A consignment management system and method thereof

Abstract

The invention discloses a consignment management system and a method thereof, which relate to the technical field of time series prediction, and include collecting multi-source data and preprocessing the data; extracting features from the data and constructing a multi-level memory unit LSTM model; predicting future consignment material demand through the multi-level memory unit LSTM model and automatically adjusting inventory management; monitoring the multi-level memory unit LSTM model according to the difference between actual sales and demand prediction data, and updating the multi-level memory unit LSTM model; based on adjusting inventory management, automating warehousing and delivery, generating relevant records, and updating an inventory management center; the invention can effectively guide inventory management through accurate prediction of the multi-level memory unit LSTM model, reduce inventory backlog and stockout, and improve the efficiency and accuracy of inventory management.

Description

Technical Field

[0001] The present invention relates to the technical field of time series prediction, in particular to a consignment management system and a method thereof. Background Art

[0002] With the rapid development of e-commerce and supply chain management, consignment system, as a flexible inventory management model, has gradually been favored by enterprises. The consignment system allows suppliers to store goods in retailers' warehouses and settle only after the goods are sold, thereby reducing retailers' financial pressure and inventory risks. However, traditional consignment management methods mainly rely on manual experience and simple statistical methods, which are difficult to cope with complex and changeable market demands and supply chain fluctuations. In recent years, with the development of big data and machine learning, more and more applications have begun to explore how to use advanced forecasting models and automated systems to optimize consignment management. For example, forecasting models based on time series analysis have shown high accuracy in sales forecasting, but these models are often limited to a single data source and simple feature extraction methods, and cannot fully consider the impact of multiple external factors.

[0003] The existing consignment management methods have several shortcomings, especially in terms of the accuracy of demand forecasting and the flexibility of inventory management. First, traditional methods usually rely only on historical sales data for forecasting, ignoring the impact of external factors such as weather, holidays, and promotional activities, resulting in large deviations in forecast results. Second, inventory adjustment strategies lack dynamism and intelligence, and cannot respond to market changes in a timely manner, which can easily lead to inventory backlogs or out-of-stock situations. In addition, the integration of automated warehousing and outbound management systems is not high, resulting in low operating efficiency and increased labor costs. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a consignment management method to solve the problems of accurate demand forecasting and efficient inventory management.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a consignment management method, which includes collecting multi-source data and preprocessing the data; extracting features from the data and constructing a multi-level memory unit LSTM model; predicting future consignment material demand through the multi-level memory unit LSTM model and automatically adjusting inventory management; monitoring the multi-level memory unit LSTM model and updating the multi-level memory unit LSTM model based on the difference between actual sales and demand forecast data; automating warehousing and outbound delivery based on adjusted inventory management, generating relevant records, and updating the inventory management center.

[0008] As a preferred solution of the consignment management method of the present invention, the following steps are performed: feature extraction is performed on the data to construct a multi-level memory unit LSTM model:

[0009] Extract time series features, external features, and lag features from the preprocessed data;

[0010] The extracted time series features, external features and lag features are saved into a feature data set, and the feature data set is converted into time series data;

[0011] Define the hidden number of the first-layer multi-level memory unit LSTM model as the maximum value, use the tanh activation function, and set true to pass it to the next layer;

[0012] Define the hidden number of the second-layer multi-level memory unit LSTM model as the median, use the tanh activation function, and set true to pass it to the next layer;

[0013] Define the hidden number of the third-layer multi-level memory unit LSTM model as the minimum value, use the tanh activation function, and set False as the output of the multi-level memory unit LSTM model;

[0014] Define the Dropout layer of the three-layer multi-level memory unit LSTM model as a fixed value;

[0015] The preprocessed time series data is input into the multi-level memory unit LSTM model for training, and the hyperparameters are adjusted through cross-validation and grid search.

[0016] As a preferred solution of the consignment management method of the present invention, the specific steps of predicting the future demand for consignment materials and automatically adjusting inventory management are as follows:

[0017] Through the trained multi-level memory unit LSTM model, the future demand for consignment materials is predicted;

[0018] Input time series data for a period of time in the future, and the model outputs future demand forecast data;

[0019] According to the demand forecast data, the inventory level is automatically adjusted, replenishment and reduction instructions are generated, and the instructions are sent through the inventory management center. The expression is:

[0020] ;

[0021] in, is the adjusted inventory, is the current inventory, is to predict demand, is the adjustment factor, is the safety stock value, is an exponential decay function, is the attenuation coefficient, is the time variable, is the minimum function, is the maximum value function.

