Food storage management system

Abstract

The invention relates to the technical field of food storage management, in particular to a food storage management system, which comprises a data acquisition layer, is composed of a plurality of devices and is used for acquiring food, environment and user operation data; the data transmission layer transmits data by using wired and wireless network technologies; the data storage layer stores various types of data in combination with a relational database and a non-relational database; the data processing layer comprises a plurality of functional modules, such as an intelligent warehousing management module, and is used for realizing automatic input verification and batch management of food information; the intelligent ex-warehouse management module is used for optimizing ex-warehouse sequence and tracking food directions; the expiration and waste management module is used for carrying out expiration reminding and surplus value evaluation; the data analysis and purchase optimization module is used for predicting demands and analyzing loss; and the dish and inventory association module is used for matching the dish formula and the inventory and predicting the sales allocation inventory. The system improves the management efficiency, reduces the waste, optimizes the purchase decision, enhances the dish and inventory association management, guarantees the system safety, and provides data support to assist the decision.

Description

Technical Field

[0001] The invention relates to the technical field of food storage management, in particular to a food storage management system. Background Art

[0002] Food storage management plays a vital role in the food industry, and is directly related to food quality and safety, the operating costs of enterprises, and the health of consumers. However, the common food storage management methods currently on the market have many problems that need to be solved urgently.

[0003] Traditional food storage management often relies on manual operation and simple recording methods. When food is put into storage, food information is mainly recorded manually. This method is inefficient and prone to errors. For example, information entry errors and blurred handwriting may occur, resulting in inaccurate and incomplete food information, affecting subsequent inventory management and quality traceability. In the inventory management process, due to the lack of effective information system support, it is difficult to grasp the key information such as food inventory quantity, batch, shelf life, etc. in real time, which is easy to cause inventory backlog or out of stock. For food outbound management, it is usually impossible to achieve accurate control of first-in-first-out, resulting in some food being expired and wasted due to backlog, increasing corporate costs. At the same time, traditional methods have obvious deficiencies in food quality monitoring. It is difficult to monitor and control the environmental conditions (such as temperature and humidity) of food during storage in real time and accurately, and it is impossible to timely discover and deal with changes in environmental factors that may affect food quality, increasing the risk of food deterioration.

[0004] Although some existing food management systems have tried to solve these problems to a certain extent, there are still many shortcomings. Although some systems have realized basic inventory management functions, they are still lacking in the accuracy and comprehensiveness of food information collection, the depth and intelligence of data processing, and the coordination with other business links (such as procurement and food preparation). For example, in terms of food information traceability, it may not be possible to provide a complete and tamper-proof traceability chain, making it difficult to meet the strict requirements of enterprises and regulatory authorities for food safety traceability; in terms of procurement optimization, there is a lack of accurate demand forecasting models and optimization algorithms that take into account multiple factors, resulting in procurement decisions that are not scientific and reasonable; in the management of the relationship between dishes and inventory, efficient recipe matching and intelligent inventory allocation have not been achieved, and the refined management needs of catering companies cannot be effectively supported.

[0005] Therefore, a food storage management system is proposed in response to the above problems. Summary of the invention

[0006] The object of the present invention is to provide a food storage management system to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A food storage management system, comprising:

[0009] Data collection layer: The data collection layer consists of temperature sensors, humidity sensors, RFID readers, barcode scanners, IoT devices, and user operation terminals. The data collection layer is used to collect food information, environmental data, and user operation data.

[0010] Data storage layer: The data transmission layer uses wired network and wireless network communication technology to transmit the collected data to the data storage layer and data processing layer;

[0011] Data storage layer: The data storage layer uses a combination of relational databases and non-relational databases to store food inventory information, sales data, purchase data, environmental data, user information, and log data;

[0012] Data processing layer: The data processing layer includes intelligent warehousing management module, intelligent outbound management module, expiration and waste management module, data analysis and procurement optimization module, dish and inventory association module, and user authority and behavior tracking module. It conducts in-depth analysis, processing and mining of data stored in the database, providing data support and decision-making basis for the application display layer;

[0013] Application display layer: The application display layer displays system functions and data results to users through Web applications and mobile apps. The application display layer includes inventory management interface, environmental monitoring interface, procurement management interface, data analysis reports, and early warning information display.

[0014] As a preferred solution, the intelligent warehousing management module in the data processing layer includes:

[0015] Automatic data entry and verification unit: integrated with automated scanning equipment, scans food barcodes or QR codes to enter information and conduct multiple verifications, including connecting with the food quality traceability database to verify the legality of food sources and quality compliance;

[0016] Batch management unit: Blockchain technology is used to encrypt and store food production batch information and conduct full traceability management. The detailed status information of each batch of food is displayed on the inventory management interface. The blockchain technology in the batch management unit records detailed information such as food production batch number, production date, production location, source of raw materials, etc., and forms an unalterable traceability chain. In the blockchain, the hash value calculation of the data block adopts the SHA-256 algorithm. For a data block , its hash value ,in, Represents a data block, Represents a data block The hash value of the previous data block is combined with the current data block information to generate a new data block hash value to ensure the integrity and non-tamperability of the data.

[0017] As a preferred solution, the intelligent outbound management module in the data processing layer includes:

[0018] First-in-first-out algorithm optimization unit: Based on the food time label and inventory distribution, an intelligent algorithm is used to combine the warehouse shelf layout and cargo handling path to calculate the optimal outbound order. Suppose the food inventory set is , each food element Including storage time Shelf location and volume , sorted in ascending order according to the time of entry. For foods with the same entry time, sorted according to the distance from the shelf location to the delivery port and handling efficiency coefficient Calculating priority values ,in, Indicates the priority value. Indicates the distance adjustment coefficient, and determines the order of outbound delivery according to the priority value from small to large;

[0019] Real-time tracking unit: Uses IoT technology and GPS positioning to record food delivery information, connects with the catering enterprise information system to track food usage information, and generates dynamic food consumption reports;

[0020] The real-time tracking unit also includes monitoring of the transportation link after leaving the warehouse and detailed recording and analysis of the association between food and dishes in catering enterprises. The Kalman filter algorithm is used to estimate the position and state of the transport vehicle in the transportation link monitoring. The system state vector ,in, Indicates location, Represents speed, state transfer matrix ,in, represents the time interval, the observation matrix , then the state prediction formula is: ,in, express The prior state estimate at time t, express The posterior state estimate at the moment; the covariance prediction formula is: ,in, express The prior covariance estimate of time, express The posterior covariance estimate at time , represents the process noise covariance matrix; the observation update formula is: ,in, represents the Kalman gain, represents the observation noise covariance matrix, where Represents the observation matrix; the covariance update formula is: .

[0021] As a preferred solution, the expiration and waste management module in the data processing layer includes:

[0022] Dynamic expiration reminder unit: Combine the time priority algorithm with the food attribute analysis model, and send dynamic expiration reminder information to managers in the form of system pop-up windows, SMS notifications, and email reminders based on the food production date, shelf life, inventory quantity, and expiration risk weight. The calculation formula for the expiration risk index in the time priority algorithm is: ,in, Indicates Expiration risk index of food, Indicates the expiration date of food. Indicates the current time. Indicates the shelf life. represents the weight factor related to food attributes;

[0023] Residual value evaluation unit: Through deep learning algorithms, it comprehensively considers multi-dimensional factors including the remaining shelf life of food, market demand conditions, historical data on sales price fluctuations, brand influence, and changes in consumer preferences, and formulates personalized secondary utilization recommendations for foods that are about to expire; the residual value evaluation unit formulates strategies including promotion, donation, or adjustment of packaging specifications for different foods based on the evaluation results. Among them, the residual value evaluation adopts a neural network based on a multi-layer perceptron MLP. The input layer neurons correspond to the remaining shelf life of food, market demand index, and brand influence scoring factors, and the hidden layer adopts a Sigmoid activation function: , the output layer is the food surplus value score, and the weights and thresholds are determined by training the neural network with a large amount of sample data.

