Supplier food erp full-chain inventory early warning system based on artificial intelligence

Abstract

The application provides an artificial intelligence-based supplier food ERP whole-chain inventory early warning system, relates to the technical field of food inventory monitoring, and comprises a data acquisition and processing module, an inventory prediction module, an inventory early warning module and an ERP execution module; the data acquisition and processing module is used for acquiring whole-chain inventory data and pre-processing the whole-chain inventory data; the inventory prediction module is used for predicting future inventory; the inventory early warning module is used for presetting inventory early warning rules and early warning future inventory; and the ERP execution module is used for generating a restocking strategy. The application accurately predicts future inventory changes by means of a time series model, a machine learning model and a fusion strategy, so that an enterprise can plan production and procurement in advance; potential inventory risks are discovered in time by means of multi-dimensional early warning rules and a risk assessment matrix, so as to reduce out-of-stock or overstock losses; and the ERP execution module generates an optimal restocking strategy based on a cost model and a linear programming algorithm, and the restocking strategy is dynamically adjusted in combination with a real-time state of a supply chain, so as to reduce the costs of procurement, transportation and warehousing while ensuring supply.

Description

Technical Field

[0001] This invention relates to the field of food inventory monitoring technology, and in particular to an artificial intelligence-based supplier food ERP full-chain inventory early warning system. Background Technology

[0002] With the rapid development of the food industry and the increasing complexity of the supply chain, supplier food inventory management faces numerous challenges and opportunities, and its technological background is constantly evolving. The food supply chain encompasses multiple stages from raw material procurement, production and processing, warehousing and logistics to the sales terminal, with frequent data interaction between each stage and rapidly changing inventory status. In recent years, information technology has been gradually applied to the field of food inventory management. Some companies have introduced ERP systems, achieving preliminary information management of inventory data, enabling digital storage and simple analysis of inventory quantities, inbound and outbound records, etc. Simultaneously, some basic inventory early warning technologies have emerged, issuing warning signals when inventory quantities exceed certain limits by setting fixed inventory thresholds, assisting companies in making inventory decisions. However, the food industry has its own unique characteristics. Food categories are numerous, and the shelf life, storage conditions, and market demand fluctuations vary greatly among different categories; moreover, the food supply chain is significantly affected by external factors such as seasonal changes, shifts in consumer preferences, and policy and regulatory adjustments. This requires inventory management technology not only to have data processing capabilities but also to deeply analyze the characteristics of the food industry and combine market dynamics for accurate forecasting and intelligent decision-making. Against this backdrop, building a full-chain inventory early warning system based on artificial intelligence technology has become a key direction for food suppliers to improve their inventory management level and enhance their supply chain competitiveness. The aim is to achieve refined and intelligent management of food inventory through advanced algorithms and models.

[0003] Typical food inventory early warning systems often employ simple statistical methods or fixed models, failing to incorporate in-depth feature extraction and analysis of data in conjunction with market dynamics. They also lack model fusion strategies, making it difficult to accurately predict fluctuations in food category demand and easily leading to inventory buildup or stockouts. Furthermore, early warning rules are often formulated solely based on inventory quantity thresholds, without considering multiple dimensions such as food category characteristics and supply chain risks, thus failing to promptly identify potential risks. Early warning responses are also insufficiently timely, lacking tiered processing of warning information and making it difficult to efficiently address situations of varying urgency.

[0004] To address the shortcomings of the existing technologies, this technical solution proposes an artificial intelligence-based supplier food ERP full-chain inventory early warning system. Summary of the Invention

[0005] This invention provides an artificial intelligence-based supplier food ERP full-chain inventory early warning system to address the shortcomings of existing technologies.

[0006] On the one hand, this invention provides an artificial intelligence-based supplier food ERP full-chain inventory early warning system, including:

[0007] The data acquisition and processing module is used to collect full-chain inventory data and preprocess the full-chain inventory data to generate standardized inventory data.

[0008] The inventory forecasting module is used to forecast future inventory based on standardized inventory data and market data, and output the inventory forecast results.

[0009] The inventory early warning module is used to preset inventory early warning rules and, in conjunction with inventory forecast results, issue early warnings about future inventory levels and output early warning logs.