[0022] As a preferred solution of the consignment management method of the present invention, wherein: the multi-level memory unit LSTM model is monitored and the multi-level memory unit LSTM model is updated, and the specific steps are:

[0023] Monitor the difference between actual sales data and demand forecast data in real time, set thresholds, and trigger alarms when the difference exceeds the threshold;

[0024] The multi-level memory unit LSTM model is updated regularly and retrained according to the new time series data. The expression is:

[0025] ;

[0026] in, is the new learning rate, is the current learning rate, is the validation set loss, is the training set loss, is the exponential decay factor, is the natural logarithm function, is the piecewise function during the training phase, is the training progress rate, is the stage control parameter, is the natural logarithm of the current learning rate;

[0027] After updating the multi-level memory unit LSTM model, re-perform demand forecasting.

[0028] As a preferred solution of the consignment management method of the present invention, the multi-source data collected includes historical sales data, market trend data, weather data, holiday information and promotion activity data.

[0029] As a preferred solution of the consignment management method of the present invention, the preprocessing includes cleaning the data, removing abnormal values, missing values ​​and data standardization.

[0030] As a preferred solution of the consignment management method of the present invention, the automated warehousing and delivery, and the generation of relevant records and the updating of the inventory management center are specifically performed as follows:

[0031] After receiving the replenishment instruction, the inventory management center automatically generates a purchase order;

[0032] When receiving consignment materials from suppliers, the material information is identified through RFID tags and barcodes;

[0033] Record material information to the inventory management center and generate inventory records;

[0034] Deliver materials to designated locations through automated equipment and record storage location and status, update inventory records at the inventory management center, and generate storage records;

[0035] Automatically generate picking tasks based on outbound orders;

[0036] Pick materials from shelves according to task instructions and verify through RFID tags and barcodes;

[0037] If the verification is successful, the materials will be delivered to the outbound area; if the verification fails, the materials will be picked again;

[0038] Based on the verification, a picking verification record is generated;

[0039] Use automated packaging equipment to package the picked materials, generate shipping documents, record shipping information, and update the inventory management center;

[0040] Generate outbound records and shipping records based on shipping documents and outbound records.

[0041] In a second aspect, the present invention provides a consignment management system, including a data collection module, a construction module, an adjustment module, a monitoring and updating module and a management module; the data collection module is used to collect multi-source data and pre-process the data; the construction module is used to extract features from the data and construct a multi-level memory unit LSTM model; the adjustment module is used to predict future consignment material demand through the multi-level memory unit LSTM model and automatically adjust inventory management; the monitoring and updating module is used to monitor the multi-level memory unit LSTM model according to the difference between actual sales and demand forecast data, and update the multi-level memory unit LSTM model; the management module is used to automate warehousing and outbound delivery based on adjusted inventory management, generate relevant records, and update the inventory management center.

[0042] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the consignment management method as described in the first aspect of the present invention is implemented.

[0043] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the consignment management method as described in the first aspect of the present invention is implemented.

[0044] The beneficial effects of the present invention are as follows: by constructing a multi-level memory unit LSTM model and defining multiple MLMU-LSTM layers and Dropout layers, the long-term memory capacity and anti-overfitting ability of the model are enhanced; a standardized clean data set is input into the LSTM model for training, and the model performance is optimized by adjusting the hyperparameters to ensure the generalization ability of the model in different scenarios; finally, through the accurate prediction of the multi-level memory unit LSTM model, inventory management can be effectively guided, inventory backlogs and out-of-stock phenomena can be reduced, and the efficiency and accuracy of inventory management can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0046] Figure 1 This is a flow chart of the consignment management method in Example 1.

[0047] Figure 2 This is a module diagram of the consignment management system in Example 1. DETAILED DESCRIPTION

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0049] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0050] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0051] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a consignment management method, comprising the following steps:

[0052] S1. Collect multi-source data and pre-process the data.

[0053] Furthermore, historical sales data, market trend data, weather data, holiday information and promotion activity data are collected from ERP systems, market research reports, meteorological websites, holiday information platforms and promotion activity records.

