070] S33: Determine the inventory adjustment ratio based on the excess value.

[0071] Inventory increase ratio refers to the percentage increase planned for the material inventory threshold in response to anticipated customer traffic growth.

[0072] The inventory adjustment percentage is obtained by consulting a pre-defined range adjustment comparison table. This table records the inventory adjustment percentages corresponding to different ranges of exceeding the range. The range adjustment comparison table is pre-formed by those skilled in the art by associating and setting different inventory adjustment percentages corresponding to different inventory adjustment percentages, and will not be elaborated here.

[0073] S34: Adjust the material inventory threshold based on the inventory increase ratio.

[0074] The current material inventory threshold is multiplied by the inventory increase ratio to raise the material inventory threshold.

[0075] It also includes stress response methods: S40: When the passenger flow impact index exceeds the preset stress threshold, collect the real-time consumption rate of materials, current order data, and booked order data.

[0076] The pressure threshold refers to a second-level benchmark comparison value that is higher than the passenger flow threshold. When the passenger flow impact index exceeds this value, the system determines that the current passenger flow pressure has reached a high level and needs to collect real-time material consumption rates, current order data, and reserved order data for subsequent steps. The pressure threshold is preset by those skilled in the art and will not be elaborated here.

[0077] Real-time consumption rate refers to the actual amount of materials consumed per unit of time (e.g., per hour). This data is obtained by connecting to the inventory change data of the kitchen management system's outbound records in real time and calculating the consumption per unit of time.

[0078] Current order data refers to the material requirements details for orders placed but not yet completed in the restaurant's point-of-sale (POS) system. This data is obtained by calling the POS system's real-time order interface.

[0079] Reservation order data refers to the estimated material requirements included in orders entered into the reservation system for in-store consumption within a specific future time period. This data is obtained by calling the data interface of the reservation management system. The specific future time period is preset by those skilled in the art and will not be elaborated here.

[0080] S41: Calculate the estimated consumption rate based on current order data and booked order data.

[0081] The estimated consumption rate refers to the rate at which materials will be consumed within a future unit of time (e.g., one hour) based on known order demand. This is achieved by analyzing current and reserved order data to obtain the corresponding material demand, allocating these demand quantities to a future timeline according to the order's consumption time, and then calculating the total demand for each material within the future unit of time window. This total demand is defined as the estimated consumption rate within that time window. The specific conversion and statistical algorithms are standard data processing methods in this field and will not be elaborated upon here.

[0082] S42: The real-time consumption rate is superimposed with the estimated consumption rate to obtain the comprehensive consumption rate.

[0083] The overall consumption rate is a comprehensive rate indicator that combines real-time consumption with future known order consumption, reflecting the overall consumption trend of materials over a period of time.

[0084] The overall consumption rate is obtained by adding the real-time consumption rate to the estimated consumption rate.

[0085] S43: Obtain the material's sustainability based on the overall consumption rate and real-time inventory data.

[0086] Material sustainability refers to the theoretical length of time that an inventory can sustain supply, based on the current real-time inventory level, assuming the current overall consumption rate remains constant. It is calculated using the formula: Material sustainability = Real-time inventory data / Overall consumption rate.

[0087] S44: When the material can be maintained for no more than the preset second warning duration, the replenishment urgency coefficient is determined based on the material's maintainable duration and the preset safety redundancy duration.

[0088] The second warning duration refers to a preset time threshold used to determine whether inventory is in a state of urgent shortage. When the material's shelf life is less than or equal to this value, the replenishment urgency factor must be determined for subsequent steps. This threshold is preset by those skilled in the art and will not be elaborated upon here.

[0089] Safety redundancy time refers to a preset buffer period reserved to cope with uncertainties, used to provide a benchmark when calculating replenishment urgency. This time is set by those skilled in the art based on historical data of supplier delivery reliability, and will not be elaborated here.

[0090] The replenishment urgency factor is an indicator used to quantify the urgency of replenishment. It is calculated using the formula: Replenishment Urgency Factor = (Safety Redundancy Duration - Material Sustainability Duration) / Safety Redundancy Duration. The higher the value of this factor, the greater the urgency.

[0091] S45: Generate expedited replenishment orders based on replenishment urgency.