[0010] The ERP execution module is used to generate replenishment strategies based on the warning logs.

[0011] According to the artificial intelligence-based supplier food ERP full-chain inventory early warning system provided by the present invention, the steps of the data acquisition and processing module outputting standardized inventory data include:

[0012] By using a rules engine to identify and correct abnormal data in the entire chain of inventory data, preprocessed inventory data is obtained.

[0013] A distributed file system is used to store the cleaned preprocessed inventory data;

[0014] Establish a data quality scoring system to quantitatively assess data quality and output standardized inventory data.

[0015] According to the artificial intelligence-based supplier food ERP full-chain inventory early warning system provided by the present invention, the steps of identifying and correcting abnormal data in the data acquisition and processing module include:

[0016] Based on historical databases, reasonable ranges and constraints are set for various types of data; and historical abnormal data are classified.

[0017] Based on the type of abnormal data, a structured language is used to write rules and output abnormal rules; and a rule base is established to classify, store and manage the versions of abnormal rules.

[0018] Based on anomaly rules, abnormal data in the entire chain inventory data is identified, and the identification results are output.

[0019] Based on the identification results and combined with the inbound and outbound data, abnormal data is corrected.

[0020] According to the AI-based supplier food ERP full-chain inventory early warning system provided by the present invention, the data processing unit identifies and corrects abnormal data in the full-chain inventory data through a rule engine, including the following steps:

[0021] Based on historical databases, reasonable ranges and constraints are set for various types of data; and historical abnormal data are classified.

[0022] Based on the abnormal data type, a structured language is used to write rules and output abnormal rules; and a rule base is established to classify, store and manage the versions of abnormal rules.

[0023] Based on anomaly rules, abnormal data in the entire chain inventory data is identified, and the identification results are output.

[0024] Based on the identification results and combined with the inbound and outbound data, abnormal data is corrected.

[0025] According to the artificial intelligence-based supplier food ERP full-chain inventory early warning system provided by the present invention, the inventory prediction module includes a feature extraction unit, a model building unit, and a predictive analysis unit. The feature extraction unit is used to extract and derive features from standardized inventory data by combining market data to obtain standard data features. The model building unit is used to establish a time series model and a machine learning model based on a historical warehousing database, and to train the model on the demand fluctuation characteristics of food categories through a model fusion strategy to obtain a fused prediction model. The predictive analysis unit is used to update the fused prediction model in real time based on the full-chain inventory data using a rolling prediction mechanism and output the inventory prediction results.

[0026] According to the AI-based supplier food ERP full-chain inventory early warning system provided by the present invention, the inventory early warning module includes an early warning rule formulation unit, an early warning triggering unit, and an early warning response unit; the early warning rule formulation unit is used to preset multi-dimensional inventory early warning rules; the early warning triggering unit is used to monitor the full-chain inventory data in real time based on the inventory forecast results, determine whether the early warning rules are met, perform composite early warning judgment and trace the cause of the early warning, and generate early warning analysis data; the early warning response unit is used to organize the early warning analysis data into an early warning log and classify the early warning log into levels of urgency.

[0027] According to the artificial intelligence-based supplier food ERP full-chain inventory early warning system provided by the present invention, the step of the early warning rule formulation unit to preset multi-dimensional inventory early warning rules includes:

[0028] Based on the characteristics of food categories, a dynamic safety stock threshold model is established, and the safety stock range is adjusted in real time according to the basic factors and market factors of food.

[0029] Construct a supply chain risk assessment matrix and, in conjunction with external factors, generate risk-weighted early warning rules;

[0030] By comparing and analyzing historical early warning data with actual business results, the parameters of the risk-weighted early warning rules are automatically adjusted.

[0031] According to the AI-based supplier food ERP full-chain inventory early warning system provided by the present invention, the ERP execution module includes a log parsing unit, a strategy generation unit, and a strategy execution unit. The log parsing unit is used to classify and organize the early warning logs and associate them with supplementary basic transportation data to obtain complete early warning logs. The strategy generation unit is used to automatically generate replenishment strategies based on the complete early warning logs, through a preset basic replenishment strategy library and an intelligent strategy matching algorithm, combined with cost factors. The strategy execution unit is used to convert the replenishment strategies into business instructions, link the ERP system to execute the instructions, track the execution progress in real time, and provide feedback on the execution results.