[0024] As a preferred solution, the data analysis and procurement optimization module in the data processing layer includes:

[0025] Procurement suggestion algorithm unit: Based on the demand forecasting model built with big data analysis and deep learning technology, the optimization algorithm combining linear programming and dynamic programming is used to generate a procurement plan, which includes the types and quantities of food to be purchased, the procurement time window and the recommended supplier list. The demand forecasting model uses the random forest algorithm combined with the gradient boosting decision tree GBDT algorithm. The random forest determines the prediction result through voting of multiple decision trees. The information gain calculation in the decision tree construction process is as follows: ,in, represents a data set, Represents feature attributes, Representation attributes The value set of Represents information baking function: , Indicates that the data set belongs to The GBDT algorithm performs iterative optimization by continuously fitting the residuals of the previous round of models. When optimizing the procurement plan, the procurement cost is set to , the inventory holding cost is ®, and the stockout cost is , the purchase volume is , the demand forecast value is , then the objective function is: , until the constraints are met ,in, express Inventory levels at any given moment, express Purchase quantity at any time, express Demand forecast value at each moment;

[0026] Loss analysis unit: Use data mining and statistical analysis techniques to mine multi-source data, identify food loss points and causes in storage, transportation, and use, establish correlation models, and propose optimization measures. The loss analysis unit analyzes the temperature, humidity, and other conditions in the storage link, the routes and methods in the transportation link, and the production process in the use link of catering enterprises. A multivariate linear regression model is used in the loss analysis of the storage link: ,in, Indicates the amount of food loss, Indicates factors including temperature and humidity. represents the regression coefficient, , represents the random error term;

[0027] Decision support unit: Generate multi-dimensional visual data reports and interactive charts through data visualization technology, support data drilling and multi-dimensional analysis functions, and assist management decision-making. The reports and charts of the decision support unit cover inventory turnover rate, food storage cycle, procurement cost analysis, loss rate, and are displayed in multiple dimensions.

[0028] As a preferred solution, the dish and inventory association module in the data processing layer includes:

[0029] Intelligent recipe matching unit, establishes a recipe database, matches raw material inventory information according to dish production requirements, calculates inventory consumption and generates a stock-out list and sends it to the purchasing department. Need raw material collection , the raw material inventory set is , for each raw material ,The formula for calculating inventory consumption is: ,in, Indicates the inventory consumption, Indicates dishes Medium Raw Materials The dosage, Indicates dishes The estimated production volume and the out-of-stock judgment formula are: ;

[0030] The forecasting and allocation unit uses a combination of time series analysis algorithms and machine learning models to predict the sales volume of dishes, and uses an inventory allocation optimization algorithm to reasonably allocate the inventory of commonly used raw materials based on the forecast results. When forecasting dish sales, the forecasting and allocation unit comprehensively considers historical sales trends, seasonal factors, marketing promotion effects, and changes in surrounding consumer groups. The seasonal autoregressive moving average SARIMA model is used to predict the sales volume of dishes, and its formula is: ,in, represents the backshift operator, represents nonseasonal autoregressive and moving average polynomials, represents seasonal autoregressive and moving average polynomials, Represents the seasonal cycle, represents the difference order, Indicates that time series data is The value of the moment, Represents a white noise sequence; inventory allocation optimization uses a genetic algorithm to encode the raw material allocation plan into chromosomes, and continuously optimizes chromosomes through genetic operations such as selection, crossover, and mutation. The fitness function is determined based on inventory cost and out-of-stock risk.

[0031] As a preferred solution, the user rights and behavior tracking module in the data processing layer includes:

[0032] The authority classification unit builds a multi-level user authority system, grants corresponding operation permissions according to different job responsibilities, and uses biometric technology combined with dynamic password authentication for user identity authentication;

[0033] The behavior tracking unit records user operation behaviors in real time through the log recording function, and uses the abnormal behavior detection algorithm to monitor and analyze user operation behaviors in real time; the abnormal behavior detection algorithm builds a normal user operation behavior model based on user behavior pattern learning and data statistical analysis technology and monitors anomalies accordingly, and uses the hidden Markov model HMM to model the user operation behavior sequence. Suppose the hidden state set is , the set of observed states is , state transition probability matrix ,in, , indicating that from the state Transfer to state The probability of observation probability matrix ,in, , indicating that in the state Observed below The probability of the initial state probability vector ,in, , indicating that the initial state is The probability of HMM model parameters is trained by Baum-Welch algorithm, and then the Viterbi algorithm is used to determine the most likely hidden state sequence according to the observation sequence to determine whether there is abnormal behavior.

[0034] It can be seen from the technical solution provided by the present invention that the food storage management system provided by the present invention has the following beneficial effects:

[0035] 1. Improve food management efficiency:

[0036] Accurate inventory management: The intelligent warehousing management module ensures accurate and fast entry of food information through automated scanning equipment and multiple verification mechanisms, and uses blockchain technology to achieve encrypted storage and full traceability management of batch information, so that each batch of food has a detailed status display on the inventory management interface; this not only improves the efficiency of warehousing operations, but also ensures the accuracy and traceability of inventory information, which helps companies accurately grasp inventory dynamics, adjust inventory strategies in a timely manner, and reduce inventory backlogs or out-of-stock phenomena;

[0037] Optimize the outbound process: The intelligent outbound management module optimizes the unit based on the first-in-first-out algorithm and calculates the optimal outbound sequence based on the actual warehouse situation, thereby reducing the time that food is in the warehouse and reducing the risk of expiration. The real-time tracking unit uses the Internet of Things and GPS technology to record the whereabouts and usage information of food after outbound delivery in detail, and generates dynamic consumption reports, which facilitates enterprises to monitor the flow of food in real time, ensure the transparency and traceability of the supply chain, and improve the efficiency and management level of the entire outbound process.

[0038] 2. Effectively reduce food waste:

[0039] Accurate expiration reminder: The dynamic expiration reminder unit of the expiration and waste management module uses the time priority algorithm and food attribute analysis model to comprehensively consider multiple factors to calculate the expiration risk index, and sends dynamic reminder information to managers in a timely manner, so that managers can know and deal with food that is about to expire in advance, avoiding food expiration and waste due to negligence;

[0040] Maximizing the use of residual value: The residual value assessment unit uses a deep learning algorithm to comprehensively assess the residual value of food that is about to expire, and formulates personalized secondary utilization suggestions based on multi-dimensional factors, such as promotion, donation, or adjustment of packaging specifications, etc., providing companies with scientific and reasonable processing strategies, which helps to minimize food waste, improve resource utilization, and reduce corporate costs to a certain extent;

[0041] 3. Optimize purchasing decisions and cost control:

[0042] Accurate demand forecasting: The data analysis and procurement optimization module uses a demand forecasting model based on big data analysis and deep learning technology, and uses advanced algorithms to generate procurement plans based on multiple factors such as inventory levels, costs, and supplier delivery cycles. Accurate demand forecasting helps companies avoid blind procurement, reduce inventory backlog costs, ensure that the quantity and type of food purchased matches actual market demand, and improve the scientificity and accuracy of procurement decisions.