[0054] It should be noted that the ERP system is an enterprise resource planning system, which is an integrated management information system used to integrate and manage various business processes of an enterprise, including finance, procurement, production, inventory, sales, etc. It provides detailed operational data within the enterprise and helps to understand the company's sales trends, inventory status and procurement needs.

[0055] Market research reports are reports compiled through market surveys and research, containing information such as market trends, consumer behavior, and competitor analysis, providing macro information on the external market and helping to predict future market demand and industry trends;

[0056] Meteorological websites are websites or services that provide weather forecasts and historical weather data. Weather conditions have a significant impact on the sales of certain products, such as rain gear, air conditioners, etc. Meteorological data can help more accurately predict sales for a specific time period.

[0057] The holiday information platform provides information about holiday arrangements and related activities, including national statutory holidays, local holidays, etc. Holidays usually affect consumers' purchasing behavior. These data help predict sales peaks and troughs during holidays.

[0058] Promotion records record detailed information about all previous promotions of an enterprise, including promotion time, promotion method, promotion products, promotion effect, etc. Promotions have a significant impact on sales in the short term. These records help analyze promotion effects and optimize future promotion strategies.

[0059] Historical sales data refers to the sales records of an enterprise over a period of time in the past, including sales quantity, sales amount, sales time, sales location, etc. The data includes daily / weekly / monthly sales volume, sales amount, product category, sales channel, customer information, etc. Historical sales data is the basis for predicting future sales. By analyzing past sales trends, we can find seasonal changes, cyclical fluctuations and other patterns, providing important reference for demand forecasting;

[0060] Market trend data is data that reflects the overall development trend of the market, including industry growth rate, changes in consumer behavior, competitor dynamics, etc. The data includes market growth rate, consumer preferences, competitor market share, industry reports, market research results, etc. It mainly helps to understand the changes in the entire industry and the evolution of consumer behavior, helps to predict future market demand changes, and formulate more reasonable sales strategies;

[0061] Weather data is data describing weather conditions, including temperature, humidity, precipitation, wind speed, weather conditions, etc. The data includes daily / weekly / monthly temperature, humidity, precipitation, wind speed, weather conditions (sunny, rainy, snowy, etc.), which mainly has a significant impact on the sales of certain products, such as rain gear, air conditioners, winter clothing, etc. By analyzing weather data, sales in a specific time period can be more accurately predicted, especially for seasonal products;

[0062] Holiday information refers to holiday arrangements and related information, including statutory holidays, local holidays, special events, etc. The data includes holiday dates, holiday lengths, special event arrangements, holiday consumption habits, etc. Holidays usually lead to changes in consumer purchasing behavior, such as an increase or decrease in sales during holidays. By analyzing holiday information, sales peaks and troughs during holidays can be predicted, and inventory and promotion activities can be arranged reasonably.

[0063] Promotional activity data records detailed information on a company's previous promotional activities, including promotion time, promotion methods, promotional products, promotion effects, etc. The data includes the specific content of the promotional activities (such as discounts, buy one get one free), sales data during the promotion period, customer feedback, promotional expenses, etc. Promotional activities have a significant impact on sales in the short term. By analyzing promotional activity data, we can evaluate promotional effects, optimize future promotional strategies, and improve sales performance.

[0064] S1.2. Furthermore, the collected data is cleaned to remove outliers and missing values ​​to ensure the quality of the data;

[0065] Standardize the data so that it is on the same scale to facilitate subsequent model training;

[0066] Divide the data into training and test sets, ensure that the time sequence is not disrupted, and save the divided data into corresponding files or databases.

[0067] It should be noted that by collecting comprehensive historical sales data, market trend data, weather data, holiday information and promotion activity data from ERP systems, market research reports, meteorological websites, holiday information platforms and promotion activity records, the diversity and integrity of the data are ensured; the collected data are cleaned to remove outliers and missing values ​​to ensure data quality; data standardization makes the data on the same scale to facilitate subsequent model training; the data are divided into training sets and test sets to ensure that the time sequence is not disrupted, which improves the accuracy and generalization ability of model training; finally, high-quality data preprocessing lays a solid foundation for the high-precision prediction of the model.

[0068] S2. Extract features from the data and build a multi-level memory unit LSTM model.