[0032] According to the artificial intelligence-based supplier food ERP full-chain inventory early warning system provided by the present invention, the step of the strategy generation unit automatically generating a replenishment strategy based on complete early warning logs includes:

[0033] By using a preset strategy mapping table, different early warning categories are automatically associated with basic replenishment strategies;

[0034] Establish a multi-dimensional cost model and update cost coefficients in real time based on market dynamics to output cost analysis results;

[0035] Based on the linear programming algorithm and the cost analysis results, the optimal replenishment plan is solved to obtain the initial replenishment strategy; and the feasibility of supplier capacity, logistics network and warehouse space is verified.

[0036] The emergency strategy is automatically switched according to the real-time status of the supply chain, and the safety stock coefficient is dynamically adjusted in combination with seasonal demand fluctuations.

[0037] Generate a comprehensive recommendation report that includes replenishment quantity, timing, and supplier selection.

[0038] The supplier food ERP full-chain inventory early warning system based on artificial intelligence provided by the present invention also includes a system monitoring and maintenance module. The system monitoring and maintenance module is used to monitor the operating status of the system in real time, and to monitor and analyze the system performance indicators in real time. When a system abnormality is detected, the fault diagnosis process is automatically triggered to quickly locate the fault point and generate maintenance suggestions.

[0039] This invention provides an AI-based supplier food ERP full-chain inventory early warning system. Through its inventory forecasting module, leveraging time series models, machine learning models, and fusion strategies, it accurately predicts future inventory changes, enabling companies to plan production and procurement in advance. By employing multi-dimensional early warning rules and a risk assessment matrix, it promptly identifies potential inventory risks, reducing losses from stockouts or overstocking. The ERP execution module generates optimal replenishment strategies based on cost models and linear programming algorithms, dynamically adjusting these strategies in conjunction with real-time supply chain status, ensuring supply while reducing costs related to procurement, transportation, and warehousing. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram of the structure of the supplier food ERP full-chain inventory early warning system based on artificial intelligence provided in an embodiment of the present invention;

[0042] Figure 2 This is a flowchart of the pre-setting multi-dimensional inventory early warning rules provided by the early warning rule formulation unit in this embodiment of the invention;

[0043] Figure 3 This is a flowchart of the strategy generation unit automatically generating a replenishment strategy based on complete early warning logs, provided in an embodiment of the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0045] Example 1:

[0046] The following is combined Figures 1-3 This invention describes an artificial intelligence-based supplier food ERP full-chain inventory early warning system.

[0047] like Figure 1 As shown in the embodiment of the present invention, the supplier food ERP full-chain inventory early warning system based on artificial intelligence includes:

[0048] The data acquisition and processing module is used to collect inventory data across the entire supply chain and preprocess this data to generate standardized inventory data. The module's scope covers raw material procurement (quantity, batches, supplier information); production and processing (raw material consumption, semi-finished product inventory); warehousing and logistics (inventory quantity, storage location, inbound / outbound time); and sales data from retail outlets. Data collection methods include IoT sensors, API interfaces with enterprise management systems, and manual data entry, ensuring the comprehensiveness and real-time nature of the data.

[0049] The steps by which the data acquisition and processing module outputs standardized inventory data include:

[0050] Preprocessed inventory data is obtained by identifying and correcting anomalous data in the entire supply chain inventory data using a rules engine. The steps for identifying and correcting anomalous data in the entire supply chain inventory data using a rules engine include:

[0051] Based on historical databases, reasonable ranges and constraints are set for various types of data. Historical anomalies are also categorized. For example, for food shelf-life data, minimum remaining shelf-life thresholds are set according to different food categories; for inventory quantity data, upper and lower limits are set based on historical sales data and production plans. Furthermore, historical anomalies are meticulously categorized according to data type, cause, and other dimensions, such as missing data, numerical errors, and logical contradictions, for targeted processing later.