[0043] Reduce procurement costs and losses: Through the loss analysis unit, we can deeply explore the loss points and causes of food in each link, and propose targeted optimization measures to effectively reduce procurement costs; at the same time, reasonable procurement plans and optimization measures can reduce out-of-stock costs, improve inventory turnover, achieve effective cost control from multiple aspects, and improve the overall operational efficiency of the enterprise;

[0044] 4. Enhance the management of dishes and inventory:

[0045] Intelligent recipe matching and out-of-stock warning: The intelligent recipe matching unit of the dish and inventory association module can quickly match the raw material inventory information according to the dish production requirements, accurately calculate the inventory consumption and timely generate an out-of-stock list and push it to the procurement department; this ensures the timeliness and accuracy of the raw material supply during the dish production process, avoids production delays caused by raw material shortages, and improves the production efficiency and customer satisfaction of catering companies;

[0046] Scientific inventory allocation: The forecasting and allocation unit uses advanced time series analysis algorithms and machine learning models to predict the sales volume of dishes, and uses inventory allocation optimization algorithms to reasonably allocate the inventory of commonly used raw materials based on the forecast results; this helps companies optimize inventory structure, reduce inventory costs, improve capital utilization efficiency, and achieve coordinated optimization of dish production and inventory management while meeting dish sales demand;

[0047] 5. Ensure system security and user management:

[0048] Strict user rights management: The rights classification unit of the user rights and behavior tracking module has built a multi-level user rights system, which grants corresponding operation rights according to different job responsibilities, and uses a combination of biometric technology and dynamic password authentication to authenticate users; this effectively prevents unauthorized access and operation, ensures the security and integrity of system data, and ensures that sensitive corporate information is not leaked;

[0049] Real-time behavior tracking and abnormality monitoring: The behavior tracking unit records user operation behaviors in real time through the log recording function, and uses the abnormal behavior detection algorithm to build a normal behavior model based on user behavior pattern learning and data statistical analysis technology to monitor and analyze user operations in real time. Once abnormal behavior is found, it can issue a warning in time, which helps enterprises to timely discover and prevent internal risks, such as employee illegal operations, data leakage risks, etc., to ensure the stable operation of the system;

[0050] 6. Provide comprehensive data support and visual decision-making assistance:

[0051] Multi-dimensional data analysis: With the synergy of various modules in the data processing layer, the system can deeply analyze, process and mine various data in the food storage management process, covering inventory, sales, procurement, environment, food association and other aspects; this comprehensive data analysis provides enterprises with rich data resources, which helps to gain a deeper understanding of business operations and discover potential problems and optimization opportunities;

[0052] Intuitive visual decision support: The application display layer displays system functions and data results in an intuitive and visual way through Web applications and mobile apps, including multiple functional interfaces such as inventory management interface, environmental monitoring interface, procurement management interface, as well as data analysis reports, early warning information display, etc.; multi-dimensional visual data reports and interactive charts support data drilling and multi-dimensional analysis functions, allowing management to easily obtain key information, deeply analyze data from different angles, assist in decision-making, improve the scientificity and timeliness of decision-making, and promote efficient operation and sustainable development of the enterprise. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 The figure is a schematic diagram of the overall structure of a food storage management system of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0055] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0056] like Figure 1 As shown, an embodiment of the present invention provides a food storage management system, including a data collection layer, a data transmission layer, a data storage layer, a data processing layer and an application display layer.

[0057] In this embodiment, the data collection layer is composed of a temperature sensor, a humidity sensor, an RFID reader / writer, a barcode scanner, an Internet of Things device, and a user operation terminal. The data collection layer is used to collect food information, environmental data, and user operation data. The specific operation steps are as follows:

[0058] Food information collection:

[0059] Use a barcode scanner or RFID reader to scan the barcode or label on the food package:

[0060] For foods with barcodes, use a barcode scanner to scan and read the barcode information, which may include the name, specification, batch, etc. of the food;

[0061] For foods equipped with RFID tags, RFID readers are used to obtain relevant data stored in the tags, such as the unique identification code of the food, manufacturer and other detailed information, to achieve non-contact fast reading and improve data collection efficiency, especially when batch processing of food is more advantageous;

[0062] Manually enter some food information through the user operation terminal (if necessary):

[0063] In some cases, such as when the barcode or label on the food package is damaged and cannot be identified, or when additional information is required, staff can manually enter relevant information about the food through the user operation terminal, including the name, type, production date, shelf life, etc. of the food;

[0064] Environmental data collection:

[0065] The temperature sensor monitors the ambient temperature in real time:

[0066] Temperature sensors are distributed in key locations in food storage warehouses to continuously detect the temperature data of the surrounding environment;

[0067] The sensor automatically collects temperature values ​​at regular intervals (e.g., a few minutes) and converts them into digital signals;

[0068] The collected temperature data is transmitted to the data processing system via a wired or wireless network;

[0069] Humidity sensor monitors ambient humidity in real time:

[0070] Humidity sensors are also installed at multiple locations in the warehouse to measure the humidity conditions of the environment;

[0071] It collects humidity data at set time intervals (such as synchronization with the temperature sensor or according to actual needs), converts the humidity information into digital signals and transmits them;

[0072] User operation data collection:

[0073] The user operation terminal records the user's various operations in the system:

[0074] When a user logs into the system to perform operations such as warehousing, warehousing, and inventory, the user's operating terminal records in detail the type of operation, time, food batches involved, and quantity;

[0075] For example, during the warehousing operation, the detailed information of the food entering the warehouse, the time of entry, the operator, etc. are recorded; during the outbound operation, the key data such as the type, quantity, outbound destination and the operator of the outbound food are recorded;

[0076] IoT devices (if relevant operations are involved) assist in collecting additional data related to user operations (such as device usage, etc.):

[0077] If IoT devices are used in the food storage management process, such as intelligent handling equipment, automated sorting equipment, etc., these devices will transmit their own operating status, operation records and other data to the system;

[0078] For example, intelligent handling equipment records the starting position, end position, handling time, and equipment status during the handling process (such as whether it is operating normally, whether there are any fault alarms, etc.), providing comprehensive data support for subsequent analysis of user operating behavior and equipment usage efficiency.

[0079] In this embodiment, the data transmission layer uses wired network and wireless network communication technology to transmit the collected data to the data storage layer and the data processing layer. The specific operation steps are as follows:

[0080] Data packaging and identification:

[0081] Each device in the data collection layer (temperature sensor, humidity sensor, RFID reader / writer, barcode scanner, IoT device and user operation terminal) packages the collected data in a predetermined format;

[0082] For example, add identification information such as collection timestamp and device number to each data so that the data can be accurately classified and processed later;

[0083] Perform a preliminary integrity and accuracy check on the packaged data to ensure that there are no obvious errors or omissions in the data before transmission;

[0084] Select a transmission network:

[0085] Automatically or manually select the appropriate transmission network according to the location of the data acquisition device and the network environment;

[0086] When within the coverage of the wired network and the device supports wired connection, wired network should be used for data transmission to ensure the stability and high speed of data transmission; for example, fixed equipment in the warehouse (such as some sensors, servers, etc.) is connected to the LAN switch through Ethernet cables, and then connected to the wired network;

[0087] For areas that cannot be covered by mobile devices or wired networks, wireless networks are used for transmission; wireless networks can use technologies such as Wi-Fi, Bluetooth, 4G / 5G, etc., and wireless networks with strong signals and good stability are selected for connection according to actual conditions; for example, handheld barcode scanners, mobile IoT devices, etc. communicate with the system through Wi-Fi hotspots or mobile data networks;

[0088] Establish connection and transmission:

[0089] The device establishes a connection with the network interface of the data transmission layer;

[0090] For wired network connection, after the device establishes a physical connection with a network device such as a switch or router through a network cable, network configuration (such as assigning an IP address, etc.) is performed to ensure that the device can communicate with the server at the data transmission layer;

[0091] When connecting to a wireless network, the device searches for available wireless network signals, enters the correct password or performs identity authentication, and then establishes a connection with the wireless access point to obtain network access rights;

[0092] After the connection is successfully established, the packaged and verified data is transmitted to the servers of the data storage layer and the data processing layer through the selected network;

[0093] The data is segmented and encapsulated according to the network protocol (such as TCP / IP protocol, etc.), and the necessary network header information is added so that it can be accurately transmitted to the target server in the network;

[0094] During the transmission process, the data transmission layer monitors the data transmission status in real time, performs flow control and error detection on the transmitted data; if data transmission errors or losses are found, timely retransmission or corresponding error correction measures are taken to ensure the integrity and accuracy of the data;

[0095] Data reception and confirmation:

[0096] After receiving the transmitted data, the servers at the data storage layer and the data processing layer perform unpacking and format conversion operations to restore the data to the original format or a format that can be processed by the system;

[0097] The server rechecks the integrity and accuracy of the received data and compares it with the verification results of the data collection layer to ensure that the data has not been changed or damaged during transmission;

[0098] The server sends a data reception confirmation signal to the data acquisition layer to inform the data acquisition device that the data has been successfully received; if data reception fails, the data acquisition layer will resend the data or perform troubleshooting based on the specific situation.