[0069] Furthermore, time series features, external features, and lag features are extracted from the preprocessed data;

[0070] It should be noted that the time series feature data are timestamp, day, week, month, and quarter;

[0071] External feature data include weather data, holiday information, and promotional activities;

[0072] The lagged features are the sales data of the previous week and month so that the model can capture historical sales trends;

[0073] The extracted time series features, external features, and lag features are saved into feature datasets, and the feature datasets are converted into time series data, i.e., a three-dimensional array (number of samples, time steps, number of features);

[0074] Define the first layer of multi-level memory unit LSTM model hidden number as the maximum value, capture high-level features, use tanh activation function, set true to pass to the next layer;

[0075] Define the hidden number of the second-layer multi-level memory unit LSTM model as the median, further refine the features, use the tanh activation function, and set true to pass it to the next layer;

[0076] Define the hidden number of the third-layer multi-level memory unit LSTM model as the minimum value to capture the finest-grained features, use the tanh activation function, and set False as the output of the multi-level memory unit LSTM model;

[0077] Define the Dropout layer of the three-layer multi-level memory unit LSTM model as a fixed value to reduce the fitting risk;

[0078] It should be noted that the number of hidden units of the first layer of MLMU-LSTM is defined as 128, the activation function is tanh, and the return sequence (return_sequences=True) is used to pass the output to the next layer;

[0079] The number of hidden units of the second layer of MLMU-LSTM is 64, the activation function is tanh, the sequence is returned (return_sequences=True), and the output is passed on;

[0080] The number of hidden units of the MLMU-LSTM in the third layer is 32, the activation function is tanh, and the return sequence (return_sequences=False) is used for the final output through the fully connected layer;

[0081] Define the Dropout layer. The Dropout layer is a regularization technique that randomly discards some neurons to reduce the risk of overfitting and improve the generalization ability of the model.

[0082] First layer Dropout: Add a Dropout layer after the first layer of MLMU-LSTM and set the dropout rate (e.g. 0.2);

[0083] The second layer of Dropout: add a Dropout layer after the second layer of MLMU-LSTM and set the dropout rate (for example, 0.2);

[0084] The third layer of Dropout, add a Dropout layer after the third layer of MLMU-LSTM and set the dropout rate (for example, 0.2).

[0085] The preprocessed time series data is input into the multi-level memory unit LSTM model for training, and the hyperparameters are adjusted through cross-validation and grid search to optimize the model performance;

[0086] Save the trained multi-level memory unit LSTM model to a file, and save the metadata of the multi-level memory unit LSTM model to the database.

[0087] It should be noted that by extracting time series features (timestamp, day, week, month and quarter), external features (weather data, holiday information and promotional activities) and lagged features (sales data of the previous week and the previous month) from the preprocessed data, and constructing a multi-level memory unit LSTM model, the extraction and combination of these features enables the model to comprehensively capture the historical sales trends and the impact of the external environment, improve the accuracy and robustness of the prediction, define multiple MLMU-LSTM layers and Dropout layers, enhance the long-term memory capacity and anti-overfitting ability of the model, input the standardized clean data set into the model for training, and optimize the model performance by adjusting the hyperparameters, and finally save the trained model to the database.

[0088] S3. Use the multi-level memory unit LSTM model to predict future consignment material demand and automatically adjust inventory management.

[0089] Furthermore, the future demand for consignment materials is predicted through the trained multi-level memory unit LSTM model;

[0090] Input time series data for a period of time in the future, and the model outputs future demand forecast data;

[0091] Save the demand forecast data to the database and generate a visualization chart of the demand forecast data to intuitively display the forecast results;

[0092] According to the demand forecast data, the inventory level is automatically adjusted, replenishment and reduction instructions are generated, and the instructions are sent through the inventory management center. The expression is:

[0093] ;

[0094] in, is the adjusted inventory. The final value after inventory adjustment is subject to the joint constraints of inventory upper limit (U), safety stock (S) and dynamic adjustment strategy. is the current inventory, indicating the current actual inventory quantity. It is the forecast demand, which means the future demand based on historical data or forecast model. is the adjustment factor used to control the difference between inventory adjustment and demand sensitivity, The larger the value, the greater the adjustment. It is the upper limit of inventory, indicating the maximum storage capacity of the warehouse. The adjusted inventory cannot exceed this value. It is the safety stock value, which indicates the minimum inventory level to ensure that the inventory will not fall below this value to cope with sudden demand or unpredictable supply chain problems. is an exponential decay function used to dynamically adjust inventory. is the attenuation coefficient, controlling the adjustment speed, The larger it is, the faster it adjusts and the less sensitive the inventory is to future demand forecasts. is a time variable, indicating the time span of the forecast (such as days, hours, etc.), which affects the dynamic changes of inventory adjustments. is the minimum function, is the maximum value function;