[0052] Based on the abnormal data type, a structured language is used to write rules and output abnormal rules. A rule base is established to categorize, store, and manage the versions of abnormal rules. Specific anomaly detection rules are written using Python's Pandas library or SQL, such as using conditional statements to determine whether inventory quantities exceed a set range. The rule base adopts a hierarchical architecture, categorizing and storing rules according to food category, data source, and anomaly type, while also recording rule version numbers, creation times, and modification records for easy rule updates and traceability.

[0053] Based on anomaly rules, the system identifies anomalous data in the entire supply chain inventory data and outputs the identification results. The rule engine automatically traverses the entire supply chain inventory data, matches it according to the anomaly rules, and compiles the identified anomalous data and its related information, such as data source, anomaly type, and involved business processes, into an identification result report.

[0054] Based on the identification results and inbound / outbound data, abnormal data is corrected. For missing data, methods such as mean imputation and regression prediction are used to supplement it; for data with numerical errors, cross-checking with upstream and downstream data and correcting it in conjunction with business logic are conducted; for data with logical contradictions, the data source is traced back, and adjustments are made after communication and confirmation with relevant business departments.

[0055] A distributed file system is used to store the cleaned preprocessed inventory data. Distributed storage systems such as Hadoop Distributed File System (HDFS) or Ceph are selected to partition the preprocessed inventory data according to a specific storage strategy, ensuring high availability, fault tolerance, and read / write performance, while also facilitating subsequent data querying and analysis.

[0056] Establish a data quality scoring system to quantitatively assess data quality and output standardized inventory data. Design scoring indicators from multiple dimensions, including data completeness, accuracy, consistency, and timeliness. For example, completeness is measured by the statistical data missing rate, and accuracy is assessed by comparing the differences between the data before and after correction. Assign weights to each indicator and calculate the data quality score using a weighted average method. Classify the data according to the scores, and identify data that meets certain quality standards as standardized inventory data.

[0057] The steps for establishing a data quality scoring system in the data acquisition and processing module include:

[0058] Define the data quality dimensions for preprocessed inventory data. These include: assessing for missing values ​​in the data, such as key fields like inventory quantity, batch number, and shelf life; verifying the match between the data and actual inventory status; checking for consistent standards across systems / departments; measuring data update frequency, such as the real-time nature of inventory change records and latency thresholds for market data; and confirming data format / value range compliance.

[0059] Assign quantitative metrics to each data quality dimension and set dimension weights. Dimension weights are typically set as follows: completeness 30%, accuracy 25%, consistency 20%, timeliness 15%, and validity 10%, but can be adjusted according to business priorities.

[0060] Establish a tiered scoring system and map the scores for each data quality dimension to quality levels, including Excellent, Good, Satisfactory, and Unsatisfactory. Good data requires monitoring, Satisfactory data requires source tracing and confirmation, and Unsatisfactory data triggers a data correction process.

[0061] A quality score for each quality level is calculated periodically using a sliding window. First, the time window size and time step are set. Then, based on the current time and window parameters, "full-chain inventory data" within the corresponding time range is automatically extracted. Next, for the data within the extracted window, scores for each indicator are calculated according to a preset data quality scoring system. Finally, the scores for each dimension are weighted and summed based on preset weights for each dimension to obtain the overall data quality score for that window.

[0062] The inventory forecasting module is used to predict future inventory levels based on standardized inventory data and market data, and outputs the inventory forecast results. Its core function lies in accurately predicting future inventory trends through in-depth analysis of massive amounts of data.

[0063] The inventory forecasting module comprises a feature extraction unit, a model building unit, and a predictive analysis unit. The feature extraction unit combines market data to extract and derive features from standardized inventory data, resulting in standardized data features. Market data covers multiple aspects, including macroeconomic indicators, industry development trends, changes in consumer preferences, and competitor dynamics. The feature extraction unit utilizes data mining techniques to extract key features from standardized inventory and market data, such as seasonal sales fluctuations and holiday consumption growth characteristics. These features are then used to derive new features through mathematical transformations and combinations, providing a more comprehensive reflection of the factors influencing inventory changes.

