Material inventory data management method and system

Abstract

The invention relates to the technical field of material inventory data management, discloses a material inventory data management method and system, and realizes comprehensive tracking of material operation by obtaining a material operation request and associating the material operation request with an operation voucher. The material identification information and the current state information are inquired from the plurality of material management systems and are compared, so that a uniform material identity can be determined according to inherent attributes of the materials, and the problem of inconsistent data sources caused by incompatible material codes of the plurality of systems is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of material inventory data management technology, and in particular to a material inventory data management method and system. Background Technology

[0002] In modern industrial production and warehousing management, meticulous management of material inventory is crucial for ensuring smooth production and controlling costs. However, with the rapid expansion of enterprise scale, such as through mergers and acquisitions or the establishment of new production bases, existing material management systems face unprecedented challenges. Material management systems from different sources may have adopted their own independent material coding rules at the initial design stage, and these rules have inherent incompatibilities. Under the pressure of rapid integration, the interoperability between systems is often difficult to achieve perfectly, resulting in ambiguous or inaccurate material coding mappings, thus causing inconsistencies in the data source at the logical level.

[0003] Meanwhile, newly introduced production processes or products may require the use of novel materials that are difficult to monitor effectively with traditional equipment, such as lightweight, high-value components, leading to discrepancies in physical inventory data. For example, the existing load-bearing racks in the warehouse may have built-in pressure sensors designed primarily for traditional heavy materials. Their measurement accuracy and minimum sensing weight are insufficient to effectively monitor inventory changes for these new components. Therefore, when these lightweight materials enter or leave the warehouse, the physical sensing system cannot record valid weight changes, and significant discrepancies begin to appear between the data and the records in the management system.

[0004] Dual data contamination at both the logical and physical levels makes it difficult for enterprises to obtain a true and unified inventory view. This, in turn, affects the identification of obsolete materials, the accurate prediction of inventory levels, and the timely warning of urgent material shortages. Ultimately, it may lead to a complete loss of control over materials management, severely hindering the enterprise's operational efficiency and market responsiveness. This long-term inconsistency between system records and actual inventory means that the judgment thresholds originally used to identify obsolete materials cannot accurately identify those hidden obsolete materials that have actually been sitting for a long time but have been "masked" by the chaotic system data due to excessive errors in the input data.

[0005] Faced with frequent shortages caused by predictive model failures, companies attempted to manually adjust the alert rule set for emergency material shortages to compensate for the inadequacy of forecasts. However, due to a lack of fundamental understanding of the underlying data chaos, these adjustments based on surface phenomena made the alert system either overly sensitive or insensitive. This human intervention, coupled with a weak data foundation, further weakened the system's ability to proactively identify abnormal states, ultimately leading to a complex technical dilemma of complete loss of control over material inventory management.

[0006] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0007] This application discloses a material inventory data management method, which aims to solve the technical difficulties in modern industrial production and warehousing management, such as incompatibility between different material management systems, difficulty in monitoring new materials, and discrepancies between logical and physical data leading to inconsistent inventory data, difficulty in identifying stagnant materials, inaccurate inventory forecasting, and failure of early warning systems.

[0008] In a first aspect, this application discloses a method for managing material inventory data, comprising the following steps: Obtain a material operation request and associate an operation document with that material operation request; Based on this operation voucher, query the material identification information and current status information from multiple material management systems; By comparing and identifying information, a unified material identity is determined based on the inherent properties of the material; Compare the current status information to identify the quantity of materials; If the material is a specific material, the physical change signal of the material is acquired, and the visual change information of the material is acquired through the imaging device; Compare physical change signals, visual change information, and material handling requests to identify discrepancies between physical and logical data; If discrepancies exist, the data reconciliation process is initiated, and a unified and corrected view of the material data is generated. Send a pre-commit instruction to multiple material management systems. This pre-commit instruction is used to update internal data based on a unified and revised material data view. Receive confirmation signals from multiple material management systems, and after confirmation, instruct the multiple material management systems to record material operations; Based on the recorded material handling data, analyze the duration of material in stock to identify materials that have not been used for a long time; Based on officially recorded material handling data, material demand trend analysis is performed to predict future material consumption. Based on the analysis results of material in-stock duration and material demand trend analysis, material management early warning information is generated.

[0009] This technical solution enables comprehensive and refined management of material inventory data, effectively solving problems such as inconsistencies between multiple systems and differences between physical and logical data. It can also proactively identify stagnant materials and predict material consumption, thereby improving the accuracy and efficiency of inventory management.

[0010] Furthermore, the steps described above, such as initiating a data reconciliation process to generate a unified and corrected material data view if discrepancies exist, include: Collect contextual information about the storage area, including information on the intensity of local airflow disturbances and the type of material packaging. Based on the contextual information, extract the numerical features of the contextual information, which include the airflow disturbance intensity value and the packaging identification matching degree value; Based on the airflow disturbance intensity value and the packaging identification matching degree value, the confidence score of the physical perception data source is calculated. The physical perception data source includes machine vision system and microgravity fluctuation capture device. Based on the confidence score, the quantity of materials reported by the machine vision system and the quantity of materials inferred by the microgravity fluctuation capture device are weighted and fused to obtain the preliminary physical quantity. Compare the preliminary physical quantities with the expected operational quantities, and assess the deviation between the preliminary physical quantities and the expected operational quantities; When the deviation exceeds the preset threshold, the confidence level of the overall physical verification result is reduced, an alarm is triggered, and a verification work order is generated. Receive the actual quantity and the reason for the problem after manual verification. The reason for the problem includes the verifier's description of the problem and the confidence assessment of the verification results. Adjust the weights and parameters in the confidence assessment model based on the actual number and the cause of the problem.

[0011] This technical solution enables intelligent reconciliation of differences between physical and logical data. Through multi-source data fusion and confidence assessment, it effectively improves data accuracy and triggers manual verification in case of anomalies, forming a self-correcting closed loop for data.

[0012] Furthermore, the steps of analyzing material inventory duration based on recorded material handling data to identify materials that have not been used for a long time include: Obtain information on the material type, historical inbound and outbound records, and current production project information of the material to be analyzed; Based on the material type information, query the baseline in-stock duration threshold for the material; Based on the seasonal fluctuation patterns in historical inbound and outbound records, the benchmark inbound duration threshold is periodically adjusted. Based on the project phase information in the production project information, the in-stock duration threshold is adjusted in stages; The adjusted in-stock duration threshold is compared with the actual in-stock duration of the material to be analyzed; If the actual inventory duration exceeds the adjusted inventory duration threshold, the material to be analyzed will be marked as a long-term unused material.

[0013] This technical solution enables accurate identification of materials that have not been used for a long time. By dynamically adjusting the in-stock duration threshold, it avoids misjudgments caused by traditional fixed thresholds and improves the precision of stagnant material management.

[0014] Building upon the above, this application further proposes that the step of conducting material demand trend analysis based on formally recorded material handling data to predict future material consumption includes: Obtain the type information of the material to be analyzed; Determine if the historical operational data for the material to be analyzed is sufficient; When historical operation data is insufficient, query historical consumption data of similar or alternative materials. Based on the bill of materials information in the production plan, deduce the basic consumption amount; Adjust the basic consumption based on the new production line's capacity ramp-up plan or the product life cycle stage; When there is sufficient historical operational data but the pattern is inconsistent with the future, external events are identified; Based on external events, historical operation data is weighted or truncated. Adjust the base consumption based on current orders, forecasted orders, and safety stock strategy; Generate predictions of future material consumption.

[0015] This technical solution enables accurate prediction of future material consumption. Through multi-dimensional data fusion and dynamic adjustment mechanisms, it effectively addresses situations where historical data is insufficient or patterns are inconsistent, thereby improving the robustness of predictions.

[0016] In some preferred embodiments, the steps of reducing the confidence level of the overall physical verification result and triggering an alarm and generating a verification work order when the deviation exceeds a preset threshold include: Obtain the value level information of the material to be operated and the urgency level information of the current operation; Based on the value level information, the preset threshold is adjusted to obtain the adjusted deviation threshold. Based on the urgency level information, adjust the alarm level to obtain the adjusted alarm level; Based on the value level information of the materials and the urgency level information of the operation, adjust the reduction in the confidence level of the overall physical verification results, and adjust the priority and response time of the verification work order; When the deviation between the initial physical quantity and the expected operational quantity exceeds the adjusted deviation threshold, the confidence level of the overall physical verification result is reduced, and an alarm corresponding to the adjusted alarm level is triggered, generating a verification work order with the adjusted priority and response time limit.

[0017] This technical solution enables differentiated responses to abnormal situations, dynamically adjusting thresholds, alarm levels, and work order priorities based on material value and operational urgency, ensuring that high-value or emergency situations are handled more promptly and effectively.

[0018] Based on the above, this application further proposes that the step of adjusting the weights and parameters in the confidence assessment model according to the actual quantity and the cause of the problem includes: Compare the actual quantity with the expected number of operations recorded by the system and historical operation data to identify the deviation of the actual quantity; The process involves analyzing the descriptions of the problems provided by the inspectors, extracting key information from these descriptions, and matching them with pre-defined problem cause classifications to obtain a preliminary problem classification. Based on the confidence assessment of the preliminary problem classification and verification results, the preliminary problem classification is revised to obtain the revised problem classification; Based on the deviation of the actual quantity, the corrected problem classification is weighted and weighted to obtain a weighted problem classification. Based on the weighted problem classification, adjust the weights and parameters of the corresponding sensing devices in the confidence assessment model.

[0019] This technical solution enables adaptive optimization of the confidence assessment model. By manually verifying the results, the model can be adjusted based on feedback, thereby improving its adaptability and accuracy to real-world situations.