[0095] This formula can be used to achieve dynamic inventory adjustment, ensuring that the inventory level is between the safety stock S and the inventory limit U, combined with forecast demand and time decay Optimize replenishment to avoid out-of-stock risks and excessive inventory backlogs, thereby improving inventory management efficiency;

[0096] ;

[0097] in, is the historical inventory change rate, is the historical sales change rate.

[0098] It should be noted that the trained multi-level memory unit LSTM model has achieved high-precision prediction of future consignment material demand, combined with time series characteristics, external characteristics and lag characteristics, to improve the accuracy and robustness of the prediction. The time series data for a period of time in the future is input, and the model outputs future demand forecast data, ensuring the real-time and accuracy of the prediction results. The demand forecast data is saved in the database and a visual chart is generated, providing intuitive decision support for the management, improving the transparency and scientificity of the decision. According to the forecast data, the inventory level is automatically adjusted, and replenishment and reduction inventory instructions are generated to ensure the dynamic and intelligent inventory management.

[0099] S4. According to the difference between the actual sales and demand forecast data, the multi-level memory unit LSTM model is monitored and updated.

[0100] Furthermore, the difference between actual sales data and demand forecast data can be monitored in real time, and a threshold can be set. When the difference exceeds the threshold, an alarm is triggered;

[0101] The difference between actual sales data and demand forecast data is recorded in the database, and a visualization chart of monitoring results is generated to intuitively display the monitoring results;

[0102] Regularly update the multi-level memory unit LSTM model and retrain the multi-level memory unit LSTM model according to the new time series data to improve the prediction accuracy. The expression is:

[0103] ;

[0104] in, is the new learning rate, which indicates the new learning rate used in the current training phase after dynamic adjustment. is the current learning rate, which indicates the learning rate being used in the current training phase, is the validation set loss, which represents the loss value calculated on the validation set and is used to measure the generalization ability of the model. is the training set loss, which represents the loss value calculated on the training set and is used to measure the degree of fit of the model on the training data. is an exponential decay factor, which means that the learning rate is adjusted dynamically according to the difference between the validation set and the training set loss. The larger the difference, the smaller the update. It is the absolute difference between the validation set loss and the training set loss, which is used to measure the degree of overfitting of the model. The larger the difference, the more serious the overfitting of the model. It is a natural logarithmic function, which is used to amplify the effect of the difference between the validation set and the training set loss on the learning rate adjustment. is the training progress rate, which indicates the completion ratio of training, ranging from [0, 1]. is a stage control parameter used to control the rate of change of learning rate updates at different training stages. is a piecewise function of the training phase, indicating the phased nature of the training progress. The value is rounded down. Is to achieve sub-update, It is the natural logarithm of the current learning rate, which is used to dynamically adjust the learning rate amplitude to ensure that the adjustment amplitude changes with the current learning rate;

[0105] This formula can be used to dynamically adjust the learning rate based on the difference in loss between the training set and the validation set. and training progress Flexible optimization of learning rate can accelerate model convergence, suppress overfitting, and improve training efficiency and model performance;

[0106] It should be noted that the setting of the threshold is explained with examples as follows:

[0107] Suppose a retail company decides to use the percentage method combined with the dynamic adjustment method to set the threshold when conducting demand forecasting;

[0108] By analyzing the data from the past year, it was found that the average difference between sales data and forecast data was 5%, with a standard deviation of 2%;

[0109] Then based on historical data, the threshold is initially set at ±10% of the forecast demand;

[0110] Considering the greater volatility of air conditioner sales in the summer, management decided to relax the threshold to ±15% in the summer;

[0111] Then for high-value goods (such as high-end home appliances), a stricter threshold is set, such as ±8%, even in summer;

[0112] Based on the above, set the final threshold:

[0113] For general commodities (relatively low prices), if the difference between actual sales data and forecast data exceeds ±10%, an alarm is triggered;

[0114] For summer air conditioners, if the difference between actual sales data and forecast data exceeds ±15%, an alarm is triggered;

[0115] For high-value items, an alert is triggered if the difference between actual sales data and forecast data exceeds ±8%.