[0092] An expedited replenishment order is a structured data message containing information about an urgent replenishment request. Expedited replenishment orders are generated by querying a pre-defined expedited order lookup table. This table records expedited replenishment orders corresponding to different replenishment urgency levels. The lookup content in the expedited order lookup table was created by those skilled in the art by recording the expedited replenishment orders corresponding to different replenishment urgency levels, and will not be elaborated upon here.

[0093] S46: When the shelf life of a material falls within the preset inventory synchronization range, an inventory synchronization prompt will be reported.

[0094] The inventory synchronization range refers to a preset range of how long a material can be maintained. The upper and lower limits of this range are pre-set by those skilled in the art based on management requirements and risk tolerance. Specifically, the upper limit is higher than the second warning duration in S44, indicating that while the inventory has not reached an emergency state, it has entered a warning range requiring attention; the lower limit is equal to the second warning duration, serving as a transition boundary from warning attention to emergency response. The specific numerical range of this range will not be elaborated here. When the material's maintainable duration falls within this range, it indicates that the inventory level warrants attention, and the system will automatically report an inventory synchronization alert to remind relevant personnel to synchronize information with suppliers in advance.

[0095] It also includes internal adjustment methods: S50: When the material can be maintained for no longer than the second warning period, the remaining inventory of the material is determined based on real-time inventory data, and the capacity load data of each workstation in the back kitchen is collected.

[0096] Material remaining inventory refers to the real-time available quantity of each material recorded in the system at the current moment. This data is obtained by reading the inventory quantity of the corresponding material from the real-time inventory data collected in S10.

[0097] Capacity load data refers to quantitative information reflecting the current workload and available remaining capacity of each processing station in the kitchen (such as wok stoves, steamers, and cutting stations). This data is obtained by calling the real-time status monitoring interface of the kitchen management system. The real-time status monitoring interface records quantitative information on the current workload and available remaining capacity of each processing station in the kitchen.

[0098] When the material's remaining inventory duration is no longer than the second warning duration, the remaining inventory of the material must be determined first, and then capacity load data should be collected for subsequent steps.

[0099] S51: Determine the inventory-related dishes based on the remaining material inventory and the preset daily menu.

[0100] The daily menu list refers to the set of identifiers for all dishes that the restaurant plans to serve that day. This list is obtained by reading the pre-set daily menu plan from the restaurant's operating system.

[0101] Inventory-related dishes refer to dishes in the daily menu whose standard recipes include scarce materials (i.e., materials corresponding to the remaining inventory of the materials).

[0102] By querying a pre-set recipe database, all materials required for each dish in the daily menu can be obtained. The required materials for each dish are compared with scarce materials (i.e., materials corresponding to the remaining inventory in S50). Dishes whose required materials include the scarce material are identified as inventory-related dishes. The recipe database is pre-set by those skilled in the art and will not be elaborated upon here.

[0103] S52: Determine alternative dishes based on inventory-related dishes.

[0104] Substitute dishes refer to dishes in the daily menu that are similar in taste and category to inventory-related dishes, and whose standard recipes do not contain scarce ingredients. Substitute dishes are obtained by querying a pre-defined menu substitution relationship database. This database records different substitute dishes corresponding to different inventory-related dishes. The menu substitution relationship database is pre-established by those skilled in the art by associating different substitute dishes corresponding to different inventory-related dishes, and will not be elaborated upon here.

[0105] S53: Combine alternative dishes and capacity load data to screen out feasible alternative dishes.

[0106] Executable substitute dishes refer to those selected from the list of substitute dishes whose main processing stations still have sufficient capacity to process them under the current capacity load data. The selection process involves querying a pre-defined standard process library for dishes to obtain the main processing stations and standard working hours required for each substitute dish, and comparing this information with the remaining available capacity of the corresponding stations in the capacity load data. If the remaining available capacity of all main processing stations required for a substitute dish can meet its standard working hour requirements, then that dish is selected as an executable substitute dish. The standard process library records the main processing stations and standard working hours for each dish, which are pre-set by those skilled in the art and will not be elaborated upon here.

[0107] S54: Generate internal adjustment schemes based on executable alternative dishes.

[0108] An internal adjustment plan is a structured set of operational instructions that specifies which executable replacement dishes should be used during periods of inventory shortage, their priority order for replacing related dishes in the original inventory, and includes corresponding kitchen task guidance. The specific method for generating the internal adjustment plan will be explained in detail in subsequent sections S60 to S63, and will not be repeated here.