[0020] In some preferred embodiments, the step of assigning weights to the corrected problem classification based on the deviation magnitude of the actual quantity to obtain a weighted problem classification includes: Obtain information on the value level, operational urgency level, and historical consumption fluctuations of the materials to be analyzed; Based on the value grade information of the materials, determine the deviation sensitivity to high-value materials; Based on the urgency level information of the operation, determine the priority of the deviation response to the emergency operation; Determine the inherent fluctuation range of materials based on historical consumption fluctuation information; The deviation range of the actual quantity is comprehensively judged in conjunction with the deviation sensitivity of high-value materials, the deviation response priority of emergency operations, and the inherent fluctuation range of the materials. When the deviation of the actual quantity exceeds the inherent fluctuation range, and the material is a high-value material or the operation is an emergency operation, the weight allocation coefficient of the corrected problem classification is increased. When the deviation of the actual quantity is within the inherent fluctuation range, and the material is a low-value material or the operation is a non-urgent operation, reduce the weight allocation coefficient of the corrected problem classification. The corrected problem classification is weighted according to the weighting coefficients after the increase or decrease.

[0021] This technical solution enables intelligent allocation of problem classification weights, and allows for more precise judgment of deviation magnitude based on material characteristics and operational urgency, making model adjustments more targeted.

[0022] Based on the above, this application further proposes that the step of adjusting the weights and parameters of the corresponding sensing devices in the confidence assessment model according to the weighted problem classification includes: Based on the weighted problem classification, determine the sensing devices that need adjustment and the direction of adjustment; Based on a preset adjustment strategy, the weights and parameters of the sensing devices are incrementally adjusted to obtain the adjusted model. Acquire and adjust relevant historical scenario data; The adjusted model is used to retrospectively evaluate historical scene data to obtain retrospective evaluation results; Acquire forward-looking scenario data related to adjustments; The adjusted model is used to conduct a forward-looking assessment of the prospective scenario data, and the forward-looking assessment results are obtained. Compare the retrospective evaluation results, the prospective evaluation results, and the model performance before adjustment; If the comparison results show that the performance of the model before adjustment has decreased or the decrease exceeds the preset threshold, the model will be restored to its state before adjustment, and a model performance anomaly report will be generated. If the comparison results show that the performance of the model before the adjustment has improved or remained stable, then the adjustment will take effect; Regularly cross-validate the confidence assessment models for all sensing devices.

[0023] This technical solution enables robust adjustment and validation of the confidence assessment model. Through retrospective and prospective assessments, it ensures the effectiveness and stability of model adjustments and avoids the negative impact of blind adjustments.

[0024] In some preferred embodiments, the step of periodically cross-validating the confidence assessment model for all sensing devices as described above includes: During the preset maintenance window or a period of low system load, select one or a group of devices from all deployed sensing devices as the target of this verification. Generate a set of test cases for the selected sensing device model, which covers typical operating scenarios and abnormal situations that the selected sensing device may encounter in different contexts; The confidence assessment model of the selected sensing device is temporarily isolated, and the weights and parameters are adjusted in a pre-set, small-scale simulation. The adjusted sensing device model is run in a simulated environment using test cases, and its performance on the test cases is evaluated to obtain the simulated performance of the adjusted sensing device model. The performance of the simulated and adjusted sensing device model is compared with its performance before adjustment and the performance of other unadjusted sensing device models on the same test cases.

[0025] This technical solution enables the systematic verification of the confidence assessment model for sensing devices. Through simulation adjustments and performance comparisons, the accuracy and reliability of the model under different scenarios can be ensured.

[0026] Secondly, this application also discloses a material inventory data management system, which includes: The detection end is used to acquire material operation requests and associate an operation credential with the material operation request; based on the operation credential, it queries multiple material management systems for the material's identification information and current status information; The comparison end is used to compare and identify information, determine a unified material identity based on the inherent properties of the material; compare the current status information to identify the quantity of the material; if the material is a specific material, it acquires the physical change signal of the material and acquires the visual change information of the material through the imaging device; it compares the physical change signal, the visual change information and the material operation request to identify the difference between physical and logical data. The processing unit initiates a data reconciliation process and generates a unified and corrected material data view if discrepancies exist. It sends pre-submission instructions to multiple material management systems to update internal data based on the unified and corrected material data view. It receives confirmation signals from multiple material management systems and, upon confirmation, instructs them to record material operations. Based on the recorded material operation data, it performs material inventory duration analysis to identify materials that have not been used for a long time. Based on the officially recorded material operation data, it performs material demand trend analysis to predict future material consumption. Based on the results of the material inventory duration analysis and the material demand trend analysis, it generates material management early warning information.

[0027] This technical solution provides a system platform for implementing material inventory data management. Through modular design, it automates data acquisition, comparison, processing, and early warning, effectively supporting the needs of refined inventory management. Beneficial effects

[0028] This application discloses a material inventory data management method and system that achieves comprehensive tracking of material operations by acquiring material operation requests and associating them with operation vouchers. By querying and comparing material identification information and current status information from multiple material management systems, a unified material identity can be determined based on the inherent attributes of the material, effectively solving the problem of inconsistent data sources caused by incompatible material codes across multiple systems. For specific materials, by acquiring physical change signals and visual change information and comparing them with operation requests, differences between physical and logical data can be identified, overcoming the technical challenge of traditional equipment's inability to effectively monitor new materials, leading to deviations in physical inventory data. When discrepancies exist, a data reconciliation process is initiated to generate a unified and corrected material data view, and internal data is updated through pre-submission instructions and confirmation mechanisms to ensure the authenticity and consistency of inventory data. Furthermore, based on recorded material operation data, in-stock duration analysis and demand trend analysis can be performed to identify long-term unused materials and predict future material consumption, ultimately generating material management early warning information, effectively solving the problems of inaccurate identification of stagnant materials, ineffective inventory forecasting, and untimely early warning of emergency material shortages. In summary, the method of this application can provide a unified and accurate view of material inventory, significantly improving the operational efficiency and market responsiveness of enterprises, and overcoming the complex technical dilemma of complete loss of control over material management in the prior art. Attached Figure Description

[0029] Figure 1 This is a flowchart of a material inventory data management method provided by an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a material inventory data management system provided in an embodiment of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] Reference Figure 1 , Figure 1 This is a flowchart of a material inventory data management method provided by an embodiment of the present invention, including the following steps: S11, Obtain a material operation request and associate an operation voucher with the material operation request; Based on the operation voucher, query the material identification information and current status information from multiple material management systems; S12, compare the identification information to determine a unified material identity based on the inherent attributes of the material; compare the current status information to identify the quantity of the material; S13, if the material is a specific material, then obtain the physical change signal of the material and obtain the visual change information of the material through the imaging device; S14, compare the physical change signal, the visual change information and the material operation request to identify the difference between the physical and logical data; if the difference exists, start the data reconciliation process and generate a unified and corrected material data view; S15, send a pre-submission instruction to the plurality of material management systems, the pre-submission instruction being used to update internal data according to the unified and corrected material data view; S16, receive confirmation signals from the plurality of material management systems, and after confirmation, instruct the plurality of material management systems to record material operations; S17. Based on the recorded material operation data, perform material in-stock duration analysis to identify materials that have not been used for a long time; based on the officially recorded material operation data, perform material demand trend analysis to predict future material consumption; based on the material in-stock duration analysis results and the material demand trend analysis results, generate material management early warning information.

[0032] The material inventory data management method proposed in this application aims to solve problems such as inconsistent material data and low inventory management efficiency in a multi-system environment. Here, "material operation request" refers to an instruction or event for operations such as material receiving, issuing, transferring, and inventory counting. "Operation voucher" is a unique identifier generated for each material operation request, used to track and associate data throughout the entire operation process. The "material management system" can be an Enterprise Resource Planning (ERP) system, a Warehouse Management System (WMS), a Manufacturing Execution System (MES), etc., each managing a portion of the material data. "Identification information" includes the material's code, name, specifications, etc. "Current status information" refers to the material's inventory quantity, location, batch, etc., in each system. "Specific materials" refers to materials with high inventory accuracy requirements, prone to physical changes, or of high value, such as precision electronic components and chemical reagents. "Physical change signals" can be acquired through sensors (such as weight sensors and volume sensors) to reflect changes in the material's actual physical state. "Visual change information" is acquired through imaging equipment (such as industrial cameras) to identify visual characteristics such as the material's shape, quantity, and packaging. The "Data Reconciliation Process" is a series of processing steps initiated when discrepancies exist between physical and logical data. Its purpose is to eliminate these discrepancies and generate an accurate, unified data view. The "Pre-Submission Instruction" is an instruction sent to each material management system to prepare for internal data updates based on the reconciled data before final confirmation. The "Confirmation Signal" is a signal returned by each material management system after completing its internal data update preparation, indicating that final recording can proceed. "Material In-Stock Duration Analysis" statistically analyzes the time from material receipt to issuance or inventory count to identify materials that have not been used for a long time. "Material Demand Trend Analysis" predicts future material consumption based on historical consumption data and future plans. "Material Management Early Warning Information" is generated based on the analysis results to alert users to potential inventory risks or provide management recommendations.

[0033] In material inventory data management, the first step is to obtain material operation requests and associate each request with an operation voucher. Material operation requests can be obtained in various ways. For example, they can be manually entered into the management system as outbound or inbound orders, at which point the system will automatically generate an operation voucher, such as a unique transaction ID. Alternatively, after an automated warehousing device (such as an AGV) completes a material handling operation, its control system can automatically generate a material operation request, and the central management platform will assign it an operation voucher.

[0034] Based on the operation voucher, the system queries multiple material management systems for material identification information and current status information. For example, when a material operation request is submitted, the system uses the operation voucher as the basis for the query and simultaneously sends query requests to the ERP system, WMS system, and MES system. The ERP system may return the material's general code, name, and current book inventory; the WMS system may return the material's location, batch, and actual inventory in a specific warehouse; and the MES system may provide the material's status information on the production line.

[0035] Next, the identification information needs to be compared to determine a unified material identity based on the inherent properties of the material. For example, the system can receive material codes from different material management systems, such as "A123" in ERP and "W-A123-001" in WMS. Through preset mapping rules or the material master data management system, these different codes are compared, and based on the inherent properties of the material (such as chemical composition, physical dimensions, supplier model, etc.), it is determined that they all point to the same material, and a unified internal material identity identifier is assigned to it.

[0036] Then, the system compares the current status information to identify the material quantity. For example, the system receives a report from the ERP system stating a material quantity of 100 units, while the WMS report states a material quantity of 98 units. By comparing these quantities, the system can initially identify the book quantity of the materials and prepare for subsequent discrepancy identification.