[0116] It should be noted that through the trained multi-level memory unit LSTM model, high-precision prediction of future consignment material demand is made. The accuracy and robustness of the prediction are improved by combining time series data, external characteristics and lag characteristics. The demand forecast data is saved in the database and visual charts are generated, providing intuitive decision support for management. According to the prediction results, the inventory level is automatically adjusted, and replenishment and inventory reduction instructions are generated to ensure the dynamic and intelligent inventory management.

[0117] S5. Based on the adjustment of inventory management, automate warehousing and outbound delivery, generate relevant records, and update the inventory management center.

[0118] Furthermore, after receiving the replenishment instruction, the inventory management center automatically generates a purchase order;

[0119] It should be noted that after receiving the replenishment instruction, the inventory management center automatically generates a purchase order based on the preset inventory threshold and demand forecast data, calculates the quantity to be purchased based on the current inventory level, forecast demand and safety stock, and generates a purchase order. The purchase order will contain the name, quantity, supplier information, etc. of the material.

[0120] When receiving consignment materials from suppliers, the material information is identified through RFID tags and barcodes;

[0121] It should be noted that RFID tags and barcodes are commonly used material identification technologies. RFID tags are read through wireless radio frequency signals and barcodes are read through scanners. When the consignment materials sent by suppliers arrive at the warehouse, the staff uses RFID readers or barcode scanners to read the labels or barcodes on the materials, automatically obtain the name, quantity, supplier information, etc. of the materials, and record this information in the inventory management center.

[0122] Record material information to the inventory management center, including the name, quantity, storage time, etc. of the material, and generate storage records, which are saved in the database to ensure the integrity and accuracy of the storage information;

[0123] Deliver materials to designated locations through automated equipment and record storage location and status, update inventory records in the inventory management center to ensure the accuracy of inventory information, generate storage records, save them in the database, and ensure the integrity and accuracy of storage information;

[0124] It should be noted that automated equipment (such as AGV carts, stackers, etc.) will transport materials to designated storage locations based on the generated storage tasks. The storage locations will be pre-allocated and this information will be sent to the automated equipment. After receiving the task, the equipment will transport the materials to the designated location along the predetermined path and record the storage location and status.

[0125] Automatically generate picking tasks based on outbound orders and save picking task records in the database to ensure the completeness and accuracy of picking tasks;

[0126] AGVs or other automated equipment pick materials from shelves according to task instructions and verify them through RFID tags and barcodes;

[0127] If the verification is passed, the materials will be transported to the outbound area. If the verification fails, the materials will be picked again or notified for manual intervention.

[0128] It should be noted that if the verification fails, the failure information will be recorded and the manual intervention personnel will be notified, and the manual intervention personnel will re-pick or check the material information;

[0129] Based on the verification, a picking verification record is generated and saved in the database;

[0130] Use automated packaging equipment to package the picked materials, generate shipping documents, record shipping information, and update the inventory management center;

[0131] It should be noted that the shipping document contains the name, quantity, consignee information, etc. of the materials;

[0132] Save the shipping documents and outbound records in the database to ensure the integrity and accuracy of the data, and generate outbound and shipment records and save them in the database to ensure the integrity and accuracy of outbound and shipment.

[0133] It should be noted that through automated warehousing and outbound management, the efficiency and accuracy of inventory management are achieved. After receiving the replenishment instruction, the inventory management center automatically generates a purchase order to ensure the timeliness of replenishment. The material information is identified through RFID tags and barcodes, and the material information is recorded in the inventory management center to generate warehousing records to ensure the integrity and accuracy of the warehousing information. The automated equipment transports the materials to the designated location and records the storage location and status, updates the inventory records to ensure the accuracy of the inventory information, and automatically generates picking tasks based on the outbound order, and picks them through AGV carts or other automated equipment to ensure the integrity and accuracy of the picking tasks. After verification, the materials are transported to the outbound area, and a picking verification record is generated. The materials are packaged through automated packaging equipment, and a shipping document is generated to record the shipping information. The inventory management center is updated to ensure the integrity and accuracy of outbound and delivery. Ultimately, this series of automated operations significantly improves the efficiency of warehousing and outbound, reduces human errors, ensures the integrity and accuracy of data, and improves overall operational efficiency.