[0099] In this embodiment, the data storage layer uses a combination of relational database and non-relational database to store food inventory information, sales data, purchase data, environmental data, user information, and log data. The specific operation steps are as follows:

[0100] Data classification and planning:

[0101] Detailed classification of data to be stored, and clear characteristics and storage requirements of each type of data;

[0102] For example, food inventory information (such as food name, batch, inventory quantity, storage location, etc.), sales data (sales order, sales time, sales quantity, sales amount, etc.), purchase data (purchase order, supplier information, purchase time, purchase quantity, etc.) and other data with clear structure and relationship are planned and stored in a relational database to facilitate complex query, association and transaction processing;

[0103] Environmental data (large amounts of time series data collected by sensors such as temperature and humidity) and user operation log data (including detailed records of various operation behaviors, with relatively flexible data structures) are suitable for storage in non-relational databases to cope with the requirements of large data volumes, non-fixed structures, and high read and write performance.

[0104] Relational database storage operations:

[0105] Database Design:

[0106] Design the table structure of the relational database according to the needs of the food storage management system; for example, create an inventory information table, including fields such as food ID (associated with the food information table), batch number, inventory quantity, storage time, shelf location, etc.; a sales data table, including fields such as sales order ID, sales time, food ID, sales quantity, sales unit price, etc.; a purchase data table, including fields such as purchase order ID, purchase time, supplier ID, food ID, purchase quantity, purchase unit price, etc., and ensure data integrity and consistency by reasonably designing the table structure and the relationship between tables (such as primary key and foreign key constraints);

[0107] Data insertion and update:

[0108] When data is transmitted from the data transmission layer, it is inserted into the corresponding relational database table according to the data type and business logic; for example, during the warehousing operation, the inventory information of the newly-arrived food is inserted into the inventory information table, and the inventory quantity of the related food is updated at the same time; when the sales business occurs, the sales record is inserted into the sales data table, and the inventory quantity in the inventory information table is updated according to the sales quantity; during the data update process, the atomicity, consistency, isolation and persistence principles of database transactions are followed to ensure the correctness and reliability of data operations;

[0109] Data query and association:

[0110] When the system needs to obtain data from a relational database, it retrieves data by writing SQL query statements. For example, to query the sales data of a certain food in a specific time period and the corresponding inventory changes, the system can obtain accurate data results by associating related tables such as the sales data table and the inventory information table, and using SQL connection operations (such as INNER JOIN, LEFT JOIN, etc.) and conditional filtering statements (WHERE clauses) to provide support for business decisions, such as analyzing sales trends, inventory turnover rates, etc.

[0111] Non-relational database storage operations:

[0112] Choose the appropriate non-relational database type and storage method:

[0113] According to the characteristics of environmental data and log data, select a suitable non-relational database. For example, for time series data such as environmental data, a special time series database such as InfluxDB can be selected, while log data can be stored in document databases such as MongoDB. Taking InfluxDB as an example, it organizes data into tables in a relational database. Each table in a relational database contains multiple tags (used to identify data attributes, such as sensor location, device number, etc.) and fields (storing specific data values, such as temperature and humidity values), as well as timestamp information. This storage method is very suitable for storing and querying large amounts of time series data.

[0114] Data writing and storage optimization:

[0115] The collected environmental data and log data are converted and written to the format according to the requirements of the selected non-relational database. For example, the temperature data collected by the temperature sensor and the corresponding timestamp, sensor location and other information are combined into data points in a specific format and written in batches to the table in the corresponding relational database of InfluxDB. For log data, it is stored in MongoDB in the form of documents, and each document contains detailed information such as operation time, operation type, operation object, and operator. At the same time, storage optimization is performed according to the characteristics of the database, such as setting a reasonable data retention policy in InfluxDB, and determining the length of time to keep data according to the importance and query frequency of the data, so as to save storage space and improve query performance.

[0116] Data query and analysis:

[0117] Use the query interface and functions provided by non-relational databases to query and analyze stored data; for example, use a specific query language (such as Flux) in InfluxDB to query the temperature change trend of a warehouse area within a certain period of time, obtain the required data by setting the time range, filtering conditions (such as specific sensor locations), and perform data aggregation operations (such as calculating average values, maximum values, minimum values, etc.) to meet the needs of environmental monitoring and data analysis; for log data, you can query in MongoDB based on conditions such as operation type and operator, quickly retrieve the operation records of a specific user or the system operation log within a specific period of time, and facilitate tracking of system operation status and user behavior analysis;

[0118] Data consistency maintenance and backup:

[0119] Establish a data consistency maintenance mechanism to ensure that the data in relational databases and non-relational databases are logically consistent. For example, when performing inventory management operations, update the inventory information in the relational database and the inventory-related log records in the non-relational database at the same time. Ensure the consistency of the data on both sides through transaction processing or data synchronization strategies to avoid business errors caused by data inconsistency.

[0120] Regularly back up relational and non-relational databases, formulate backup strategies (such as the cycle of full backup and incremental backup), and store backup data in a safe and reliable location (such as off-site backup storage) to prevent data loss or damage, ensure data security and recoverability, and meet the requirements of the food storage management system for long-term data preservation and data integrity.

[0121] In this embodiment, the data processing layer includes an intelligent warehousing management module, an intelligent outbound management module, an expiration and waste management module, a data analysis and procurement optimization module, a dish and inventory association module, and a user authority and behavior tracking module, which deeply analyzes, processes and mines the data stored in the database to provide data support and decision-making basis for the application display layer. The specific operations are as follows:

[0122] Intelligent warehousing management module:

[0123] Food information collection and verification:

[0124] When food is put into storage, use automated scanning equipment (such as barcode scanner or RFID reader) to scan the barcode or label on the food packaging to obtain basic information about the food, including name, specification, batch number, production date, shelf life, etc.;

[0125] The scanned information is connected to the food quality traceability database to verify the legality of the food source and quality compliance; at the same time, the system performs multiple internal checks on the collected food information to ensure that the data format is accurate, complete and consistent with the existing food information;

[0126] Batch information management and storage:

[0127] The production batch information of foods that have passed the verification is encrypted through blockchain technology; key information such as food production batch number, production date, production location, source of raw materials, etc. is recorded in detail and stored according to the blockchain data structure;

[0128] In the blockchain, the hash value calculation of the data block adopts a specific algorithm (SHA-256 algorithm), and the hash value of the previous data block and the current data block information jointly generate a new hash value to form an unalterable traceability chain; then, the encrypted and stored batch information is associated with the food inventory information, and the detailed status information of each batch of food is displayed on the inventory management interface;

[0129] Intelligent outbound management module:

[0130] Optimization calculation of outbound order:

[0131] The system obtains food inventory data from the database, including information such as storage time, shelf location and volume;

[0132] First, sort the food in ascending order by the time of entry; for food with the same entry time, calculate the priority value based on the distance from the shelf position to the delivery port and the handling efficiency coefficient, so as to determine the order of delivery, generate the optimal delivery plan, guide the delivery operation, and reduce the risk of food expiration;

[0133] Post-delivery tracking and analysis:

[0134] With the help of IoT technology and GPS positioning, the location information of the transport vehicle is obtained when the food is shipped out of the warehouse and the shipping records are linked to record the shipping destination; it is connected with the catering enterprise information system to obtain information on the food usage link and generate dynamic food consumption reports;

[0135] The transportation link uses a specific algorithm (Kalman filter algorithm) to estimate the position and state of the transport vehicle in real time. By setting relevant parameters (system state vector, state transfer matrix, observation matrix, etc.) and combining sensor data, a series of formulas (state prediction formula, covariance prediction formula, observation update formula and covariance update formula) are used to optimize the estimation, monitor the transportation process, detect abnormalities in time, ensure transportation safety, and record data for subsequent analysis and optimization;

[0136] Expiration and waste management module:

[0137] Expired risk assessment and reminder:

[0138] The system regularly retrieves food inventory information to obtain key data such as production date, shelf life, inventory quantity, etc.

[0139] The food expiration risk index is calculated by combining the time priority algorithm with the food attribute analysis model; the food is sorted according to the expiration risk index. When the index reaches the preset threshold, dynamic expiration reminder information is sent to the management personnel through various methods (system pop-up window, SMS notification, email reminder, etc.) so as to handle it in time and avoid food expiration waste;

[0140] Residual value assessment and utilization strategy formulation:

[0141] Collect multi-dimensional data such as remaining shelf life of food, market demand, historical data on sales price fluctuations, brand influence, changes in consumer preferences, etc., and comprehensively evaluate the remaining value of food that is about to expire;

[0142] The evaluation is conducted using a neural network based on a multi-layer perceptron (MLP). The input layer neurons correspond to relevant evaluation factors, the hidden layer uses a specific activation function (Sigmoid activation function) to process data, and the output layer is the food residual value score. After training with a large amount of sample data to determine the weights and thresholds, personalized secondary utilization recommendations are formulated for different foods based on the evaluation results, such as promotion, donation, or adjustment of packaging specifications, to reduce food waste and improve resource utilization.

[0143] Data analysis and procurement optimization module:

[0144] Procurement plan generation:

[0145] Build a demand forecasting model based on big data analysis and deep learning technology to obtain historical sales data, inventory level data, supplier delivery cycle data, purchase price fluctuation data and other relevant data from the database;

[0146] The random forest algorithm is combined with the gradient boosting decision tree (GBDT) algorithm to predict demand. The random forest algorithm determines the prediction result by building multiple decision trees for voting, and the decision tree construction selects the optimal feature split based on information gain. The GBDT algorithm is iteratively optimized by fitting the residual of the previous round of models. Based on the prediction results combined with factors such as inventory level and cost, a purchase plan is generated using an optimization algorithm combining linear programming and dynamic programming to determine the type, quantity, purchase time window and recommended supplier list of purchased food, providing a scientific basis for purchasing decisions, optimizing inventory structure and reducing costs.

[0147] Loss analysis and optimization measures formulation:

[0148] Use data mining and statistical analysis techniques to integrate multi-source data, including temperature and humidity data in the storage process, route, method, and time data in the transportation process, and production process and loss data in the use process of catering enterprises;

[0149] A multivariate linear regression model is used to analyze the impact of environmental conditions on food loss in the storage stage and determine the loss point; similar analysis is conducted on the data of the transportation and use stages to find out the causes of loss in each stage and establish a correlation model; targeted optimization measures are proposed based on the analysis results, such as adjusting warehouse environmental parameters, optimizing transportation routes, improving production processes, etc., to reduce the loss rate, improve management efficiency, and ensure food safety quality;

[0150] Decision support and data visualization:

[0151] Through data visualization technology, key data such as inventory turnover rate, food storage period, procurement cost analysis, loss rate, etc. are extracted, and statistical analysis is performed according to multiple dimensions (time, food category, warehouse area, supplier, etc.);

[0152] Use visualization tools (bar charts, line charts, pie charts, radar charts, interactive data dashboards, etc.) to convert data into intuitive charts, providing data drilling and multi-dimensional analysis functions; users can dig deep into data through interactive operations, assist management to understand business conditions, discover problem trends, and provide data support for decision-making, such as adjusting procurement strategies, optimizing inventory layout, improving operational processes, etc., to enhance corporate competitiveness and efficiency;

[0153] Module for associating dishes with inventory:

[0154] Intelligent formula matching and out-of-stock warning:

[0155] Establish a recipe database to store information such as the types of raw materials, quantities, and production processes required for the dishes;

[0156] When a catering company has a dish production plan, it retrieves the required raw material collection information from the recipe database according to the dish name and queries the inventory quantity at the same time; calculates the raw material inventory consumption for making the dish, compares the consumption with the existing inventory to generate a shortage list, including the name and quantity of the shortage raw materials, and pushes it to the procurement department to ensure the smooth progress of dish production and avoid the impact of raw material shortage on production;

[0157] Product sales forecast and inventory allocation optimization:

[0158] When predicting the sales volume of dishes, we comprehensively consider historical sales trends, seasonal factors, marketing promotion effects, changes in surrounding consumer groups and other multi-dimensional factors, and obtain historical sales data from the database for analysis;

[0159] The seasonal autoregressive moving average SARIMA model is used for forecasting. The model parameters are determined based on historical data and related factors to generate future sales forecast data to guide production planning and inventory management. Based on the forecast results, the inventory allocation optimization algorithm is used to reasonably allocate the inventory of common raw materials. The raw material allocation plan is encoded into chromosomes and optimized through genetic operations such as selection, crossover, and mutation. The fitness function is determined based on inventory cost and out-of-stock risk to optimize inventory resource allocation and improve turnover rate.

[0160] User permissions and behavior tracking module:

[0161] User rights hierarchical management and identity authentication:

[0162] Build a multi-level user authority system and set corresponding operation permissions in the system according to different job responsibilities (warehouse managers, purchasing staff, sales staff, financial staff, management staff, etc.);

[0163] User identity authentication is performed by combining biometric technology (fingerprint recognition, facial recognition, iris recognition, etc.) with dynamic password authentication; when a user logs in, biometric recognition is first performed to verify the authenticity of the identity.