[0037] If the material is specific, the system acquires signals indicating physical changes in the material and uses imaging equipment to obtain visual information about those changes. For example, for high-value, precision electronic components, when they are removed from or placed on a shelf, the shelf's built-in microgravity fluctuation capture device senses minute weight changes and generates physical change signals. Simultaneously, imaging equipment installed above the shelf captures images or videos of the material area to obtain visual information about changes in the material, such as how it is stacked, the integrity of its packaging, or a visual estimate of its quantity.

[0038] Subsequently, the system compares physical change signals, visual change information, and material handling requests to identify discrepancies between physical and logical data. For example, a material handling request might indicate that 10 items should be shipped, but the microgravity fluctuation capture device only detects a weight change equivalent to 8 items, and the imaging device identifies a reduction of 8 units in the material stack. In this case, the system would identify a difference of 2 items between the physical data (8 items) and the logical data (10 items).

[0039] If discrepancies exist, a data reconciliation process is initiated, generating a unified and corrected material data view. For example, when the system detects discrepancies between physical and logical data, a data reconciliation module is automatically triggered. This module may consider various factors, such as the confidence level of sensor data, historical discrepancy patterns, and manual verification results, ultimately generating a corrected material quantity. For instance, the quantity of the discrepancy might be corrected to 9 units, and a unified and corrected material data view is generated accordingly.

[0040] Next, pre-commit instructions are sent to multiple material management systems. These instructions are used to update internal data based on a unified and revised material data view. For example, the system sends the revised material data view to the ERP and WMS systems, instructing them to update their respective internal material inventory data according to this view. However, at this point, the update has not yet taken effect and remains in a pre-commit state.

[0041] Then, the central management system receives confirmation signals from multiple material management systems and, upon confirmation, instructs these systems to record the material operation. For example, after both the ERP and WMS systems have successfully completed the pre-update of their internal data and returned confirmation signals, the central management system sends a final recording instruction, causing each system to officially record the material operation and its corrected data into its respective database.

[0042] Based on recorded material handling data, the system analyzes the duration of materials in stock to identify materials that have not been used for a long time. For example, the system periodically analyzes the timestamps of all materials' entry and most recent exit or inventory count. For materials whose actual stock time exceeds a preset threshold (e.g., 6 months), the system marks them as long-term unused materials and generates a corresponding report.

[0043] Based on officially recorded material handling data, the system performs material demand trend analysis to predict future material consumption. For example, the system uses historical inbound and outbound records, production plans, sales orders, and other data, employing methods such as time series analysis and regression analysis, to predict material consumption trends for the next quarter or half a year, providing a basis for procurement and inventory planning.

[0044] Finally, based on the analysis results of material inventory duration and material demand trend analysis, material management early warning information is generated. For example, if the material inventory duration analysis finds a large amount of materials that have not been used for a long time, and the material demand trend analysis shows that the future demand for these materials is low, the system may generate an early warning message suggesting "scrapping or reselling". If a shortage of a certain key material is predicted in the future, the system will generate an early warning message suggesting "procure in advance".

[0045] The material inventory data management method proposed in this application, by introducing an operation voucher mechanism, achieves unified tracking and management of material operation requests, effectively solving the problem of inconsistent material data sources in a multi-system environment. By comparing identification information and current status information from different material management systems, and combining this with the inherent attributes of the materials, this application can determine a unified material identity and accurate material quantity, laying the foundation for subsequent data processing.

[0046] Of particular note is that this application introduces a mechanism for acquiring and comparing physical change signals and visual change information for specific materials. Traditional methods often rely solely on system accounting data or data from a single sensor, making it difficult to accurately reflect the actual inventory of high-value, perishable, or difficult-to-measure materials. This application, by integrating visual information acquired from physical sensing devices such as microgravity fluctuation capture devices and imaging equipment, can more comprehensively and accurately capture the actual physical state changes of materials, thereby identifying the differences between physical and logical data.

[0047] When discrepancies are detected between physical and logical data, this application can initiate a data reconciliation process and generate a unified and corrected view of material data. This mechanism effectively avoids the inventory management chaos caused by data inconsistency in traditional methods, ensuring the accuracy and reliability of inventory data. By sending pre-submission instructions to multiple material management systems and receiving confirmation signals, this application achieves synchronized updates of data within each system, ensuring the consistency of material data across the entire enterprise.

[0048] Furthermore, this application innovatively combines material inventory duration analysis and material demand trend analysis. Traditional methods are often limited by insufficient data accuracy and a single analytical dimension in identifying stagnant materials and predicting material consumption. Based on corrected and accurate material handling data, this application can more accurately identify long-term unused materials and combine historical consumption data and production plans to predict material demand. Ultimately, based on these two analytical results, material management early warning information is generated, providing enterprises with forward-looking decision support and effectively avoiding problems such as stagnant material backlog and emergency material shortages.

[0049] Compared to existing technologies, this application has the following advantages: First, by comparing and querying data from multiple systems using operational vouchers, it fundamentally solves the problems of incompatible material codes and inconsistent data across multiple systems. Second, by introducing a dual physical and visual perception mechanism for specific materials, it significantly improves the inventory accuracy of high-value or special materials. Third, the data reconciliation process ensures the consistency of physical and logical data, providing a reliable data foundation for subsequent inventory analysis and forecasting. Finally, by combining inventory duration analysis and demand trend analysis, it can generate more intelligent and instructive material management early warning information, thereby comprehensively improving the precision and efficiency of material inventory management.

[0050] Specifically, if the aforementioned discrepancies exist, a data reconciliation process is initiated to generate a unified and corrected material data view, which may further include the following steps: If the aforementioned discrepancies exist, the data reconciliation process is initiated, generating a unified and corrected view of the material data, including: Collect contextual information about the storage area, including local airflow disturbance intensity information and material packaging type information; Based on the context information, extract the numerical features of the context information, including the airflow disturbance intensity value and the packaging identification matching degree value; Based on the airflow disturbance intensity value and the packaging identification matching degree value, the confidence score of the physical perception data source is calculated. The physical perception data source includes machine vision system and microgravity fluctuation capture device. Based on the confidence score, the quantity of materials reported by the machine vision system and the quantity of materials inferred by the microgravity fluctuation capture device are weighted and fused to obtain a preliminary physical quantity. Compare the preliminary physical quantities with the expected operational quantities, and assess the deviation between the preliminary physical quantities and the expected operational quantities; When the deviation exceeds a preset threshold, the confidence level of the overall physical verification result is reduced, an alarm is triggered, and a verification work order is generated. Receive the actual quantity and the reason for the problem after manual verification. The reason for the problem includes the verification personnel's description of the problem and the confidence assessment of the verification results. Based on the actual number and the cause of the problem, adjust the weights and parameters in the confidence assessment model.

[0051] Specifically, when initiating the data reconciliation process, the first step is to collect contextual information about the storage area. This contextual information aims to provide background data about the environment in which the material is located, and may include information on the intensity of local airflow disturbances and the type of material packaging. Information on the intensity of local airflow disturbances reflects the severity of airflow within the storage area, such as airflow changes caused by fans, air conditioning, or human activity, which may affect the stability of lightweight materials and the accuracy of sensing devices. Information on the type of material packaging refers to the form of packaging, such as whether it is bulk, boxed, bagged, or palletized; different packaging types may affect the acquisition and interpretation of physical sensing data.

[0052] Furthermore, based on the collected contextual information, numerical features of the contextual information can be extracted. These numerical features are a quantitative representation of the contextual information, and may include airflow disturbance intensity values ​​and packaging identification matching scores. The airflow disturbance intensity value can be a specific value obtained through sensor measurement or model estimation, used to quantify the degree of airflow disturbance. The packaging identification matching score can be a confidence score or similarity score obtained by identifying the material packaging through image recognition or RFID technologies and then matching it with a preset packaging type.

[0053] Based on this, a confidence score for the physical sensing data source can be calculated according to the airflow disturbance intensity value and the packaging identification matching degree value. The physical sensing data source may include a machine vision system and a microgravity fluctuation detection device. The machine vision system can identify the type, quantity, and location of materials through image processing technology. The microgravity fluctuation detection device can infer the quantity or state of materials by detecting minute changes in material weight. The confidence score reflects the reliability of the data provided by different sensing devices in the current context. For example, in an environment with strong airflow disturbance, the confidence score of a machine vision system for counting lightweight materials may decrease; for materials with specific packaging types, the identification accuracy of certain sensing devices may be higher.

[0054] Subsequently, based on the confidence scores, a weighted fusion is performed on the material quantity reported by the machine vision system and the material quantity inferred by the microgravity fluctuation capture device to obtain a preliminary physical quantity. Weighted fusion refers to assigning different weights to the confidence scores of each sensing data source, and then combining these weighted quantities to obtain a more accurate and reliable estimate of the material quantity. For example, the quantity reported by sensing devices with higher confidence scores will be given greater weight.

[0055] Next, the preliminary physical quantity is compared with the expected operational quantity to assess the deviation between the two. The expected operational quantity refers to the logical quantity of materials that should occur based on the material handling request. This comparison reveals any inconsistencies between the physical inventory and the logical records.

[0056] When the deviation exceeds a preset threshold, the confidence level of the overall physical verification results can be reduced, triggering an alarm and generating a verification work order. The preset threshold can be a configurable value used to define the acceptable range of deviation. Once the deviation exceeds this range, it indicates a significant discrepancy between the physical and logical data, requiring further manual intervention. Reducing the confidence level of the overall physical verification results means the system is skeptical of the reliability of the currently automatically acquired physical quantities. Triggering an alarm notifies relevant personnel of this anomaly, while generating a verification work order initiates the manual verification process, assigning personnel to conduct an on-site inventory or investigation.

[0057] After manual verification is completed, the system can receive the verified actual quantity and the reason for the problem. The reason for the problem can include the verifier's description of the problem and an assessment of the confidence level of the verification results. The actual quantity is the accurate quantity of materials confirmed manually on-site. The reason for the problem can detail the specific circumstances that led to the deviation, such as misplaced materials, damaged packaging, or system entry errors. Simultaneously, the verifier can assess the reliability of the verification results.