[0064] The model building unit is used to establish time-series and machine learning models based on historical warehouse databases. It trains these models on the fluctuation characteristics of food category demand using a model fusion strategy to obtain a fused prediction model. For the time-series model, models such as ARIMA and Prophet are used to capture the time-series patterns of inventory data. In the machine learning model, algorithms such as random forests, gradient boosting trees, and neural networks are applied to uncover complex relationships within the data. Through model fusion strategies, such as weighted average fusion, the prediction results of different models are integrated to fully leverage the advantages of each model and improve the accuracy and stability of predictions. Taking the ARIMA time-series model as an example, the formula is expressed as:

[0065]

[0066]

[0067]

[0068] Among them, X t Let B be the observation value of the time series at time t, and B be the lag operator, satisfying... . Here, θ(B) is an autoregressive polynomial, θ(B) is a moving average polynomial, p is the autoregressive order, d is the difference order, and q is the moving average order. It is a white noise sequence.

[0069] The model fusion strategy employs a weighted average fusion method. Specifically:

[0070]

[0071] Among them, Y t To fuse predicted values, n is the number of prediction models, i is the index of the prediction model, and y is the index of the prediction model. i,t Let w be the predicted value of the i-th model. i Let be the weights corresponding to the predicted values ​​of the i-th model. By adjusting the weights of each model, the advantages of each model can be fully utilized to improve the accuracy and stability of the predictions.

[0072] The predictive analytics unit employs a rolling forecasting mechanism to update the fusion forecasting model in real time based on end-to-end inventory data, outputting inventory forecast results. The rolling forecasting mechanism periodically retrains and adjusts the parameters of the fusion forecasting model based on the latest end-to-end inventory data, ensuring the model always adapts to dynamic changes in the market and business.

[0073] The inventory early warning module is used to preset inventory early warning rules and, based on inventory forecast results, issue early warnings for future inventory levels and output early warning logs. The inventory early warning module includes an early warning rule setting unit, an early warning triggering unit, and an early warning response unit. The early warning rule setting unit is used to preset multi-dimensional inventory early warning rules.

[0074] The early warning triggering unit monitors the entire supply chain inventory data in real time based on inventory forecasting results, determines whether early warning rules are met, performs a composite early warning judgment, traces the cause of the early warning, and generates early warning analysis data. This unit uses real-time data stream processing technology to analyze the entire supply chain inventory data in real time. Once the data is found to meet the early warning rule conditions, the composite early warning judgment process is immediately initiated. Through correlation analysis, causal analysis, and other methods, the root cause of the early warning is traced, such as supplier delays in delivery or a sudden surge in market demand.

[0075] The early warning response unit is used to organize early warning analysis data into early warning logs and classify the logs according to their urgency level. Based on factors such as the severity and scope of impact of the early warning, early warning logs are typically divided into three levels: urgent, severe, and general.

[0076] like Figure 2 As shown, the steps for the early warning rule formulation unit to preset multi-dimensional inventory early warning rules include:

[0077] Based on the characteristics of different food categories, a dynamic safety stock threshold model is established, and the safety stock range is adjusted in real time according to fundamental and market factors. Different food categories have different characteristics such as shelf life, sales cycle, and demand stability. For example, fresh foods have short shelf lives and fluctuating demand, while dried foods have long shelf lives and relatively stable demand. The dynamic safety stock threshold model comprehensively considers fundamental factors such as food procurement cycle, production cycle, and sales speed, as well as market factors such as changes in market demand, promotional activities, and holidays. Through mathematical models and algorithms, it calculates and adjusts the safety stock range in real time to ensure that inventory meets market demand without causing excessive stockpiling.

[0078] A supply chain risk assessment matrix is ​​constructed, and risk-weighted early warning rules are generated by incorporating external factors. The matrix assesses supply chain risks from multiple dimensions, including supplier risk, logistics risk, and market risk. Each dimension is further subdivided into specific indicators; for example, supplier risk includes indicators such as supplier reputation, supply capacity, and quality stability. The risk weights of each indicator are determined through historical data analysis and other methods. These are then combined with external factors, such as natural disasters, policy and regulatory changes, and exchange rate fluctuations, to generate risk-weighted early warning rules. When supply chain risks reach a certain threshold, the corresponding early warning mechanism is triggered.