[0109] S55: Calculate the net material demand gap based on the internal adjustment plan, and revise the expedited replenishment order based on the net material demand gap.

[0110] The net material demand gap refers to the quantity of specific materials that still need to be procured externally to meet expected operational needs, calculated after the estimated reduction in material consumption from the application of internal adjustment schemes (i.e., partially replacing the original related dishes with executable alternative dishes).

[0111] The system directly reads the internal adjustment plan generated by S54, which explicitly includes the material consumption adjustment amount corresponding to each replacement suggestion. The system performs routine numerical calculations, subtracting the sum of the adjustment amounts of all replacement suggestions in the plan from the original total material demand; the result is the net material demand gap.

[0112] The obtained net material demand gap is used to adjust the purchase quantity requirements of the expedited replenishment order.

[0113] It also includes a method for generating internal adjustment schemes: S60: Determine the dish matching degree based on the replaceable dishes and the daily menu.

[0114] Dish matching degree refers to a numerical indicator used to quantify the substitutability between a replaceable dish and an original, related dish in the inventory. This indicator is obtained by querying a pre-set replacement matching library. This matching library records the matching degree scores between different dishes, which are pre-set by those skilled in the art based on the similarity of the dishes in terms of flavor, ingredients, and processing methods, and will not be elaborated here.

[0115] S61: Determine the load increment based on capacity load data and alternative dishes.

[0116] Load increment refers to the additional workload that is expected to be generated at the relevant processing stations if a certain alternative dish is prepared, based on the current capacity load of the kitchen.

[0117] This data is calculated by matching the standard preparation time data of replaceable dishes with the remaining available capacity of the corresponding workstations in the capacity load data collected by the S50. The specific matching calculation is a conventional data processing method for those skilled in the art and will not be elaborated here.

[0118] S62: Combine the dish matching degree and load increment to comprehensively score and rank each replaceable dish to obtain the ranking result.

[0119] The sorting result refers to the ordered list generated after prioritizing the list of replaceable dishes. This result is obtained through the system's decision module, which calculates a comprehensive score for each replaceable dish based on two dimensions: dish matching degree (higher is better) and load increment (lower is better), according to a preset weighted scoring rule. The dishes are then sorted from highest to lowest score. The specific weighting rules and scoring algorithms are conventional techniques set by those skilled in the art based on optimization objectives and will not be elaborated upon here.

[0120] S63: Generate an internal reconciliation plan based on the sorting results and remaining material inventory.

[0121] By reading the sorting results and starting with the top-ranked dishes, and combining the remaining inventory of materials, the proportion of each dish that can be substituted is calculated until the estimated total consumption that can be substituted matches the inventory shortage situation, thereby generating a structured solution that includes specific substitute dishes, suggested replacement order and quantity.

[0122] It also includes heat perception and pre-adjustment methods: S70: collect local catering topic data.

[0123] Local restaurant-related data refers to publicly available posts, discussions, and interactions related to the city where the restaurant is located and its food category, scraped from social media platforms. This data is obtained by calling the public data interfaces (APIs) of the relevant social media platforms and using preset keyword combinations for targeted collection. The specific keyword combinations are pre-defined by those skilled in the art and will not be elaborated here.

[0124] S71: Extract volume growth rate, sentiment data, and local user engagement data from local food and beverage topic data.

[0125] The volume growth rate refers to the rate at which the discussion intensity of a specific topic increases within a unit of time (such as 24 hours). This data is obtained by statistically analyzing the volume indicators of a topic at different time points and calculating its rate of change.

[0126] Sentiment data refers to the proportion of text expressing positive, neutral, and negative sentiment within a topic discussion. This data is obtained by processing the collected text content using sentiment analysis algorithms. Sentiment analysis algorithms are common knowledge in this field and will not be elaborated upon here.

[0127] Local user engagement data refers to the proportion of users participating in topic discussions whose geographical location information is displayed as the city where the restaurant is located. This data is obtained by analyzing and statistically processing the location information in users' publicly available profiles.

[0128] S72: Combine volume growth rate, sentiment data, and local user engagement data to calculate topic relevance.