[0058] Finally, based on the actual number of errors and the reasons for the problems, the weights and parameters in the confidence assessment model can be adjusted. This adjustment process aims to optimize the confidence assessment model of the sensing device by learning from the results of human verification, enabling it to more accurately determine the reliability of various sensing data sources in different scenarios in the future. For example, if a sensing device frequently errs in a specific scenario and is corrected through human verification, the weight of that sensing device in that scenario may be reduced, or its parameters may be adjusted to reduce similar errors.

[0059] This application's solution improves the accuracy of initial physical quantities by dynamically assessing the confidence level of physical perception data sources through the introduction of contextual information and performing weighted fusion based on this assessment. When there is a significant deviation between the physical quantity and the expected operational quantity, the system can promptly identify and trigger manual verification, avoiding potential inventory data errors. By receiving the verified actual quantity and the cause of the problem, the system can adaptively adjust the confidence assessment model, forming a closed-loop optimization mechanism that makes the acquisition and reconciliation process of physical inventory data more intelligent and reliable. This mechanism effectively solves the problem of inconsistency between physical and logical inventory data in traditional material management and provides the ability to automatically detect, warn, and correct discrepancies.

[0060] Through the above technical solution, this application can achieve refined reconciliation of physical and logical discrepancies in material inventory data. Specifically, by collecting and utilizing contextual information from the storage area, the system can more accurately assess the data reliability of different physical sensing devices (such as machine vision systems and microgravity fluctuation capture devices) under specific environments, thereby performing intelligent weighted fusion to obtain preliminary physical quantities that are closer to reality. When a significant deviation is detected between the physical quantity and the expected operational quantity, the system can promptly trigger an alarm and generate a verification work order, effectively avoiding production interruptions or resource waste caused by data inconsistencies. Furthermore, by continuously optimizing the confidence assessment model through manual verification results, the system can continuously learn and adapt to complex inventory environments, significantly improving the accuracy, reliability, and management efficiency of material inventory data, while reducing the frequency and cost of manual intervention.

[0061] In some of the embodiments described above in this application, a method for analyzing material holding time based on recorded material operation data is proposed to identify materials that have not been used for a long time. However, in actual inventory management, the "long-term unused" status of materials is not static, and its judgment criteria are often affected by factors such as the characteristics of the materials themselves, changes in the market environment, and adjustments to production plans. If only a single or fixed holding time threshold is used for judgment, it may lead to misjudgment of the material status. For example, seasonal materials with normal inventory during the off-season may be misjudged as long-term unused, or high-value materials that are actually idle may fail to be identified in a timely manner, thereby affecting the accuracy and efficiency of inventory management. To address this, this application further proposes a scheme to optimize the material holding time analysis by dynamically adjusting the holding time threshold to more accurately identify long-term unused materials.

[0062] The above-mentioned material handling data is used to analyze the duration of materials in stock in order to identify materials that have not been used for a long time, including: Obtain information on the material type, historical inbound and outbound records, and current production project information of the material to be analyzed; Based on the material type information, query the baseline in-stock duration threshold for the material; Based on the seasonal fluctuation patterns in the historical inbound and outbound records, the benchmark inbound duration threshold is periodically adjusted. The in-stock duration threshold is adjusted in stages based on the project stage information in the production project information. The adjusted in-stock duration threshold is compared with the actual in-stock duration of the material to be analyzed; If the actual in-stock duration exceeds the adjusted in-stock duration threshold, the material to be analyzed will be marked as a long-term unused material.

[0063] Specifically, when analyzing the duration of materials in stock, it is first necessary to obtain information on the material type, historical inbound and outbound records, and the current production project information of the material to be analyzed. Material type information can include the material's classification, attributes, and value level, used to distinguish the characteristics of different materials; historical inbound and outbound records provide information on the material's movement over a past period, helping to identify its usage patterns and seasonal characteristics; and production project information is linked to the material's purpose and project progress, such as whether the project is in the R&D, pilot production, or mass production stage.

[0064] Furthermore, based on the obtained material type information, the baseline in-stock duration threshold for that material can be queried. This baseline threshold is a preset upper limit for the in-stock time that is acceptable under normal circumstances for a specific material type. For example, the baseline in-stock duration threshold may be shorter for perishable materials, while it may be longer for certain strategic materials that are stored for a long time.

[0065] Based on this, and taking into account the seasonal fluctuation patterns reflected in historical inventory records, the benchmark inventory duration threshold is periodically adjusted. For example, for materials with significant seasonal demand, the inventory duration threshold can be appropriately extended before the peak demand season to meet stockpiling needs; while during the off-season, the threshold should be shortened to avoid inventory buildup. This adjustment allows the threshold to better adapt to the natural demand cycle of materials.

[0066] Meanwhile, the inventory duration threshold is adjusted periodically based on the project stage information in the production project information. For example, when the project to which the material belongs is in the R&D or initial trial production stage, the inventory duration threshold can be appropriately relaxed due to the high uncertainty of demand; while when the project enters the mass production stage, the demand is stable and the production pace accelerates, the inventory duration threshold should be tightened to improve inventory turnover efficiency.

[0067] Subsequently, the in-stock duration threshold, after the aforementioned periodic and phased adjustments, is compared with the actual in-stock duration of the material to be analyzed. Actual in-stock duration refers to the time elapsed from the material's entry into the warehouse to the current moment.

[0068] If the actual inventory duration exceeds the adjusted inventory duration threshold, the material to be analyzed will be marked as long-term unused material. This marking helps with subsequent inventory optimization decisions, such as material allocation, scrapping, or demand forecast revision.

[0069] This application's solution addresses the potential misjudgment of long-term unused materials by incorporating material type information, seasonal fluctuation patterns from historical inventory records, and project stage information from production project data. Specifically, material type information ensures that different materials have distinct long-term storage duration standards, avoiding a "one-size-fits-all" approach. The introduction of seasonal fluctuation patterns allows the threshold to adapt to cyclical changes in market demand, preventing misjudgments due to seasonal factors. Furthermore, the consideration of project stage information ensures that the threshold matches the actual progress of the production plan, avoiding inventory management deviations caused by varying project durations. Through these dynamic adjustment mechanisms, the system can more accurately determine the actual status of materials, ensuring that only truly idle or abnormally turnover-prone materials are marked as long-term unused, thereby improving the precision of inventory analysis.

[0070] Through the above technical solution, this application enables intelligent and adaptive adjustment of the material in-stock duration threshold, significantly improving the accuracy of identifying materials that have not been used for a long time. Compared with the traditional method using fixed thresholds, this application can effectively avoid misjudgments caused by differences in material characteristics, seasonal demand, or project stages, reducing unnecessary inventory counting or processing costs. As a result, inventory managers can obtain more reliable material status information, enabling them to formulate inventory optimization strategies more promptly and accurately, such as material allocation, accelerated consumption, or timely scrapping, effectively reducing the risk of inventory backlog and improving capital turnover efficiency and overall inventory management level.

[0071] This application further proposes the following steps for conducting material demand trend analysis based on the aforementioned formally recorded material handling data to predict future material consumption: Obtain the type information of the material to be analyzed; Determine whether the historical operation data of the material to be analyzed is sufficient; If the historical operation data is insufficient, query the historical consumption data of similar or alternative materials. Based on the bill of materials information in the production plan, deduce the basic consumption amount; Adjust the basic consumption based on the new production line's capacity ramp-up plan or the product life cycle stage; When the historical operation data is sufficient but the pattern is inconsistent with the future, an external event is identified; Based on the external events, the historical operation data is weighted or truncated. The base consumption level is adjusted by combining current orders, forecasted orders, and safety stock strategy; Generate predictions of future material consumption.

[0072] Specifically, obtaining the type information of the material to be analyzed refers to the system automatically extracting attributes such as the material's classification, specifications, and uses from the material master data for subsequent targeted analysis. Determining whether the historical operational data of the material to be analyzed is sufficient can be understood as the system evaluating the quantity and quality of existing historical data based on preset rules (e.g., requiring at least the monthly consumption data of the past 12 months). In practical applications, when the historical operational data is insufficient, the system queries the historical consumption data of similar or alternative materials. For example, if historical data for a new type of screw is insufficient, the system can query the historical consumption data of screws of the same type and specifications but from different suppliers, or query the historical data of other fasteners used as substitutes in the same product. The purpose is to fill data gaps and provide a more comprehensive reference for subsequent analysis.

[0073] Furthermore, by combining the bill of materials (BOM) information in the production plan, the basic consumption is derived. This refers to the system calculating the theoretical consumption of a specific material under normal production load based on current and future production orders, product structure (BOM), and production process flow. As a preferred implementation, the basic consumption is adjusted according to the capacity ramp-up plan of the new production line or the product lifecycle stage. For example, for a newly commissioned production line, its capacity will gradually increase, and material consumption will also increase accordingly; for products at the end of their product lifecycle, their material requirements may gradually decrease, thus requiring dynamic adjustment of the basic consumption.

[0074] Furthermore, when historical operational data is sufficient but the pattern is inconsistent with the future, external events are identified, such as changes in market policies, fluctuations in raw material prices, major holidays, and new product launches by competitors, which may affect material demand. Based on these external events, the historical operational data is weighted or truncated. Specifically, for recent external events, relevant historical data can be assigned higher weights; outdated data that is no longer of reference value can be truncated to avoid negatively impacting the forecast results. Therefore, by combining current orders, forecasted orders, and safety stock strategies, the basic consumption level is revised. This means that based on the basic consumption level, confirmed customer orders, future order volumes derived from market forecasting models, and safety stock levels set to address uncertainty, the material demand is finally revised. Finally, generating a forecast of future material consumption means that the system integrates the data after multiple adjustments and revisions and outputs a quantitative, time-dimensional material consumption forecast report.