[0079] By comparing and analyzing historical early warning data with actual business results, the parameters of the risk-weighted early warning rules are automatically adjusted. In-depth analysis of historical early warning data and actual business results is conducted regularly to evaluate the effectiveness and accuracy of the early warning rules. If certain early warning rules are found to frequently generate false alarms or misses, machine learning algorithms are used to automatically adjust the parameters of the risk-weighted early warning rules, optimizing the rules and improving their reliability and usability.

[0080] The ERP execution module generates replenishment strategies based on early warning logs. It comprises a log parsing unit, a strategy generation unit, and a strategy execution unit. The log parsing unit categorizes and organizes the early warning logs, and correlates them with supplementary basic transportation data to obtain a complete early warning log. The log parsing unit uses natural language processing technology to perform semantic analysis on the text content of the early warning logs, extracting key information such as the early warning type, the food categories involved, and changes in inventory quantities. This information is then correlated and integrated with basic transportation data, such as transportation routes, transportation times, and transportation costs, to form complete early warning log information.

[0081] The strategy generation unit is used to automatically generate replenishment strategies based on complete early warning logs, through a preset basic replenishment strategy library and intelligent strategy matching algorithm, combined with cost factors.

[0082] The strategy execution unit translates replenishment strategies into business instructions, coordinates with the ERP system to execute these instructions, tracks execution progress in real time, and provides feedback on results. The unit is deeply integrated with the enterprise's ERP system, transforming replenishment strategies into specific business instructions such as purchase orders, production plans, and logistics scheduling. It obtains the execution status of these instructions in real time through interfaces, including whether orders have been placed, goods have been shipped, and inventory has been updated. If any anomalies are detected during execution, such as suppliers failing to deliver on time or logistics delays, the unit promptly reports the information to relevant departments and activates emergency response mechanisms.

[0083] like Figure 3 As shown, the steps by which the strategy generation unit automatically generates a replenishment strategy based on complete early warning logs include:

[0084] A pre-defined strategy mapping table automatically associates different alert categories with basic replenishment strategies. This table pre-sets corresponding basic replenishment strategies based on factors such as alert type, food category, and inventory status. For example, when the alert type is "inventory shortage" and the food category is fresh produce, a rapid replenishment strategy is automatically associated; when the alert type is "inventory backlog," a clearance sale strategy is associated. This method enables rapid matching of replenishment strategies, improving decision-making efficiency.

[0085] A multi-dimensional cost model is established, and cost coefficients are updated in real time based on market dynamics to output cost analysis results. The multi-dimensional cost model considers multiple cost factors such as procurement costs, transportation costs, warehousing costs, and stockout costs, calculating the total cost of different replenishment plans through mathematical models and algorithms. Simultaneously, real-time monitoring of market dynamics, such as raw material price fluctuations and logistics cost adjustments, ensures timely updates to cost coefficients and guarantees the accuracy and timeliness of cost analysis results.

[0086] Based on the linear programming algorithm and cost analysis results, the optimal replenishment plan is solved to obtain the initial replenishment strategy. Feasibility verification is then performed on supplier capacity, logistics network, and warehouse space. The linear programming algorithm uses cost minimization or profit maximization as the objective function, combined with various constraints such as supplier supply capacity limitations, logistics transportation capacity limitations, and warehouse space limitations, to solve for the optimal replenishment plan. The feasibility of the initial replenishment strategy is verified by communicating with suppliers to confirm capacity, verifying transportation capacity with logistics companies, and assessing warehouse space with the warehousing department, ensuring the smooth implementation of the replenishment strategy in actual business operations.

[0087] When solving for the optimal replenishment plan using the linear programming algorithm, let the objective function be cost minimization, expressed as:

[0088]

[0089] in, Let Z be the total cost of the j-th replenishment plan, minZ be the objective function to minimize the cost, Z be the cost of the replenishment plan, m be the quantity of replenishment plans, and j be the index of the replenishment plan. Constraints include supplier supply capacity limitations, logistics transportation capacity limitations, and warehouse space limitations. By solving this linear programming problem, the optimal replenishment plan can be obtained.