[0134] The present embodiment also provides a consignment management system, including: a data collection module, a construction module, an adjustment module, a monitoring and updating module and a management module; the data collection module is used to collect multi-source data and pre-process the data; the construction module is used to extract features from the data and build a multi-level memory unit LSTM model; the adjustment module is used to predict future consignment material demand through the multi-level memory unit LSTM model and automatically adjust inventory management; the monitoring and updating module is used to monitor the multi-level memory unit LSTM model according to the difference between actual sales and demand forecast data, and update the multi-level memory unit LSTM model; the management module is used to automate warehousing and outbound delivery based on adjusted inventory management, generate relevant records, and update the inventory management center.

[0135] This embodiment also provides a computer device suitable for the consignment management method, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to implement the consignment management method proposed in the above embodiment.

[0136] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0137] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the consignment management method proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0138] In summary, the present invention enhances the long-term memory capacity and anti-overfitting ability of the model by constructing a multi-level memory unit LSTM model and defining multiple MLMU-LSTM layers and Dropout layers; inputs a standardized clean data set into the LSTM model for training, and optimizes the model performance by adjusting hyperparameters to ensure the generalization ability of the model in different scenarios; finally, through the precise prediction of the multi-level memory unit LSTM model, it can effectively guide inventory management, reduce inventory backlogs and out-of-stock phenomena, and improve the efficiency and accuracy of inventory management.

[0139] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of the consignment management method are provided.

[0140] First, the experimental team collected detailed historical sales data, market trend data, weather data, holiday information and promotional activity data from multiple channels such as the target retail companies' ERP systems, market research reports, meteorological websites, holiday information platforms and promotional activity records, ensuring the diversity and richness of data sources.

[0141] Secondly, the collected data were strictly preprocessed, including cleaning the data to remove outliers and missing values, standardizing the data to make them on the same scale to facilitate subsequent modeling, and dividing the data set into a training set (70%) and a test set (30%) in chronological order to maintain the continuity of the time series.

[0142] Next, time series features (such as date, week, month, etc.), external features (such as weather conditions, holiday schedules, etc.) and lagged features (such as sales data from the previous week and month) were extracted from the preprocessed data and converted into a three-dimensional array suitable for use in the LSTM model.

[0143] Finally, a multi-level memory unit LSTM model was constructed. The model consists of three LSTM layers, each followed by a Dropout layer to avoid overfitting. After sufficient training, the model was applied to predict daily demand in the next month. The prediction results were not only saved in the database, but also generated visual charts. Based on the prediction results, the inventory level was automatically adjusted and replenishment or reduction instructions were issued to the inventory management center.

[0144] The details are shown in Table 1 below:

[0145] Table 1 Demand forecasting model performance comparison

[0146]

[0147] The table compares four demand forecasting models, namely the traditional ARIMA model, the single LSTM model, the multi-level memory unit LSTM model and the improved multi-level memory unit LSTM model. Through the four key indicators of prediction accuracy, response time, inventory adjustment error and model training time, it can be seen that with the advancement of model complexity and technology, the prediction accuracy and response speed have been significantly improved, and the practicality and efficiency of the model have also been enhanced.

[0148] It is worth noting that although the improved multi-level memory unit LSTM model achieved the highest prediction accuracy of 94.8%, it also showed outstanding performance in other aspects, such as shorter response time and extremely low inventory adjustment error, which shows that the model not only surpasses other methods numerically, but also truly provides enterprises with a more reliable and efficient solution. The continuous reduction in the time spent on model training means that technological advances not only improve the performance of the model, but also reduce the computing cost and time investment required to achieve these high performances, which is especially important for enterprises pursuing rapid iteration and agile response.

[0149] In addition, the single LSTM model has obvious advantages over the traditional ARIMA, and the multi-level memory unit LSTM and its improved version further strengthen these advantages. The hierarchical structural design enables the model to better handle complex patterns in time series, especially when facing multi-source heterogeneous data, and can effectively capture potential correlations and trends. By introducing regularization methods such as the Dropout layer, the model not only enhances the learning ability, but also ensures stability and generalization ability, making the prediction results more robust and better able to cope with uncertainties in practical applications.