[0075] This application's solution, through a multi-stage, adaptive analysis process, effectively addresses the problems of insufficient data, environmental changes, and the impact of external events in traditional material demand forecasting. Through the aforementioned technical solution, this application significantly improves the accuracy and reliability of material demand forecasting. Specifically, by introducing alternative material data, it effectively solves the problem of insufficient historical data for new or low-frequency materials, avoiding forecast blind spots caused by data gaps. Simultaneously, by dynamically adjusting basic consumption to adapt to the ramp-up of new production lines and changes in product lifecycles, the forecast results better reflect actual production, reducing forecast errors caused by changes in the production environment. Furthermore, by identifying and responding to external events and intelligently processing historical data, the forecasting model possesses stronger environmental adaptability and anti-interference capabilities, effectively reducing the impact of market fluctuations on material planning. Therefore, this application can help enterprises formulate more accurate procurement plans and inventory strategies, effectively reduce inventory costs, reduce material shortage risks, and improve the overall efficiency and responsiveness of the supply chain.

[0076] When the deviation exceeds a preset threshold, the confidence level of the overall physical verification result is reduced, an alarm is triggered, and a verification work order is generated, including: Obtain the value level information of the material to be operated and the urgency level information of the current operation; Based on the value level information, the preset threshold is adjusted to obtain the adjusted deviation threshold; Based on the urgency level information, the alarm level is adjusted to obtain the adjusted alarm level; Based on the value level information of the material and the urgency level information of the operation, adjust the confidence reduction of the overall physical verification result and adjust the priority and response time of the verification work order; When the deviation between the initial physical quantity and the expected operation quantity exceeds the adjusted deviation threshold, the confidence level of the overall physical verification result is reduced, and an alarm corresponding to the adjusted alarm level is triggered, generating a verification work order with the adjusted priority and response time limit.

[0077] The acquisition of the value level information of the material to be operated and the urgency level information of the current operation refers to the system extracting value attributes (e.g., high value, medium value, low value) and urgency attributes (e.g., urgent, high priority, routine) related to the material from the material master data or business process data when processing material operation requests. This information forms the basis for subsequent dynamic adjustments. For example, the value level of a material can be defined based on its procurement cost, scarcity, or impact on production criticality; the urgency level of the operation can be determined based on factors such as whether it involves urgent production orders, whether it is a critical inventory count, or whether it affects product delivery cycles.

[0078] Furthermore, based on the value level information, the preset threshold is adjusted to obtain the adjusted deviation threshold. This means that the system no longer uses a single fixed deviation threshold, but dynamically adjusts the threshold according to the value level of the material. For example, for high-value materials, even a small quantity deviation can lead to significant economic losses, so the system will lower the preset deviation threshold to make it more sensitive to minor deviations; while for low-value materials, the deviation threshold can be appropriately increased to avoid frequently triggering alarms due to unimportant minor deviations, thereby reducing unnecessary intervention.

[0079] Simultaneously, based on the urgency level information, the alarm level is adjusted to obtain the adjusted alarm level. This step ensures that the alarm response level matches the actual urgency of the operation. For example, when a deviation occurs during an urgent operation, the system will trigger a higher-level alarm (such as a Level 1 alarm, immediately notifying key personnel) to ensure the problem receives the fastest possible response; while for non-urgent operations, the alarm level can be appropriately lowered to avoid excessive disruption to non-critical business processes.

[0080] Furthermore, based on the value level information of the materials and the urgency level information of the operations, the confidence level reduction of the overall physical verification results is adjusted, and the priority and response time of the verification work orders are also adjusted. When a deviation is detected, the system's confidence in the physical verification results decreases. This adjustment refers to determining the degree of confidence reduction based on the combination of material value and operational urgency. For example, deviations in high-value materials or urgent operations will lead to a significant decrease in confidence, prompting the system to be more cautious. Simultaneously, to ensure that problems can be resolved quickly, verification work orders for high-value materials or urgent operations will be assigned higher priority and shorter response times.

[0081] Finally, when the deviation between the initial physical quantity and the expected operational quantity exceeds the adjusted deviation threshold, the system will execute a series of dynamically adjusted response measures. Specifically, the system will lower the confidence level of the overall physical verification results and trigger an alarm corresponding to the adjusted alarm level, while generating a verification work order with the adjusted priority and response time limit. This ensures that the system can provide differentiated and intelligent processing solutions based on the actual importance of materials and operations.

[0082] This application's solution addresses the limitations of existing solutions' fixed thresholds and response mechanisms by dynamically adjusting key parameters in the data reconciliation process by introducing material value level information and operational urgency level information. Through this technical solution, the application can dynamically adjust various parameters for deviation handling based on the actual value of the material and the urgency of the operation, thereby achieving intelligent optimization of the material inventory data management process. Compared to the fixed thresholds and response mechanisms in basic solutions, this solution avoids excessive alerts and resource investment for low-value materials, while ensuring sufficient attention and rapid response to deviations in high-value materials or urgent operations. This not only improves the efficiency and accuracy of the data reconciliation process and reduces unnecessary intervention, but also significantly reduces the economic losses and production delay risks that may be caused by material deviations. Through differentiated response strategies, this solution makes the material inventory data management system more flexible and efficient, better adaptable to complex and ever-changing business needs, and improves the overall refinement level of inventory management and risk control capabilities.

[0083] The steps for adjusting the weights and parameters in the confidence assessment model based on the actual quantity and the cause of the problem can be further refined into the following operations: By comparing the actual quantity with the expected number of operations recorded by the system and historical operation data, the deviation of the actual quantity is identified. The problem descriptions provided by the inspectors are analyzed, key information in the descriptions is extracted, and matched with preset problem cause classifications to obtain a preliminary problem classification. Based on the preliminary problem classification and the confidence assessment of the verification results, the preliminary problem classification is revised to obtain the revised problem classification; Based on the deviation range of the actual quantity, the corrected problem classification is weighted to obtain a weighted problem classification; Based on the weighted problem classification, the weights and parameters of the corresponding sensing devices in the confidence assessment model are adjusted.

[0084] Specifically, after receiving the manually verified actual quantity, the system compares it with the expected operational quantity used when the data reconciliation process was initiated. It also references historical operational data for the material, such as past inbound and outbound records and inventory counts. Through this multi-dimensional comparison, the degree of difference between the actual quantity and the system's expectation can be accurately calculated, i.e., the deviation magnitude. This deviation magnitude can be an absolute value or a relative percentage, and its purpose is to quantify the inconsistency between physical inventory and logical inventory.

[0085] During manual verification, inspectors typically provide a written description of the cause of the problem. This application uses Natural Language Processing (NLP) technology or keyword matching algorithms to extract key information from these descriptions, such as "packaging damage," "sensor malfunction," "human error," and "environmental interference." This key information is then matched against a pre-set problem cause classification database within the system to obtain a preliminary problem classification. The pre-set problem cause classifications may include, but are not limited to: sensing equipment malfunction, environmental interference, human error, and changes in material properties.

[0086] In practical applications, when submitting verification results, inspectors also assess the confidence level of their results, such as high, medium, and low confidence levels. The system then uses this confidence assessment to revise the initial problem classification. For example, if the initial classification is "sensor failure," but the inspector's confidence assessment is low, the system may consider other possibilities or reduce the weight of that classification. The revised problem classification aims to improve the accuracy of problem cause determination. Furthermore, the system assigns weights to the revised problem classification based on the magnitude of the deviation in the identified actual number of deviations. For example, if the deviation is very large, indicating a serious problem, the weight of the problem classification related to that deviation (such as "serious sensor device failure") will increase accordingly; if the deviation is small, it may indicate a minor, intermittent problem, and the weight of the relevant classification will decrease. This weight allocation allows the model to respond more sensitively to significant deviations and prioritize the problems that caused them. Therefore, based on the final weighted problem classification, the system will selectively adjust the weights and parameters of the corresponding sensing devices in the confidence assessment model. For example, if a weighted classification problem indicates a fault in a machine vision system, its weight in the confidence assessment model might be reduced, or its parameters (such as image recognition thresholds and calibration parameters) might be adjusted to minimize its impact on the overall material quantity judgment, or to prompt it to self-calibrate. This adjustment is dynamic and adaptive, designed to improve the model's accuracy in perceiving future material quantities.

[0087] This application's solution achieves adaptive adjustment of the confidence assessment model by introducing refined analysis of the deviation range of actual quantities, the problem descriptions of verification personnel, and confidence level assessments. Through the above technical solution, this application enables dynamic and adaptive optimization of the confidence assessment model in material inventory data management methods. Specifically, by quantifying the deviation of actual quantities, intelligently identifying the causes of problems, and allocating weights based on the deviation range, the model can more accurately locate the source of problems in sensing equipment and make targeted adjustments. This significantly improves the accuracy and efficiency of the data reconciliation process, reduces the frequency and cost of manual intervention, and enhances the system's adaptability to various anomalies. Ultimately, this application ensures the consistency and accuracy of material inventory data, providing enterprises with more reliable inventory management decision support.

[0088] In some embodiments described above in this application, when assigning weights to the corrected problem classifications based on the deviation magnitude of the actual quantities, the characteristics of the materials themselves (such as value level), the urgency of the operation, and the volatility of historical material consumption may not be fully considered. This single or fixed weight allocation method may result in insufficiently refined handling of deviations under different importance or scenarios, thus affecting the accuracy and efficiency of confidence assessment model adjustments. For example, for high-value materials or urgent operations, even small deviations should be given higher weights to trigger more sensitive model adjustments, while for low-value materials or routine operations, a certain range of fluctuation can be allowed. If the above problems are not addressed, the robustness of model adjustments may be insufficient, making it unable to effectively cope with diverse material management scenarios.

[0089] In response, this application further proposes a more refined weight allocation mechanism. By comprehensively considering the value level information of materials, the operational urgency level information, and historical consumption fluctuation information, the weight allocation coefficient of the corrected problem classification is dynamically adjusted to obtain a more targeted weighted problem classification.