[0090] The system automatically switches emergency strategies based on the real-time status of the supply chain and dynamically adjusts the safety stock coefficient in conjunction with seasonal demand fluctuations. When unexpected situations arise in the supply chain, such as supplier shutdowns or logistical disruptions caused by natural disasters, the system automatically identifies and switches to the appropriate emergency strategy, such as finding alternative suppliers or adjusting transportation routes. Simultaneously, it dynamically adjusts the safety stock coefficient based on the characteristics of market demand fluctuations in different seasons, appropriately increasing safety stock levels during peak sales seasons and reducing inventory costs during off-seasons.

[0091] Generate a comprehensive recommendation report that includes replenishment quantity, timing, and supplier selection, including recommended replenishment quantity, optimal replenishment time, suitable supplier list, and related reasons.

[0092] The system monitoring and maintenance module is used to monitor the system's operational status in real time and perform real-time monitoring and analysis of system performance indicators. When a system anomaly is detected, the fault diagnosis process is automatically triggered to quickly locate the fault point and generate maintenance suggestions. This module typically employs various technologies such as monitoring probes, log analysis, and performance indicator collection to collect various operational data of the system in real time, such as server CPU utilization, memory usage, database query response time, and network traffic. Through data analysis algorithms and machine learning models, system performance is evaluated and predicted in real time, and potential performance bottlenecks and fault hazards are identified in a timely manner. Once a system anomaly is detected, the fault diagnosis process is automatically initiated, using methods such as fault tree analysis and root cause analysis to quickly locate the location and cause of the fault. Based on the fault diagnosis results, combined with historical maintenance experience and a knowledge base, detailed maintenance suggestions are generated, including fault repair steps and system optimization measures, to help technicians solve problems in a timely manner and ensure the stable operation of the system. At the same time, the system monitoring and maintenance module also regularly performs health checks and performance optimizations on the system, such as database index optimization, code refactoring, and hardware resource adjustments, to continuously improve the system's performance and reliability.

[0093] In summary, the AI-based supplier food ERP full-chain inventory early warning system provided by this invention accurately predicts future inventory changes through an inventory forecasting module that utilizes time series models, machine learning models, and fusion strategies, enabling enterprises to plan production and procurement in advance. Through multi-dimensional early warning rules and a risk assessment matrix, it promptly identifies potential inventory risks, reducing losses from stockouts or overstocking. The ERP execution module generates optimal replenishment strategies based on cost models and linear programming algorithms, dynamically adjusting these strategies in conjunction with real-time supply chain status, ensuring supply while reducing costs related to procurement, transportation, and warehousing.

[0094] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An AI-based supplier food ERP full-chain inventory early warning system, characterized in that, include: The data acquisition and processing module is used to collect full-chain inventory data and preprocess the full-chain inventory data to generate standardized inventory data. The inventory forecasting module is used to forecast future inventory based on the standardized inventory data and market data, and output the inventory forecasting results. The inventory early warning module is used to preset inventory early warning rules and, in conjunction with the inventory forecast results, issue early warnings for future inventory and output early warning logs. The inventory early warning module includes an early warning rule formulation unit, an early warning triggering unit, and an early warning response unit; The early warning rule formulation unit is used to preset multi-dimensional inventory early warning rules; The early warning triggering unit is used to monitor the full-chain inventory data in real time based on the inventory forecast results, determine whether the early warning rules are met, perform composite early warning judgment, trace the cause of the early warning, and generate early warning analysis data. The early warning response unit is used to organize the early warning analysis data into an early warning log and classify the early warning log into levels of urgency. The steps of the early warning rule formulation unit to preset multi-dimensional inventory early warning rules include: Based on the characteristics of food categories, a dynamic safety stock threshold model is established, and the safety stock range is adjusted in real time according to the basic factors and market factors of food. A supply chain risk assessment matrix is ​​constructed, and risk-weighted early warning rules are generated by combining external factors; the supply chain risk assessment matrix includes multiple dimensions such as supplier risk, logistics risk, and market risk. By comparing and analyzing historical early warning data with actual business results, the parameters of the risk-weighted early warning rules are automatically adjusted. The ERP execution module is used to generate a replenishment strategy based on the warning logs. The ERP execution module includes a log parsing unit and a strategy generation unit. The log parsing unit is used to classify and organize the early warning logs and associate them with supplementary basic transportation data to obtain complete early warning logs. The strategy generation unit is used to automatically generate replenishment strategies based on the complete early warning logs, through a preset basic replenishment strategy library and an intelligent strategy matching algorithm, combined with cost factors. The steps of the strategy generation unit automatically generating a replenishment strategy based on complete early warning logs include: By using a preset strategy mapping table, different early warning categories are automatically associated with basic replenishment strategies; Establish a multi-dimensional cost model and update cost coefficients in real time based on market dynamics to output cost analysis results; Based on the linear programming algorithm and the cost analysis results, the optimal replenishment plan is solved to obtain the initial replenishment strategy; and the feasibility of supplier capacity, logistics network, and warehouse space is verified. The emergency strategy is automatically switched according to the real-time status of the supply chain, and the safety stock coefficient is dynamically adjusted in combination with seasonal demand fluctuations. Generate a comprehensive recommendation report that includes replenishment quantity, timing, and supplier selection.