[0150] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A consignment management method, characterized by: include, Collect multi-source data and pre-process the multi-source data; Extract features from data and build a multi-level memory unit LSTM model; The multi-level memory unit LSTM model is used to predict future consignment material demand and automatically adjust inventory management; According to the difference between actual sales and demand forecast data, the multi-level memory unit LSTM model is monitored and updated; Based on adjusting inventory management, automate warehousing and outbound delivery, generate relevant records, and update the inventory management center; The data is subjected to feature extraction to construct a multi-level memory unit LSTM model, and the specific steps are: Extract time series features, external features, and lag features from the preprocessed data; The extracted time series features, external features and lag features are saved into a feature data set, and the feature data set is converted into time series data; Define the first layer of multi-level memory unit LSTM model hidden number as the maximum value, use the tanh activation function, set true to pass it to the next layer; Define the hidden number of the second-layer multi-level memory unit LSTM model as the median, use the tanh activation function, and set true to pass it to the next layer; Define the hidden number of the third-layer multi-level memory unit LSTM model as the minimum value, use the tanh activation function, and set False as the output of the multi-level memory unit LSTM model; Define the Dropout layer of the three-layer multi-level memory unit LSTM model as a fixed value; The preprocessed time series data is input into the multi-level memory unit LSTM model for training, and the hyperparameters are adjusted through cross-validation and grid search; The specific steps of predicting future consignment material demand and automatically adjusting inventory management are as follows: Through the trained multi-level memory unit LSTM model, the future demand for consignment materials is predicted; Input time series data for a period of time in the future, and the model outputs future demand forecast data; According to the demand forecast data, the inventory level is automatically adjusted, replenishment and reduction instructions are generated, and the instructions are sent through the inventory management center. The expression is: ； in, is the adjusted inventory, is the upper limit of inventory. is the current inventory, is to predict demand, is the adjustment factor, is the safety stock value, is an exponential decay function, is the attenuation coefficient, is the time variable, is the minimum function, is the maximum value function; The multi-level memory unit LSTM model is monitored and updated, and the specific steps are: Monitor the difference between actual sales data and demand forecast data in real time, set thresholds, and trigger alarms when the difference exceeds the threshold; The multi-level memory unit LSTM model is updated regularly and retrained according to the new time series data. The expression is: ； in, is the new learning rate, is the current learning rate, is the validation set loss, is the training set loss, is the exponential decay factor, is the natural logarithm function, is the piecewise function during the training phase, is the training progress rate, is the stage control parameter, is the natural logarithm of the current learning rate; After updating the multi-level memory unit LSTM model, re-perform demand forecast; The collected multi-source data includes historical sales data, market trend data, weather data, holiday information and promotion activity data.

2. The consignment management method according to claim 1, characterized in that: The preprocessing includes cleaning the data, removing outliers, missing values ​​and data standardization.

3. The consignment management method according to claim 2, characterized in that: The automated warehousing and outbound delivery, and the generation of relevant records, update the inventory management center, the specific steps are: After receiving the replenishment instruction, the inventory management center automatically generates a purchase order; When receiving consignment materials from suppliers, the material information is identified through RFID tags and barcodes; Record material information to the inventory management center and generate inventory records; Deliver materials to designated locations through automated equipment and record storage location and status, update inventory records at the inventory management center, and generate storage records; Automatically generate picking tasks based on outbound orders; Pick materials from shelves according to task instructions and verify through RFID tags and barcodes; If the verification is successful, the materials will be delivered to the outbound area; if the verification fails, the materials will be picked again; Based on the verification, a picking verification record is generated; Use automated packaging equipment to package the picked materials, generate shipping documents, record shipping information, and update the inventory management center; Generate outbound records and shipping records based on shipping documents and outbound records.

4. A consignment management system, based on the consignment management method according to any one of claims 1 to 3, characterized in that: Including data collection module, construction module, adjustment module, monitoring and updating module and management module; The data collection module is used to collect multi-source data and pre-process the data; The construction module is used to extract features from data and construct a multi-level memory unit LSTM model; The adjustment module is used to predict the future demand for consignment materials through a multi-level memory unit LSTM model and automatically adjust inventory management; The monitoring and updating module is used to monitor the multi-level memory unit LSTM model according to the difference between the actual sales and demand forecast data, and update the multi-level memory unit LSTM model; The management module is used to automate warehousing and outbound delivery based on the adjustment of inventory management, generate relevant records, and update the inventory management center.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the consignment management method described in any one of claims 1 to 3 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the consignment management method described in any one of claims 1 to 3 are implemented.

Images