[0164] In this embodiment, the application display layer displays system functions and data results to users through Web applications and mobile APPs. The application display layer includes inventory management interface, environmental monitoring interface, procurement management interface, data analysis report, and early warning information display. The specific operation steps are as follows:

[0165] Web application showcase:

[0166] The user accesses the Web application address of the food storage management system through a computer browser and enters the login page; after entering the correct user name and password, the system loads the corresponding function interface according to the user's authority;

[0167] In the inventory management interface, detailed inventory information of food is clearly displayed in a table, including key data such as food name, batch, inventory quantity, storage location, storage time, shelf life, etc. Users can enter keywords such as food name or batch number in the search box to quickly query the inventory status of specific food; they can also click on the column header to sort, so as to conveniently view food with large inventory quantity or approaching the shelf life; at the same time, the interface provides an inventory counting function. After the user clicks the corresponding button, the system automatically generates an inventory counting list, showing the difference between the actual inventory and the system record, to assist users in inventory verification and management;

[0168] The environmental monitoring interface intuitively presents the environmental data in the warehouse in the form of charts, such as the curves of temperature and humidity changes over time; users can select different time periods (such as the past 24 hours, the past week, the past month, etc.) to view the environmental data trends, so as to promptly discover environmental anomalies; when environmental data exceeds the preset safety range, the interface will mark the abnormal data points with eye-catching colors (such as red) and trigger early warning prompts (such as pop-up warning boxes showing specific abnormal information), reminding users to take corresponding measures, such as adjusting the parameters of warehouse air conditioning or ventilation equipment to ensure a suitable food storage environment;

[0169] The purchasing management interface displays relevant information of the purchase order, including the purchase order number, purchase time, supplier name, purchased food type, quantity, unit price, total price, etc.; users can view the detailed status of the purchase order (such as submitted, reviewed, shipped, received, etc.) on this interface, and operate the purchase order, such as adding a new purchase order, modifying the order content (before the order is reviewed), confirming receipt, etc.; at the same time, the interface provides supplier management functions, users can view the supplier list, add new suppliers, edit supplier information (such as contact information, address, supply capacity, etc.), and evaluate suppliers (based on factors such as supply timeliness, product quality, price, etc.), providing a reference for purchasing decisions;

[0170] The data analysis report interface displays the data results after system analysis and processing in the form of various charts (such as bar charts, line charts, pie charts, etc.) and reports; for example, a bar chart is used to compare the sales volume of different types of food in a certain period of time to clearly reflect the sales situation of various types of food; a line chart is used to show the trend of inventory turnover rate over time to help users evaluate inventory management efficiency; a pie chart is used to analyze the proportion of food loss in each link (storage, transportation, use, etc.) to facilitate users to find key loss links for improvement; the report content covers inventory turnover rate, food storage cycle, procurement cost analysis, loss rate and other key indicators, and segmented and displayed by different dimensions (such as time dimension: monthly, quarterly, annual; food category dimension: fresh food, processed food, condiments, etc.; warehouse area dimension: different warehouses or different areas in the warehouse; supplier dimension: food provided by different suppliers, etc.); users can click on the data points in the chart or report to drill down and view detailed data in depth. For example, by clicking on a monthly inventory turnover bar chart, you can view the inventory turnover data for each week or day of the month, as well as the specific food batches and inventory changes involved, to assist users in in-depth data analysis and decision-making;

[0171] The warning information display interface centrally presents various warning information generated by the system, such as food expiration warning, insufficient inventory warning, environmental abnormality warning, etc. The warning information is displayed in a list format, and each piece of information includes the warning type, the name of the food involved, the batch, the warning time, a detailed description, etc. Users can filter and view according to conditions such as warning type and time range to ensure timely processing of important warning information and avoid potential risks affecting food storage management;

[0172] Mobile APP display:

[0173] The user downloads and installs the mobile APP of the food storage management system on a mobile device (such as a mobile phone or tablet); when opening the APP for the first time, the user registers and logs in, enters the user name and password to log in to the system, and the system customizes the APP interface display content according to the user's permissions;

[0174] The inventory management interface of the mobile APP adopts a simple and intuitive layout, displaying key inventory information in a list format, which is convenient for users to quickly view in mobile scenarios; users can quickly locate the inventory information of the required food by sliding the screen or using the search function, and gesture operations are also supported (such as long pressing a food item to edit or view detailed information); in addition, the APP provides an inventory warning push function. When the food inventory quantity is lower than the preset safety inventory level or is approaching the expiration date, a notification message is pushed to the relevant users to ensure that users are aware of inventory dynamics in a timely manner and can take timely replenishment or processing measures even if they are not in front of the computer;

[0175] The environment monitoring interface displays warehouse environment data in the form of dynamic charts on the mobile terminal. Users can switch the environment data trends of different time periods by sliding their fingers, and grasp the warehouse environment conditions in real time. When the environment is abnormal, the APP immediately pushes an early warning notification to the user's mobile phone. The user clicks the notification to directly enter the environment monitoring interface to view detailed abnormal information, and can remotely operate related equipment on the APP (such as adjusting the temperature controller, checking the operating status of the ventilation equipment, etc.), solve environmental problems in a timely manner, and ensure food quality and safety.

[0176] The procurement management interface simplifies the operation process on the mobile APP, making it convenient for users to perform procurement-related operations anytime and anywhere; users can check the status of purchase orders, approve purchase applications (for users with approval authority), receive purchase arrival notifications, etc. on the APP; at the same time, using the camera and code scanning functions of mobile devices, users can directly scan the barcode on the food packaging or take photos of the supplier's qualification documents and upload them to the system at the procurement site, improving the convenience and accuracy of purchasing information entry;

[0177] The data analysis report is presented in the form of an interactive chart that adapts to the mobile device screen. Users can view data of different dimensions by pinching, clicking, and other operations. For example, pinching the chart to adjust the time range, and clicking the chart area to switch the display of different data dimensions (such as switching from the food category dimension to the warehouse area dimension) facilitates users to quickly obtain data analysis results on the move and assist in decision-making. In addition, the APP supports sharing data analysis reports in PDF or image format to other applications (such as email, instant messaging tools, etc.), which facilitates users to share data and communicate with colleagues or superiors.

[0178] Warning information is displayed in a prominent position on the mobile APP, such as setting up a warning information bar on the homepage or reminding users through push notifications; users can view detailed content by clicking on the warning information, and can directly perform processing operations on the APP (such as directly jumping to the purchase application page to apply for replenishment for insufficient inventory warnings), ensuring that users respond to warning events in a timely manner and reduce risks; at the same time, the APP records the user's processing history of warning information, which is convenient for users to trace and query.

[0179] In this embodiment, the intelligent warehousing management module in the data processing layer includes:

[0180] Automatic data entry and verification unit: integrated with automated scanning equipment, scans food barcodes or QR codes to enter information and conduct multiple verifications, including connecting with the food quality traceability database to verify the legality of food sources and quality compliance;

[0181] Batch management unit: Blockchain technology is used to encrypt and store food production batch information and conduct full traceability management. The detailed status information of each batch of food is displayed on the inventory management interface. The blockchain technology in the batch management unit records detailed information such as food production batch number, production date, production location, source of raw materials, etc., and forms an unalterable traceability chain. In the blockchain, the hash value calculation of the data block adopts the SHA-256 algorithm. For a data block , its hash value ,in, Represents a data block, Represents a data block The hash value of the previous data block is combined with the current data block information to generate a new data block hash value to ensure the integrity and non-tamperability of the data.

[0182] In this embodiment, the intelligent outbound management module in the data processing layer includes:

[0183] First-in-first-out algorithm optimization unit: Based on the food time label and inventory distribution, an intelligent algorithm is used to combine the warehouse shelf layout and cargo handling path to calculate the optimal outbound order. Suppose the food inventory set is , each food element Including storage time Shelf location and volume , sorted in ascending order according to the time of entry. For foods with the same entry time, sorted according to the distance from the shelf location to the delivery port and handling efficiency coefficient Calculating priority values ,in, Indicates the priority value. Indicates the distance adjustment coefficient, and determines the order of outbound delivery according to the priority value from small to large;

[0184] Real-time tracking unit: Uses IoT technology and GPS positioning to record food delivery information, connects with the catering enterprise information system to track food usage information, and generates dynamic food consumption reports;

[0185] The real-time tracking unit also includes monitoring of the transportation link after leaving the warehouse and detailed recording and analysis of the association between food and dishes in catering enterprises. The Kalman filter algorithm is used to estimate the position and state of the transport vehicle in the transportation link monitoring. The system state vector ,in, Indicates location, Represents speed, state transfer matrix ,in, represents the time interval, the observation matrix , then the state prediction formula is: ,in, express The prior state estimate at time t, express The posterior state estimate at the moment; the covariance prediction formula is: ,in, express The prior covariance estimate of time, express The posterior covariance estimate at time , represents the process noise covariance matrix; the observation update formula is: ,in, represents the Kalman gain, represents the observation noise covariance matrix, where Represents the observation matrix; the covariance update formula is: .