2. The supplier food ERP full-chain inventory early warning system based on artificial intelligence according to claim 1, characterized in that, The steps for the data acquisition and processing module to output standardized inventory data include: The rule engine identifies and corrects abnormal data in the full-chain inventory data to obtain preprocessed inventory data; A distributed file system is used to store the cleaned preprocessed inventory data; Establish a data quality scoring system to quantitatively assess data quality and output standardized inventory data.

3. The supplier food ERP full-chain inventory early warning system based on artificial intelligence according to claim 2, characterized in that, The steps for the data acquisition and processing module to identify and correct abnormal data include: Based on historical databases, reasonable ranges and constraints are set for various types of data; and historical abnormal data are classified. Based on the type of the abnormal data, a structured language is used to write rules and output abnormal rules; and a rule base is established to classify, store and manage the versions of the abnormal rules. Based on the aforementioned anomaly rules, abnormal data in the full-chain inventory data is identified, and the identification results are output. Based on the identification results and combined with the inbound and outbound data, the abnormal data is corrected.

4. The supplier food ERP full-chain inventory early warning system based on artificial intelligence according to claim 2, characterized in that, The steps for establishing a data quality scoring system in the data acquisition and processing module include: Define the data quality dimensions of the preprocessed inventory data; Assign quantitative metrics to each data quality dimension and set dimension weights; Establish a scoring standard hierarchy and map the scores of each of the data quality dimensions to quality levels; The quality score for the quality level is calculated periodically based on a sliding window.

5. The supplier food ERP full-chain inventory early warning system based on artificial intelligence according to claim 1, characterized in that, The inventory forecasting module includes a feature extraction unit, a model building unit, and a forecasting analysis unit. The feature extraction unit is used to extract and derive features from the standardized inventory data by combining the market data to obtain standard data features. The model building unit is used to establish a time series model and a machine learning model based on a historical warehousing database, and to train the model on the demand fluctuation characteristics of food categories through a model fusion strategy to obtain a fused forecasting model. The forecasting analysis unit is used to update the fused forecasting model in real time based on the full-chain inventory data using a rolling forecasting mechanism, and output the inventory forecasting results.

6. The supplier food ERP full-chain inventory early warning system based on artificial intelligence according to claim 1, characterized in that, The ERP execution module also includes a strategy execution unit; the strategy execution unit is used to convert the replenishment strategy into business instructions, link the ERP system to execute the instructions, track the execution progress in real time, and provide feedback on the execution results.

7. The supplier food ERP full-chain inventory early warning system based on artificial intelligence according to claim 1, characterized in that, It also includes a system monitoring and maintenance module, which is used to monitor the system's operating status in real time, monitor and analyze system performance indicators in real time; when a system anomaly is detected, it automatically triggers a fault diagnosis process, quickly locates the fault point, and generates maintenance suggestions.

Images