[0090] According to the above material inventory data management method, the step of assigning weights to the corrected problem classifications based on the deviation range of the actual quantities to obtain weighted problem classifications includes: Obtain information on the value level, operational urgency level, and historical consumption fluctuations of the materials to be analyzed; Based on the value grade information of the materials, determine the sensitivity to deviations for high-value materials; Based on the urgency level information of the operation, determine the priority of the deviation response to the emergency operation; Based on the historical consumption fluctuation information, determine the inherent fluctuation range of the material; The deviation of the actual quantity is comprehensively judged in conjunction with the deviation sensitivity of the high-value material, the deviation response priority of the emergency operation, and the inherent fluctuation range of the material. When the deviation of the actual quantity exceeds the inherent fluctuation range, and the material is a high-value material or the operation is an emergency operation, the weight allocation coefficient of the corrected problem classification is increased. When the deviation of the actual quantity is within the inherent fluctuation range, and the material is a low-value material or the operation is a non-urgent operation, the weight allocation coefficient of the corrected problem classification is reduced. The corrected problem classification is weighted according to the weight allocation coefficients after the increase or decrease.

[0091] Specifically, acquiring the value level, operational urgency level, and historical consumption fluctuation information of the materials to be analyzed refers to the system extracting attribute data related to the materials currently being analyzed from material master data, operation logs, or historical databases. The value level information can be based on classifying materials according to their unit price, strategic importance, or scarcity, such as classifying them as high-value, medium-value, and low-value materials. The operational urgency level information can be an assessment of the time requirements or impact on the production plan for current material operations (such as warehousing, outbound, and inventory counting), such as classifying them as urgent, routine, or non-urgent operations. Historical consumption fluctuation information can be obtained by analyzing the material's consumption data over a past period, calculating its standard deviation, coefficient of variation, or fluctuation range to reflect the stability of material demand.

[0092] Based on the value grade information of the materials, the deviation sensitivity for high-value materials is determined. This means that for materials with higher value grades, even small quantity deviations should be considered as having a higher risk, and therefore the system will set a higher sensitivity threshold. For example, for materials with a value grade of "high," the deviation sensitivity can be set to 0.5%, meaning that if the deviation between the actual quantity and the expected quantity exceeds 0.5%, it should attract high attention.

[0093] Based on the urgency level information of the operations, the deviation response priority for urgent operations is determined. This means that for operations with higher urgency levels, the system will assign them a higher response priority to ensure that deviations can be quickly identified and handled once they occur. For example, for operations with an urgency level of "urgent," the deviation response priority can be set to the highest level to ensure that model adjustments can reflect problems in a timely manner.

[0094] Determining the inherent fluctuation range of a material based on historical consumption fluctuation information means calculating a reasonable and acceptable range of quantity fluctuations based on the material's historical consumption data. For example, if the historical monthly consumption of a certain material fluctuates between 1000 ± 50 units, then its inherent fluctuation range can be set to ± 5%.

[0095] The comprehensive judgment based on the deviation of the actual quantity, the deviation sensitivity of the high-value material, the deviation response priority of the emergency operation, and the inherent fluctuation range of the material refers to the system comparing and logically reasoning with the currently detected deviation of the actual quantity against the aforementioned determined parameters. For example, if the deviation of the actual quantity is 10 units, and the material is a high-value material with a deviation sensitivity of 5 units, then the deviation is considered to exceed the sensitivity.

[0096] When the deviation of the actual quantity exceeds the inherent fluctuation range, and the material is a high-value material or the operation is an emergency operation, the weight allocation coefficient of the corrected problem classification is increased. This means that when the above comprehensive judgment result is a high-risk situation, the system will increase the weight coefficient related to that problem classification. For example, if the deviation exceeds the inherent fluctuation range and the material is a high-value material, then the weight coefficient related to problem classifications such as "sensor failure" or "human error" will be significantly increased to enable the model to learn and adapt to such high-risk deviations more quickly.

[0097] When the deviation of the actual quantity is within the inherent fluctuation range, and the material is a low-value material or the operation is a non-urgent operation, reducing the weight allocation coefficient of the corrected problem classification means that when the above comprehensive judgment result is a low-risk scenario, the system will reduce the weight coefficient related to that problem classification. For example, if the deviation is within the inherent fluctuation range and the material is a low-value material, then the weight coefficients related to problem classifications such as "environmental disturbance" or "normal loss" will be appropriately reduced to avoid the model overreacting to normal fluctuations.

[0098] The weight allocation for the corrected problem classifications, based on the increased or decreased weight allocation coefficients, refers to the system recalculating the final weight of each corrected problem classification according to the dynamically adjusted weight allocation coefficients, thereby obtaining weighted problem classifications. These weighted problem classifications will be used to subsequently adjust the weights and parameters in the confidence assessment model.

[0099] This application's solution, by introducing material value level information, operational urgency level information, and historical consumption fluctuation information, achieves a more refined and contextualized weight allocation for the corrected problem classification corresponding to the magnitude of actual quantity deviations. Through this technical solution, this application can dynamically and intelligently allocate weights to the problem classification of data discrepancies based on the actual importance of the material, the urgency of the operation, and the material's own fluctuation characteristics. This allows the confidence assessment model to more accurately reflect the true impact and importance of deviations in different scenarios during adjustment, avoiding the "one-size-fits-all" problem that may result from traditional fixed weight allocation. Consequently, the targeting and effectiveness of model adjustment are significantly improved, especially when dealing with deviations in high-value materials or urgent operations. It can identify the root cause of the problem and optimize the model more quickly and accurately, thereby improving the refinement level of material inventory data management and the robustness of the system.

[0100] Based on the above weighted problem classification, the weights and parameters of the corresponding sensing devices in the confidence assessment model are adjusted, including: Based on the weighted problem classification, determine the sensing devices that need adjustment and their adjustment directions; Based on a preset adjustment strategy, the weights and parameters of the sensing device are incrementally adjusted to obtain the adjusted model. Acquire and adjust relevant historical scenario data; The historical scene data is retrospectively evaluated using the adjusted model to obtain the retrospective evaluation results. Acquire forward-looking scenario data related to adjustments; The adjusted model is used to perform a forward-looking evaluation of the forward-looking scenario data to obtain forward-looking evaluation results. Compare the retrospective evaluation results, the prospective evaluation results, and the performance of the model before adjustment; If the comparison results show that the performance of the model before adjustment has decreased or the decrease exceeds a preset threshold, the model is restored to the state before adjustment, and a model performance anomaly report is generated. If the comparison results show that the performance of the model before the adjustment is improved or remains stable, then the adjustment will take effect; Regularly cross-validate the confidence assessment models for all sensing devices.

[0101] Specifically, based on the aforementioned weighted problem classification, the system first determines which sensing devices (such as machine vision systems and microgravity fluctuation capture devices) require adjustments to their confidence assessment models, and clarifies the specific adjustment direction, such as increasing the weight of a certain parameter or decreasing the weight of another parameter. Here, sensing devices refer to various sensors or systems used to acquire information about the physical state of materials; the adjustment direction can be understood as how the contribution of this device in the confidence assessment should change for specific problem causes.

[0102] Furthermore, based on a preset adjustment strategy, such as gradient descent, heuristic rules, or expert system rules, the weights and parameters of the sensing device are adjusted incrementally and in small increments. This incremental adjustment aims to avoid the instability that may result from a large, one-time modification, ensuring the smoothness and controllability of the adjustment process. This yields an adjusted model.

[0103] To verify the effectiveness of the adjustments, it is necessary to obtain historical scenario data related to the adjustments. This historical scenario data consists of actual past problem cases that have been manually verified, along with their corresponding perception data. The purpose is to evaluate the performance of the adjusted model when handling known problems. By retrospectively evaluating the historical scenario data using the adjusted model, the retrospective evaluation results can be obtained, i.e., the model's performance on the historical data.

[0104] Simultaneously, to assess the model's adaptability to future scenarios, it is also necessary to acquire forward-looking scenario data relevant to the adjustments. This data can be simulated, predicted, or recent unverified operational data, the purpose of which is to test the model's ability to generalize to unknown or new situations. By performing a forward-looking evaluation on the forward-looking scenario data using the adjusted model, a forward-looking evaluation result can be obtained, i.e., the model's predictive performance on future data.

[0105] Subsequently, the retrospective evaluation results, the prospective evaluation results, and the model performance before adjustment are compared. This comparison aims to comprehensively measure the impact of the adjustment on model performance, including its explanatory power for historical issues and its predictive power for future situations.

[0106] If the comparison results show that the performance of the model before adjustment has decreased or the decrease exceeds a preset threshold, such as an accuracy reduction of more than 5% or a significant increase in the false alarm rate, the system will restore the model to its state before adjustment to avoid introducing negative impacts and generate a model performance anomaly report so that manual intervention can be used to analyze the cause.

[0107] Conversely, if the comparison results show that the performance of the model before the adjustment has improved or remained stable, for example, the accuracy has increased or remained within an acceptable range, and the false positive rate has not increased significantly, then the adjustment will take effect and the adjusted model will be officially deployed.

[0108] Furthermore, to ensure that the confidence evaluation models for all sensing devices remain optimal in the long term, the system periodically performs cross-validation on these models. This cross-validation can be understood as training and testing the models on different datasets to evaluate their generalization ability and stability. Its purpose is to identify potential performance degradation or biases and to maintain and optimize them in a timely manner.

[0109] This application's solution effectively addresses the potential performance degradation issue that may occur during parameter adjustments in the aforementioned confidence assessment models by introducing a rigorous model adjustment and verification mechanism. By comprehensively comparing the retrospective and prospective assessment results with the model performance before adjustment, the system can comprehensively and objectively determine the actual effect of the adjustment. If the assessment results show a performance degradation or a degradation exceeding a preset threshold, the system immediately reverts to the pre-adjustment model state and generates an anomaly report, effectively preventing the deployment of erroneous or harmful adjustments and ensuring the lower limit of model performance. Conversely, if performance improves or remains stable, the adjustment takes effect, enabling continuous model optimization. Furthermore, regular cross-validation of the confidence assessment models for all sensing devices further ensures that the sensing device models throughout the system maintain a high level of accuracy and reliability over the long term, promptly identifying and correcting potential performance deviations, thereby continuously improving the overall accuracy of material inventory data management.