[0186] In this embodiment, the expiration and waste management module in the data processing layer includes:

[0187] Dynamic expiration reminder unit: Combine the time priority algorithm with the food attribute analysis model, and send dynamic expiration reminder information to managers in the form of system pop-up windows, SMS notifications, and email reminders based on the food production date, shelf life, inventory quantity, and expiration risk weight. The calculation formula for the expiration risk index in the time priority algorithm is: ,in, Indicates Expiration risk index of food, Indicates the expiration date of food. Indicates the current time. Indicates the shelf life. represents the weight factor related to food attributes;

[0188] Residual value evaluation unit: Through deep learning algorithms, it comprehensively considers multi-dimensional factors including the remaining shelf life of food, market demand conditions, historical data on sales price fluctuations, brand influence, and changes in consumer preferences, and formulates personalized secondary utilization recommendations for foods that are about to expire; the residual value evaluation unit formulates strategies including promotion, donation, or adjustment of packaging specifications for different foods based on the evaluation results. Among them, the residual value evaluation adopts a neural network based on a multi-layer perceptron MLP. The input layer neurons correspond to the remaining shelf life of food, market demand index, and brand influence scoring factors, and the hidden layer adopts a Sigmoid activation function: , the output layer is the food surplus value score, and the weights and thresholds are determined by training the neural network with a large amount of sample data.

[0189] In this embodiment, the data analysis and procurement optimization module in the data processing layer includes:

[0190] Procurement suggestion algorithm unit: Based on the demand forecasting model built with big data analysis and deep learning technology, the optimization algorithm combining linear programming and dynamic programming is used to generate a procurement plan, which includes the types and quantities of food to be purchased, the procurement time window and the recommended supplier list. The demand forecasting model uses the random forest algorithm combined with the gradient boosting decision tree GBDT algorithm. The random forest determines the prediction result through voting of multiple decision trees. The information gain calculation in the decision tree construction process is as follows: ,in, represents a data set, Represents feature attributes, Representation attributes The value set of Represents information baking function: Indicates that the data set belongs to The GBDT algorithm performs iterative optimization by continuously fitting the residuals of the previous round of models. When optimizing the procurement plan, the procurement cost is set to , the inventory holding cost is ®, and the stockout cost is , the purchase volume is , the demand forecast value is , then the objective function is: , until the constraints are met ,in, express Inventory levels at any given moment, express Purchase quantity at any time, express Demand forecast value at each moment;

[0191] Loss analysis unit: Use data mining and statistical analysis techniques to mine multi-source data, identify food loss points and causes in storage, transportation, and use, establish correlation models, and propose optimization measures. The loss analysis unit analyzes the temperature, humidity, and other conditions in the storage link, the routes and methods in the transportation link, and the production process in the use link of catering enterprises. A multivariate linear regression model is used in the loss analysis of the storage link: ,in, Indicates the amount of food loss, Indicates factors including temperature and humidity. represents the regression coefficient, , represents the random error term;

[0192] Decision support unit: Generate multi-dimensional visual data reports and interactive charts through data visualization technology, support data drilling and multi-dimensional analysis functions, and assist management decision-making. The reports and charts of the decision support unit cover inventory turnover rate, food storage cycle, procurement cost analysis, loss rate, and are displayed in multiple dimensions.

[0193] In this embodiment, the dish and inventory association module in the data processing layer includes:

[0194] Intelligent recipe matching unit, establishes a recipe database, matches raw material inventory information according to dish production requirements, calculates inventory consumption and generates a stock-out list and sends it to the purchasing department. Need raw material collection , the raw material inventory set is , for each raw material ,The formula for calculating inventory consumption is: ,in, Indicates the inventory consumption, Indicates dishes Medium Raw Materials The dosage, Indicates dishes The estimated production volume and the out-of-stock judgment formula are: ;

[0195] The forecasting and allocation unit uses a combination of time series analysis algorithms and machine learning models to predict the sales volume of dishes, and uses an inventory allocation optimization algorithm to reasonably allocate the inventory of commonly used raw materials based on the forecast results. When forecasting dish sales, the forecasting and allocation unit comprehensively considers historical sales trends, seasonal factors, marketing promotion effects, and changes in surrounding consumer groups. The seasonal autoregressive moving average SARIMA model is used to predict the sales volume of dishes, and its formula is: ,in, represents the backshift operator, represents nonseasonal autoregressive and moving average polynomials, represents seasonal autoregressive and moving average polynomials, Represents the seasonal cycle, represents the difference order, Indicates that time series data is The value of the moment, Represents a white noise sequence; inventory allocation optimization uses a genetic algorithm to encode the raw material allocation plan into chromosomes, and continuously optimizes chromosomes through genetic operations such as selection, crossover, and mutation. The fitness function is determined based on inventory cost and out-of-stock risk.

[0196] In this embodiment, the user rights and behavior tracking module in the data processing layer includes:

[0197] The authority classification unit builds a multi-level user authority system, grants corresponding operation permissions according to different job responsibilities, and uses biometric technology combined with dynamic password authentication for user identity authentication;

[0198] The behavior tracking unit records user operation behaviors in real time through the log recording function, and uses the abnormal behavior detection algorithm to monitor and analyze user operation behaviors in real time; the abnormal behavior detection algorithm builds a normal user operation behavior model based on user behavior pattern learning and data statistical analysis technology and monitors anomalies accordingly, and uses the hidden Markov model HMM to model the user operation behavior sequence. Suppose the hidden state set is , the set of observed states is , state transition probability matrix ,in, , indicating that from the state Transfer to state The probability of observation probability matrix ,in, , indicating that in the state Observed below The probability of the initial state probability vector ,in, , indicating that the initial state is The probability of HMM model parameters is trained by Baum-Welch algorithm, and then the Viterbi algorithm is used to determine the most likely hidden state sequence according to the observation sequence to determine whether there is abnormal behavior.

[0199] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A food storage management system, characterized in that: include: Data collection layer: The data collection layer is composed of temperature sensors, humidity sensors, RFID readers, barcode scanners, IoT devices and user operation terminals. The data collection layer is used to collect food information, environmental data and user operation data; Data storage layer: The data storage layer uses wired network and wireless network communication technology to transmit the collected data to the data storage layer and the data processing layer; Data storage layer: The data storage layer uses a combination of relational database and non-relational database to store food inventory information, sales data, purchase data, environmental data, user information, and log data; Data processing layer: The data processing layer includes intelligent warehousing management module, intelligent outbound management module, expiration and waste management module, data analysis and procurement optimization module, dish and inventory association module, and user authority and behavior tracking module. It conducts in-depth analysis, processing and mining of data stored in the database to provide data support and decision-making basis for the application display layer; Application display layer: The application display layer displays system functions and data results to users through Web applications and mobile APPs. The application display layer includes inventory management interface, environmental monitoring interface, procurement management interface, data analysis reports, and early warning information display.

2. A food storage management system according to claim 1, characterized in that: The intelligent warehousing management module in the data processing layer includes: Automatic data entry and verification unit: integrated with automated scanning equipment, scans food barcodes or QR codes to enter information and conduct multiple verifications, including connecting with the food quality traceability database to verify the legality of food sources and quality compliance; Batch management unit: Blockchain technology is used to encrypt and store food production batch information and conduct full traceability management. The detailed status information of each batch of food is displayed on the inventory management interface. The blockchain technology in the batch management unit records the food production batch number, production date, production location, and raw material source to form an unalterable traceability chain. In the blockchain, the hash value calculation of the data block adopts the SHA-256 algorithm. For a data block , its hash value ,in, Represents a data block, Represents a data block The hash value of the new data block is generated by combining the hash value of the previous data block with the current data block information to ensure the integrity and non-tamperability of the data.