[0129] Topic relevance is a comprehensive quantification of the impact of local food and beverage topics on potential restaurant demand. It is calculated using a pre-defined weighted fusion algorithm, combining data on volume growth rate, sentiment, and local user participation extracted by S71. The specific weighted fusion algorithm is determined by those skilled in the art based on the objectives and will not be elaborated upon here.

[0130] S73: Determine the pre-adjustment coefficient based on the relevance of the topic.

[0131] The pre-adjustment coefficient is a parameter used to dynamically adjust the weight of the passenger flow impact index in the real-time control method.

[0132] The pre-adjustment coefficients are determined by querying a pre-defined topic coefficient mapping table. This table records the pre-adjustment coefficients corresponding to different topic relevance ranges. The topic coefficient mapping table is set in advance by those skilled in the art based on the data distribution patterns and weight adjustment requirements of topic relevance, and will not be elaborated here.

[0133] S74: Based on the pre-adjustment coefficient, reduce the fusion weight of pedestrian density at intersections in the calculation of the passenger flow impact index, and recalculate the passenger flow impact index according to the reduced fusion weight.

[0134] This step reduces the value of the original weight (e.g., w1) related to "intersection pedestrian density" in the S31 passenger flow impact index calculation formula by multiplying it with a pre-adjustment coefficient, and then re-executes the S31 calculation process using the adjusted new weight combination. The specific calculation is: New weight = Original weight × (1 - Pre-adjustment coefficient).

[0135] S75: Update material inventory thresholds based on the recalculated passenger flow impact index.

[0136] This step is completed by repeating the process from S32 to S34, using the recalculated passenger flow impact index from S74 as the new input. Specifically, it determines whether the passenger flow / pressure threshold is exceeded based on the new passenger flow impact index, calculates the possible extent of the exceedance and the corresponding inventory adjustment ratio, and finally updates the material inventory threshold for the affected materials.

[0137] It also includes a hot topic cycle procurement method: S80: When the topic relevance exceeds the preset high relevance threshold, the demand growth coefficient is determined based on the topic relevance.

[0138] The high relevance threshold refers to a preset numerical limit used to determine whether the relevance of a topic has reached a level sufficient to trigger a special procurement plan. This threshold is set in advance by those skilled in the art and will not be elaborated upon here.

[0139] The demand growth coefficient is a numerical parameter used to quantify the expected additional increase in demand resulting from the popularity of a topic. It is obtained by querying a pre-defined topic coefficient mapping table. This table records the demand growth coefficients corresponding to different ranges of topic relevance, which are pre-set by those skilled in the art and will not be elaborated upon here.

[0140] S81: Combine demand growth factor, real-time inventory data and material inventory threshold to calculate additional material purchase quantity.

[0141] Additional material procurement quantity refers to the quantity of materials that should be procured outside of the regular procurement plan to cope with potential demand increases due to the popularity of a topic. It is calculated using the formula: Additional Material Procurement Quantity = Max(0, Demand Growth Coefficient × Material Inventory Threshold - (Current Actual Inventory Quantity - Material Inventory Threshold)).

[0142] S82: Generate incremental purchase orders based on additional material purchase quantities.

[0143] An incremental procurement instruction is a structured data message containing additional procurement demand information triggered by trending topics. Its content includes at least the target material identifier and the corresponding additional procurement quantity. This instruction is generated by calling a pre-set instruction generation template and filling in the specific material identifier and additional procurement quantity data. The instruction generation template is pre-set by those skilled in the art and will not be elaborated upon here.

[0144] S83: The incremental purchase order is superimposed with the material purchase quantity to generate the total purchase order.

[0145] The master procurement order is the final procurement order that integrates regular procurement needs and additional procurement needs related to trending topics. It is generated by adding the additional procurement quantity of materials specified in the incremental procurement order generated in S82 to the corresponding procurement quantity of materials in the material procurement quantity generated in S16, thereby updating the procurement quantity and forming the merged master procurement order.

[0146] It also includes a method for monitoring the decline in popularity: S90: After generating an incremental purchase order, collect subsequent data on catering topics.

[0147] Subsequent food and beverage topic data refers to data related to the same food and beverage topic continuously collected from social media platforms within the time period following the generation of the incremental purchase order. The collection method is the same as that of the S70 and will not be elaborated upon here.

[0148] S91: Update topic relevance based on subsequent catering topic data.