[0110] This application further proposes a procedure for periodically cross-validating the confidence assessment model for all sensing devices, including: During the preset maintenance window or a period of low system load, select one or a group of devices from all deployed sensing devices as the target of this verification. Generate a set of test cases for the selected sensing device model, the test cases covering typical operating scenarios and abnormal situations that the selected sensing device may encounter in different contexts; The confidence assessment model of the selected sensing device is temporarily isolated, and the weights and parameters are adjusted in a preset, small-range simulation. The adjusted sensing device model is run in a simulated environment using the test cases, and the performance of the adjusted sensing device model on the test cases is evaluated to obtain the simulated performance of the adjusted sensing device model. The performance of the simulated and adjusted sensing device model is compared with its performance before adjustment and the performance of other unadjusted sensing device models on the same test cases.

[0111] Specifically, the preset maintenance window or period of low system load refers to the time period with the least impact on system performance, planned by the system administrator or automated scheduling system based on business needs and system resource status, such as nighttime, weekends, or off-peak periods. During this period, one or a group of devices can be selected from all deployed sensing devices, such as machine vision systems and microgravity fluctuation capture devices, as the target of this verification for systematic performance checks.

[0112] This involves generating a set of test cases for the selected sensing device model. This can be understood as constructing a series of datasets simulating real-world operating scenarios and potential anomalies based on the sensing device's historical data, operation logs, and expert experience. These test cases aim to comprehensively evaluate the sensing device model's recognition accuracy, robustness, and responsiveness under different conditions. For example, test cases may include material recognition under different lighting conditions, quantity statistics under partial occlusion, recognition of different packaging types, and data acquisition under slight vibration or airflow disturbances.

[0113] In practical applications, temporarily isolating the confidence assessment model of the selected sensing device means switching or bypassing the real-time data processing path of the model to be verified, without affecting the normal operation of the current production system, so that it can be tested in an independent simulation environment. In this isolated state, the model's weights and parameters can be preset and adjusted within a small range. These adjustments can be based on optimization directions derived from historical performance data analysis, or they can be used to explore the model's sensitivity to parameter changes.

[0114] Furthermore, running the adjusted sensing device model in a simulated environment using the test cases and evaluating the performance of the adjusted sensing device model on the test cases means inputting the generated test cases into the simulated and adjusted model, observing its output results, and comparing them with the expected correct results, thereby quantifying the model's accuracy, recall, F1 score, and other performance indicators to obtain the performance of the simulated and adjusted sensing device model.

[0115] Therefore, the performance of the simulated and adjusted sensing device model is compared with its performance before adjustment and the performance of other unadjusted sensing device models on the same test cases. The purpose is to evaluate the effectiveness of this simulation adjustment and ensure the consistency and synergy between the adjusted model and other unadjusted models. This comparison helps to discover potential deviations between models and ensures that the data output of the entire sensing network has a unified confidence standard.

[0116] This application's solution effectively addresses the performance drift and cumulative error issues that can arise from relying solely on reactive adjustments in the overall sensing system by introducing a periodic cross-validation mechanism. Through this technical solution, the application enables systematic and periodic health checks and performance calibration of the confidence assessment models for all sensing devices. This not only overcomes the limitations of relying solely on reactive adjustments, effectively preventing long-term model performance drift and cumulative errors, but also minimizes the impact on daily operations by operating within the maintenance window. Furthermore, by generating comprehensive test cases and conducting simulation adjustments, model optimization can be explored in a controlled environment, ensuring the model's robustness under various complex scenarios. This proactive and preventative maintenance strategy significantly improves the data accuracy, reliability, and stability of the entire material inventory data management system, reducing the risk of data reconciliation failures or erroneous warnings due to performance degradation of sensing device models, thereby providing enterprises with a more accurate and reliable view of material inventory data.

[0117] refer to Figure 2 , Figure 2 This is a schematic diagram of a material inventory data management system provided in an embodiment of the present invention. The system includes: The detection end is used to acquire material operation requests and associate an operation credential with the material operation request; based on the operation credential, it queries the material identification information and current status information from multiple material management systems. The comparison end is used to compare the identification information and determine a unified material identity based on the inherent properties of the material; compare the current status information to identify the quantity of the material; if the material is a specific material, then obtain the physical change signal of the material and obtain the visual change information of the material through the imaging device; compare the physical change signal, the visual change information and the material operation request to identify the difference between the physical and logical data. The processing unit is configured to initiate a data reconciliation process and generate a unified and corrected material data view if the discrepancies exist; send a pre-submission instruction to the multiple material management systems, the pre-submission instruction being used to update internal data according to the unified and corrected material data view; receive confirmation signals from the multiple material management systems, and, upon confirmation, instruct the multiple material management systems to record material operations; perform material inventory duration analysis based on the recorded material operation data to identify materials that have not been used for a long time; perform material demand trend analysis based on the officially recorded material operation data to predict future material consumption; and generate material management early warning information based on the material inventory duration analysis results and the material demand trend analysis results.

[0118] Traditional material management systems face challenges when enterprises expand, including incompatible coding rules across different material management systems and difficulties in monitoring new materials, leading to dual contamination of physical and logical data. This makes it difficult for enterprises to obtain a true and unified inventory view, affecting the identification of stagnant materials, accurate prediction of inventory levels, and timely warnings of urgent material shortages, potentially resulting in a complete loss of control over material management. To address this, this application proposes a material inventory data management system. The system acquires material operation requests and associates them with operation vouchers at the detection end, then queries multiple material management systems for material identification and current status information based on these vouchers. The comparison end compares the identification information to determine a unified material identity and compares the current status information to identify the material quantity. For specific materials, the comparison end also acquires physical change signals and visual change information and compares them with the material operation requests to identify differences between physical and logical data. When discrepancies exist, the processing end initiates a data reconciliation process to generate a unified and corrected material data view and coordinates data updates and recording across multiple material management systems. Furthermore, the processing end analyzes material in-stock duration and material demand trends based on the recorded material operation data, ultimately generating material management early warning information. This system, through modular design and collaborative operation, aims to solve problems in traditional material management such as inconsistencies between logical and physical data, difficulties in identifying stagnant materials, inaccurate inventory forecasting, and failure of early warning systems, providing a comprehensive, accurate, and intelligent material inventory data management solution.

[0119] The material inventory data management system proposed in this application achieves comprehensive and accurate management of material inventory data through the collaborative work of its detection, comparison, and processing ends.

[0120] Specifically, the detection terminal is used to acquire material operation requests and associate an operation credential with each request. Based on the operation credential, it queries multiple material management systems for material identification information and current status information. The detection terminal can be implemented as a standalone software module deployed on a central server, responsible for data interaction with user interfaces, automated equipment interfaces, and other material management systems. For example, the detection terminal can be configured with an API interface to receive material operation requests from Enterprise Resource Planning (ERP) systems, Warehouse Management Systems (WMS), or Manufacturing Execution Systems (MES), and automatically generate a unique transaction ID as the operation credential. Furthermore, the detection terminal can proactively send query requests to these systems via message queues or database connections to obtain material identification information (such as material code and name) and current status information (such as inventory quantity and location). As an alternative implementation, the detection terminal can also be integrated into an existing material management system as a functional submodule, performing the above functions through internal calls.

[0121] The comparison terminal is used to compare the identification information to determine a unified material identity based on the inherent properties of the material; compare the current state information to identify the quantity of the material; if the material is a specific material, it acquires the physical change signal of the material and obtains the visual change information of the material through an imaging device; it compares the physical change signal, the visual change information, and the material operation request to identify the differences between the physical and logical data. The comparison terminal can be implemented as a data processing engine, which internally includes a data parsing module, a comparison algorithm module, and a sensor data interface module. For example, the data parsing module is responsible for extracting key information from multi-source data received from the detection terminal; the comparison algorithm module compares the material identification information from different sources according to preset rules or machine learning models to determine a unified material identity and perform preliminary identification of the material quantity. For specific materials, the comparison terminal can also acquire physical change signals and visual change information in real time through interfaces with external sensors (such as microgravity fluctuation capture devices, RFID readers) and imaging devices (such as industrial cameras). These physical and visual data are then cross-validated by the comparison endpoint against material handling requests to identify potential discrepancies between physical inventory and logical book data. As a preferred implementation, the comparison endpoint can employ a distributed computing architecture to handle large-scale, real-time data comparison tasks.

[0122] In practical applications, the processing end is used to initiate a data reconciliation process and generate a unified and corrected material data view if the discrepancies exist; send pre-submission instructions to the multiple material management systems, which are used to update internal data according to the unified and corrected material data view; receive confirmation signals from the multiple material management systems, and after confirmation, instruct the multiple material management systems to record material operations; perform material inventory duration analysis based on the recorded material operation data to identify materials that have not been used for a long time; perform material demand trend analysis based on the officially recorded material operation data to predict future material consumption; and generate material management early warning information based on the results of the material inventory duration analysis and the material demand trend analysis. The processing end can be implemented as a core business logic layer, which integrates a data reconciliation module, a data synchronization module, an analysis module, and an early warning module. For example, when the comparison end identifies data discrepancies, the data reconciliation module is activated, and uses complex algorithms (such as confidence-weighted fusion and expert system rules) to eliminate the discrepancies and generate an accurate and unified material data view. The data synchronization module is responsible for sending this view to each material management system via pre-submission instructions and, upon receiving confirmation signals, instructing each system to perform final data recording. The analysis module utilizes the officially recorded material operation data to perform material inventory duration analysis and material demand trend analysis, such as using time series forecasting models to predict future consumption. The early warning module generates and publishes material management early warning information based on the analysis results. The processing end can be deployed on a high-performance server cluster to ensure the real-time performance and accuracy of data processing and analysis.

[0123] The material inventory data management system proposed in this application represents a significant advancement compared to existing technologies. Traditional material management systems often operate in isolation, resulting in severe data silos, inconsistencies in material coding, and inaccurate inventory data. This is particularly problematic when dealing with new or high-value materials, where discrepancies between physical and logical data are difficult to identify and reconcile. This system constructs an integrated data management architecture through clearly defined detection, comparison, and processing ends. The detection end provides a unified entry point and voucher association for multi-source material operation requests, ensuring the integrity of data tracking from the source. The comparison end not only unifies identification information from multiple systems but also innovatively introduces a fusion comparison mechanism of physical change signals and visual change information, greatly improving the accuracy of inventory data for specific materials and effectively solving the problem of insufficient accuracy of traditional sensors. Building upon this foundation, the processing end ensures the consistency of physical and logical data through an intelligent data reconciliation process and further provides functions for inventory duration analysis and demand trend prediction, thereby generating more accurate and forward-looking material management early warning information. Therefore, this system can comprehensively improve the level of precision and decision-making efficiency of material inventory management, effectively avoid problems such as the backlog of stagnant materials and the shortage of emergency materials, and overcome the technical dilemmas of data chaos and prediction failure in existing technologies.