3. A food storage management system according to claim 1, characterized in that: The intelligent outbound management module in the data processing layer includes: First-in-first-out algorithm optimization unit: Based on the food time label and inventory distribution, an intelligent algorithm is used to combine the warehouse shelf layout and cargo handling path to calculate the optimal outbound order. Suppose the food inventory set is ,in, is the number of food types in inventory, each food element Including storage time Shelf location and volume , sorted in ascending order according to the time of entry. For foods with the same entry time, sorted according to the distance from the shelf location to the delivery port and handling efficiency coefficient Calculating priority values ,in, Indicates the priority value. Indicates the distance adjustment coefficient, and determines the order of outbound delivery according to the priority value from small to large; Real-time tracking unit: Uses IoT technology and GPS positioning to record food delivery information, connects with the catering enterprise information system to track food usage information, and generates dynamic food consumption reports; The real-time tracking unit also includes monitoring of the transportation link after leaving the warehouse and detailed recording and analysis of the association between food and dishes in catering enterprises. The Kalman filter algorithm is used in the transportation link monitoring to estimate the position and state of the transport vehicle. The system state vector ,in, Indicates location, Represents speed, state transfer matrix ,in, represents the time interval, the observation matrix , then the state prediction formula is: ,in, express The prior state estimate at time t, express The posterior state estimate at the moment; the covariance prediction formula is: ,in, express The prior covariance estimate of time, express The posterior covariance estimate at time , represents the process noise covariance matrix; the observation update formula is: ,in, represents the Kalman gain, represents the observation noise covariance matrix, where Represents the observation matrix; the covariance update formula is: .

4. A food storage management system according to claim 1, characterized in that: The expiration and waste management module in the data processing layer includes: Dynamic expiration reminder unit: Combine the time priority algorithm with the food attribute analysis model, and send dynamic expiration reminder information to managers in the form of system pop-up windows, SMS notifications, and email reminders based on the food production date, shelf life, inventory quantity, and expiration risk weight. The calculation formula for the expiration risk index in the time priority algorithm is: ,in, Indicates Expiration risk index of food, Indicates the expiration date of food. Indicates the current time. Indicates the shelf life. represents the weight factor related to food attributes; Residual value evaluation unit: through deep learning algorithms, comprehensively consider multi-dimensional factors including the remaining shelf life of food, market demand conditions, historical data of sales price fluctuations, brand influence, and changes in consumer preferences, and formulate personalized secondary utilization suggestions for foods that are about to expire; the residual value evaluation unit formulates strategies including promotion, donation or adjustment of packaging specifications for different foods based on the evaluation results, wherein the residual value evaluation adopts a neural network based on a multi-layer perceptron MLP, the input layer neurons correspond to the remaining shelf life of food, market demand index, and brand influence scoring factors, and the hidden layer adopts a Sigmoid activation function: , the output layer is the food surplus value score, and the weights and thresholds are determined by training the neural network with a large amount of sample data.

5. A food storage management system according to claim 1, characterized in that: The data analysis and procurement optimization module in the data processing layer includes: Procurement suggestion algorithm unit: Based on the demand forecasting model built with big data analysis and deep learning technology, the optimization algorithm combining linear programming and dynamic programming is used to generate a procurement plan, which includes the types and quantities of food to be purchased, the procurement time window and the recommended supplier list. The demand forecasting model uses the random forest algorithm combined with the gradient boosting decision tree GBDT algorithm. The random forest determines the prediction result through voting of multiple decision trees. The information gain calculation in the decision tree construction process is as follows: ,in, represents a data set, represents the sample subset in the data set whose attribute a takes the value v, Represents feature attributes, Representation attributes The value set of Represents information baking function: Indicates that the data set belongs to The GBDT algorithm performs iterative optimization by continuously fitting the residuals of the previous round of models. When optimizing the procurement plan, the procurement cost is set to , the inventory holding cost is ®, and the stockout cost is , the purchase volume is , the demand forecast value is , then the objective function is: , until the constraints are met ,in, express Inventory levels at any given moment, express Purchase quantity at any time, express Demand forecast value at each moment; Loss analysis unit: Use data mining and statistical analysis technology to mine multi-source data, identify food loss points and causes in storage, transportation, and use, establish a correlation model, and propose optimization measures. The loss analysis unit analyzes the temperature and humidity conditions in the storage link, the route and method in the transportation link, and the production process in the use link of catering enterprises. A multivariate linear regression model is used in the loss analysis of the storage link: ,in, Indicates the amount of food loss, Indicates factors including temperature and humidity. represents the regression coefficient, , represents the random error term; Decision support unit: Generate multi-dimensional visual data reports and interactive charts through data visualization technology, support data drilling and multi-dimensional analysis functions, and assist management decision-making. The reports and charts of the decision support unit cover inventory turnover rate, food storage cycle, procurement cost analysis, loss rate, and are displayed in multiple dimensions.

6. A food storage management system according to claim 1, characterized in that: The dish and inventory association module in the data processing layer includes: Intelligent recipe matching unit, establishes a recipe database, matches raw material inventory information according to dish production requirements, calculates inventory consumption and generates a stock-out list and sends it to the purchasing department. Need raw material collection , the raw material inventory set is , for each raw material ,The formula for calculating inventory consumption is: ,in, Indicates the inventory consumption, Indicates dishes Medium Raw Materials The dosage, Indicates dishes The estimated production volume and the out-of-stock judgment formula are: ; The forecasting and allocation unit uses a combination of time series analysis algorithm and machine learning model to predict the sales volume of dishes, and uses the inventory allocation optimization algorithm to reasonably allocate the inventory of commonly used raw materials according to the forecast results. The forecasting and allocation unit comprehensively considers historical sales trends, seasonal factors, marketing promotion effects and changes in surrounding consumer groups when predicting dish sales. The seasonal autoregressive moving average SARIMA model is used to predict the sales volume of dishes, and its formula is: ,in, represents the backshift operator, represents nonseasonal autoregressive and moving average polynomials, represents seasonal autoregressive and moving average polynomials, Represents the seasonal cycle, represents the difference order, Indicates that time series data is The value of the moment, Represents a white noise sequence; inventory allocation optimization uses a genetic algorithm to encode the raw material allocation plan into chromosomes, and continuously optimizes chromosomes through genetic operations such as selection, crossover, and mutation. The fitness function is determined based on inventory cost and out-of-stock risk.

7. A food storage management system according to claim 1, characterized in that: The user rights and behavior tracking module in the data processing layer includes: The authority classification unit builds a multi-level user authority system, grants corresponding operation permissions according to different job responsibilities, and uses biometric technology combined with dynamic password authentication for user identity authentication; The behavior tracking unit records user operation behaviors in real time through the log recording function, and uses the abnormal behavior detection algorithm to monitor and analyze user operation behaviors in real time; the abnormal behavior detection algorithm builds a normal user operation behavior model based on user behavior pattern learning and data statistical analysis technology and monitors anomalies accordingly, and uses the hidden Markov model HMM to model the user operation behavior sequence. Suppose the hidden state set is , the set of observed states is , state transition probability matrix ,in, , indicating that from the state Transfer to state The probability of observation probability matrix ,in, , indicating that in the state Observed below The probability of the initial state probability vector ,in, , indicating that the initial state is The probability of HMM model parameters is trained by Baum-Welch algorithm, and then the Viterbi algorithm is used to determine the most likely hidden state sequence according to the observation sequence to determine whether there is abnormal behavior.

Images