[0149] This step is completed by repeatedly executing the calculation process of S71 and S72 using newly collected subsequent catering topic data, thereby obtaining the topic relevance that reflects the latest popularity status.

[0150] S92: When the updated topic relevance is lower than the high relevance threshold and shows a continuous downward trend, collect the downward trend data.

[0151] Downtrend data refers to indicators used to quantify the rate and persistence of topic relevance decay, including the magnitude of the decrease in topic relevance per unit time and the duration of continuous decline. This data is obtained by performing time series analysis (such as linear regression to calculate the slope) on topic relevance sequences updated in the most recent few periods. The specific analysis methods are conventional techniques in this field and will not be elaborated here.

[0152] When the updated topic relevance is lower than the high relevance threshold and shows a continuous downward trend, it is necessary to collect the downward trend data first for subsequent steps.

[0153] S93: Determine the demand decay coefficient based on the downward trend data.

[0154] The demand decay coefficient is a parameter used to quantify the expected magnitude of demand decay.

[0155] The demand decay coefficient is obtained by querying a pre-defined decline coefficient mapping table. This table records the demand decay coefficients corresponding to different ranges of declining data trends, which are preset by those skilled in the art and will not be elaborated upon here.

[0156] S94: Calculate the reduction in procurement quantity based on the demand decay coefficient and the additional procurement quantity of materials.

[0157] Reduced procurement volume refers to the quantity of materials that should be cut from the planned additional procurement volume due to the waning popularity of the topic. It is calculated by multiplying the demand decay factor by the additional procurement volume of materials calculated by S81.

[0158] S95: Generate a purchase reduction order based on the reduction in purchase volume, and update the general purchase order based on the purchase reduction order.

[0159] A procurement reduction order is a structured data message that contains information about the need to reduce procurement demand due to the decline in popularity of a topic. Its content includes at least the target material identifier and the corresponding reduction in procurement quantity.

[0160] The method for generating procurement reduction orders is the same as that of S82, and will not be elaborated here.

[0161] The reduced purchase quantity in the purchase reduction order is subtracted from the corresponding purchase quantity of the material in the current total purchase order to update the purchase quantity and form the final total purchase order.

[0162] Based on the same inventive concept, embodiments of the present invention provide an AI-based prediction-based inventory threshold and purchase quantity calculation and processing system, comprising: a data acquisition module for acquiring historical data, real-time inventory data, external parameter data, material consumption data, intersection pedestrian density, parking lot vacancy rate, competitor restaurant queue length, real-time consumption rate, current order data, reserved order data, capacity load data, local catering topic data, subsequent catering topic data, and downward trend data; a memory for storing a program that implements an AI-based prediction-based inventory threshold and purchase quantity calculation and processing method; and a processor for loading and executing the program stored in the memory.

[0163] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0164] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for calculating and processing inventory thresholds and purchase quantities based on AI prediction, characterized in that, include: Collect historical data, real-time inventory data, and external parameter data; Historical data is preprocessed to form training and input datasets; The long short-term memory network model is trained using the training dataset to obtain the prediction model; Input the dataset into the prediction model for analysis and forecasting to determine future demand trends; Material inventory thresholds are determined based on future demand trends and pre-defined industry characteristic parameters; The basic purchase quantity is determined based on future demand trends, real-time inventory data, and external parameter data. The material purchase quantity is generated based on external parameter data, material inventory thresholds, and basic purchase quantity.

2. The method for calculating and processing inventory thresholds and purchase quantities based on AI prediction according to claim 1, characterized in that, It also includes model optimization methods: collecting material consumption data within a preset detection period; calculating the prediction deviation value based on the material consumption data and future demand trends; generating optimized model parameters based on the prediction deviation value when the prediction deviation value exceeds a preset deviation threshold; and updating the prediction model based on the optimized model parameters.

3. The method for calculating and processing inventory thresholds and purchase quantities based on AI prediction according to claim 1, characterized in that, It also includes real-time control methods: collecting pedestrian density at intersections, parking lot vacancy rates, and queue lengths at competing restaurants; combining pedestrian density at intersections, parking lot vacancy rates, and queue lengths at competing restaurants to calculate the passenger flow impact index; when the passenger flow impact index exceeds a preset passenger flow threshold, calculating the excess value based on the passenger flow impact index and the passenger flow threshold. The inventory adjustment ratio is determined based on the value exceeding the range. The material inventory threshold is increased based on the inventory increase ratio.