[0124] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for managing material inventory data, characterized in that, include: Obtain a material operation request and associate an operation document with the material operation request; Based on the operation voucher, query the material identification information and current status information from multiple material management systems; By comparing the identification information, a unified material identity is determined based on the inherent properties of the material; By comparing the current status information, the quantity of the material is identified; If the material is a specific material, then the physical change signal of the material is acquired, and the visual change information of the material is acquired through the imaging device; By comparing the physical change signal, the visual change information, and the material handling request, the differences between the physical and logical data are identified. If the aforementioned discrepancies exist, the data reconciliation process is initiated, and a unified and corrected view of the material data is generated; Send a pre-submission instruction to the plurality of material management systems, the pre-submission instruction being used to update internal data based on the unified and corrected material data view; Receive confirmation signals from the plurality of material management systems, and after confirmation, instruct the plurality of material management systems to record material operations; Based on the recorded material handling data, analyze the duration of material in stock to identify materials that have not been used for a long time; Based on the officially recorded material handling data, material demand trend analysis is performed to predict future material consumption. Based on the analysis results of the material's inventory duration and the analysis results of the material demand trend, material management early warning information is generated.

2. The material inventory data management method according to claim 1, characterized in that, If the aforementioned discrepancies exist, a data reconciliation process is initiated to generate a unified and corrected material data view, including: Collect contextual information about the storage area, including local airflow disturbance intensity information and material packaging type information; Based on the context information, extract the numerical features of the context information, including the airflow disturbance intensity value and the packaging identification matching degree value; Based on the airflow disturbance intensity value and the packaging identification matching degree value, the confidence score of the physical perception data source is calculated. The physical perception data source includes machine vision system and microgravity fluctuation capture device. Based on the confidence score, the quantity of materials reported by the machine vision system and the quantity of materials inferred by the microgravity fluctuation capture device are weighted and fused to obtain a preliminary physical quantity. Compare the preliminary physical quantities with the expected operational quantities, and assess the deviation between the preliminary physical quantities and the expected operational quantities; When the deviation exceeds a preset threshold, the confidence level of the overall physical verification result is reduced, an alarm is triggered, and a verification work order is generated. Receive the actual quantity and the reason for the problem after manual verification. The reason for the problem includes the verification personnel's description of the problem and the confidence assessment of the verification results. Based on the actual number and the cause of the problem, adjust the weights and parameters in the confidence assessment model.

3. The material inventory data management method according to claim 1, characterized in that, The analysis of material in-stock duration based on recorded material handling data is used to identify materials that have not been used for a long time, including: Obtain information on the material type, historical inbound and outbound records, and current production project information of the material to be analyzed; Based on the material type information, query the baseline in-stock duration threshold for the material; Based on the seasonal fluctuation patterns in the historical inbound and outbound records, the benchmark inbound duration threshold is periodically adjusted. The in-stock duration threshold is adjusted in stages based on the project stage information in the production project information. The adjusted in-stock duration threshold is compared with the actual in-stock duration of the material to be analyzed; If the actual in-stock duration exceeds the adjusted in-stock duration threshold, the material to be analyzed will be marked as a long-term unused material.

4. The material inventory data management method according to claim 1, characterized in that, The process of performing material demand trend analysis based on the officially recorded material operation data to predict future material consumption includes: Obtain the type information of the material to be analyzed; Determine whether the historical operation data of the material to be analyzed is sufficient; If the historical operation data is insufficient, query the historical consumption data of similar or alternative materials. Based on the bill of materials information in the production plan, deduce the basic consumption amount; Adjust the basic consumption based on the new production line's capacity ramp-up plan or the product life cycle stage; When the historical operation data is sufficient but the pattern is inconsistent with the future, an external event is identified; Based on the external events, the historical operation data is weighted or truncated. The base consumption level is adjusted by combining current orders, forecasted orders, and safety stock strategy; Generate predictions of future material consumption.

5. A material inventory data management method according to claim 2, characterized in that, When the deviation exceeds a preset threshold, the confidence level of the overall physical verification result is reduced, an alarm is triggered, and a verification work order is generated, including: Obtain the value level information of the material to be operated and the urgency level information of the current operation; Based on the value level information, the preset threshold is adjusted to obtain the adjusted deviation threshold; Based on the urgency level information, the alarm level is adjusted to obtain the adjusted alarm level; Based on the value level information of the material and the urgency level information of the operation, adjust the confidence reduction of the overall physical verification result and adjust the priority and response time of the verification work order; When the deviation between the initial physical quantity and the expected operation quantity exceeds the adjusted deviation threshold, the confidence level of the overall physical verification result is reduced, and an alarm corresponding to the adjusted alarm level is triggered, generating a verification work order with the adjusted priority and response time limit.

6. The material inventory data management method according to claim 2, characterized in that, The step of adjusting the weights and parameters in the confidence assessment model based on the actual quantity and the cause of the problem includes: By comparing the actual quantity with the expected number of operations recorded by the system and historical operation data, the deviation of the actual quantity is identified. The problem descriptions provided by the inspectors are analyzed, key information in the descriptions is extracted, and matched with preset problem cause classifications to obtain a preliminary problem classification. Based on the preliminary problem classification and the confidence assessment of the verification results, the preliminary problem classification is revised to obtain the revised problem classification; Based on the deviation range of the actual quantity, the corrected problem classification is weighted to obtain a weighted problem classification; Based on the weighted problem classification, the weights and parameters of the corresponding sensing devices in the confidence assessment model are adjusted.

7. A material inventory data management method according to claim 6, characterized in that, The step of assigning weights to the corrected problem classification based on the deviation magnitude of the actual quantity to obtain a weighted problem classification includes: Obtain information on the value level, operational urgency level, and historical consumption fluctuations of the materials to be analyzed; Based on the value grade information of the materials, determine the sensitivity to deviations of high-value materials; Based on the urgency level information of the operation, determine the priority of the deviation response to the emergency operation; Based on the historical consumption fluctuation information, determine the inherent fluctuation range of the material; The deviation of the actual quantity is comprehensively judged in conjunction with the deviation sensitivity of the high-value material, the deviation response priority of the emergency operation, and the inherent fluctuation range of the material. When the deviation of the actual quantity exceeds the inherent fluctuation range, and the material is a high-value material or the operation is an emergency operation, the weight allocation coefficient of the corrected problem classification is increased; When the deviation of the actual quantity is within the inherent fluctuation range, and the material is a low-value material or the operation is a non-urgent operation, the weight allocation coefficient of the corrected problem classification is reduced. The corrected problem classification is weighted according to the weight allocation coefficients after the increase or decrease.

8. A material inventory data management method according to claim 6, characterized in that, The step of adjusting the weights and parameters of the corresponding sensing devices in the confidence assessment model according to the weighted problem classification includes: Based on the weighted problem classification, determine the sensing devices that need adjustment and their adjustment directions; Based on a preset adjustment strategy, the weights and parameters of the sensing device are incrementally adjusted to obtain the adjusted model. Acquire and adjust relevant historical scenario data; The historical scene data is retrospectively evaluated using the adjusted model to obtain the retrospective evaluation results. Acquire forward-looking scenario data related to adjustments; The adjusted model is used to perform a forward-looking evaluation of the forward-looking scenario data to obtain forward-looking evaluation results. Compare the retrospective evaluation results, the prospective evaluation results, and the performance of the model before adjustment; If the comparison results show that the performance of the model before adjustment has decreased or the decrease exceeds a preset threshold, the model is restored to the state before adjustment, and a model performance anomaly report is generated. If the comparison results show that the performance of the model before the adjustment is improved or remains stable, then the adjustment will take effect; Regularly cross-validate the confidence assessment models for all sensing devices.

9. A material inventory data management method according to claim 8, characterized in that, The periodic cross-validation of the confidence assessment model for all sensing devices includes: During the preset maintenance window or a period of low system load, select one or a group of devices from all deployed sensing devices as the target of this verification. Generate a set of test cases for the selected sensing device model, the test cases covering typical operating scenarios and abnormal situations that the selected sensing device may encounter in different contexts; The confidence assessment model of the selected sensing device is temporarily isolated, and the weights and parameters are adjusted in a preset, small-range simulation. The adjusted sensing device model is run in a simulated environment using the test cases, and the performance of the adjusted sensing device model on the test cases is evaluated to obtain the simulated performance of the adjusted sensing device model. The performance of the simulated and adjusted sensing device model is compared with its performance before adjustment and the performance of other unadjusted sensing device models on the same test cases.

10. A material inventory data management system, characterized in that, The system includes: The detection end is used to acquire material operation requests and associate an operation credential with the material operation request; based on the operation credential, it queries the material identification information and current status information from multiple material management systems. The comparison end is used to compare the identification information and determine a unified material identity based on the inherent properties of the material; compare the current status information to identify the quantity of the material; if the material is a specific material, then obtain the physical change signal of the material and obtain the visual change information of the material through the imaging device; compare the physical change signal, the visual change information and the material operation request to identify the difference between the physical and logical data. The processing unit is configured to initiate a data reconciliation process and generate a unified and corrected material data view if the discrepancies exist; send a pre-submission instruction to the multiple material management systems, the pre-submission instruction being used to update internal data according to the unified and corrected material data view; receive confirmation signals from the multiple material management systems, and, upon confirmation, instruct the multiple material management systems to record material operations; perform material inventory duration analysis based on the recorded material operation data to identify materials that have not been used for a long time; perform material demand trend analysis based on the officially recorded material operation data to predict future material consumption; and generate material management early warning information based on the material inventory duration analysis results and the material demand trend analysis results.

Images