4. The method for calculating and processing inventory thresholds and purchase quantities based on AI prediction according to claim 3, characterized in that, It also includes stress response methods: when the passenger flow impact index exceeds the preset stress threshold, it collects the real-time consumption rate of materials, current order data, and booked order data; The estimated consumption rate is calculated based on current order data and booked order data; the real-time consumption rate is then superimposed with the estimated consumption rate to obtain the comprehensive consumption rate. The material's sustainability is determined based on the overall consumption rate and real-time inventory data. When the material's sustainability is no greater than the preset second warning duration, the replenishment urgency coefficient is determined based on the material's sustainability and the preset safety redundancy duration; an expedited replenishment instruction is generated based on the replenishment urgency coefficient; when the material's sustainability falls within the preset inventory synchronization range, an inventory synchronization alert is reported.

5. The method for calculating and processing inventory thresholds and purchase quantities based on AI prediction according to claim 4, characterized in that, It also includes internal adjustment methods: when the material can be maintained for no more than the second warning period, the remaining inventory of the material is determined based on real-time inventory data, and the capacity load data of each workstation in the back kitchen is collected; Determine the inventory-related dishes based on the remaining material inventory and the pre-set daily menu; Identify alternative dishes based on inventory-related dishes; Combine alternative dishes and capacity load data to screen out feasible alternative dishes; An internal adjustment plan is generated based on executable alternative dishes; the net material demand gap is calculated based on the internal adjustment plan; and the expedited replenishment order is revised based on the net material demand gap.

6. The method for calculating and processing inventory thresholds and purchase quantities based on AI prediction according to claim 5, characterized in that, It also includes a method for generating internal adjustment plans: determining the matching degree of dishes based on the list of replaceable dishes and the dishes of the day; determining the load increment based on the capacity load data and replaceable dishes; combining the matching degree of dishes and the load increment to comprehensively score and rank each replaceable dish to obtain the ranking result; and generating an internal adjustment plan based on the ranking result and the remaining inventory of materials.

7. The method for calculating and processing inventory thresholds and purchase quantities based on AI prediction according to claim 3, characterized in that, It also includes a heat perception and pre-adjustment method: collecting local catering topic data; extracting volume growth rate, sentiment data, and local user participation data from the local catering topic data; combining the volume growth rate, sentiment data, and local user participation data to calculate the topic relevance; determining the pre-adjustment coefficient based on the topic relevance; reducing the fusion weight of intersection personnel density in the calculation of the passenger flow impact index based on the pre-adjustment coefficient, and recalculating the passenger flow impact index based on the reduced fusion weight; and updating the material inventory threshold based on the recalculated passenger flow impact index.

8. The method for calculating and processing inventory thresholds and purchase quantities based on AI prediction according to claim 7, characterized in that, It also includes a trending period procurement method: when the topic relevance exceeds a preset high relevance threshold, the demand growth coefficient is determined based on the topic relevance; The additional material purchase quantity is calculated by combining the demand growth coefficient, real-time inventory data, and material inventory threshold. Incremental purchase orders are generated based on the additional purchase quantity of materials; the incremental purchase orders are then combined with the material purchase quantity to generate a total purchase order.

9. The method for calculating and processing inventory thresholds and purchase quantities based on AI prediction according to claim 8, characterized in that, It also includes a method for monitoring the decline in popularity: after generating an incremental procurement order, collect subsequent catering topic data; update the topic relevance based on the subsequent catering topic data; when the updated topic relevance is lower than the high relevance threshold and shows a continuous downward trend, collect the downward trend data. Determine the demand decay coefficient based on the downward trend data; The reduction in procurement volume is calculated based on the demand decay coefficient and the additional procurement volume of materials. Generate a purchase reduction order based on the reduction in purchase volume, and update the master purchase order based on the purchase reduction order.

10. An AI-based prediction-based inventory threshold and purchase quantity calculation and processing system, characterized in that, include: The data acquisition module is used to collect historical data, real-time inventory data, and external parameter data. A memory for storing a program that implements the AI-based prediction method for calculating inventory thresholds and purchase quantities as described in any one of claims 1 to 9; The processor is used to load and execute programs stored in memory.

Images