A warehouse inventory identification system and method for electrical supplies

Abstract

This invention relates to the field of power material warehousing management technology, and particularly to a warehouse inventory identification system and method for power materials, comprising: data acquisition, dynamic verification and early warning, predictive analysis and early warning, task fusion and scheduling, instruction execution and feedback, feedback data management, and model optimization modules. It generates anomaly warnings by real-time comparison of book and physical data streams, and predicts shortage and obsolescence risks by combining historical data; it intelligently fuses and sorts multi-source warnings to generate inventory instructions; it executes physical inventory and traces the root causes of discrepancies based on the law of conservation of inventory, completing account adjustments; it associates and stores all process data to form feedback data, and automatically optimizes warning thresholds and predictive models accordingly, achieving intelligent management of the entire process from real-time monitoring and intelligent early warning to closed-loop self-optimization, effectively solving the problems of traditional inventory lag and resource misallocation.

Description

Technical Field

[0001] This invention relates to the field of power material warehousing management technology, and in particular to a warehouse inventory identification system and method for power materials. Background Technology

[0002] Power material warehouses are crucial hubs for ensuring the safe and stable operation of the power grid. The accuracy of their inventory records directly impacts emergency repair efficiency, project material supply, and the safety of enterprise assets. Currently, mainstream power material warehouse management widely adopts warehouse management systems for record-keeping and incorporates IoT technologies such as barcodes and RFID to assist in the collection of physical inventory entry and exit information, achieving information-based management.

[0003] Currently, warehouse management of power materials has widely adopted warehouse management systems combined with Internet of Things (IoT) technology for information-based operations. However, existing technical solutions have a fundamental flaw: they rely on a manually initiated, periodic, static inventory model that consumes resources evenly. This model cannot identify high-value materials that are truly at risk of shortage or obsolescence in real time and accurately during dynamic business processes, leading to misallocation of management resources and delayed risk response. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides a warehouse inventory identification system and method for power materials, aiming to improve the existing periodic static inventory mode, which cannot identify high-value materials that are truly at risk of shortage or obsolescence in real time and accurately during dynamic business processes, leading to the problems of misallocation of management resources and delayed risk response.

[0005] In a first aspect, the present invention provides the following technical solution: a warehouse inventory and identification system for power materials, comprising the following modules:

[0006] The data acquisition module is used to collect the book data stream of power materials leaving the warehouse and the physical movement data stream obtained through IoT sensing devices, and to acquire historical inventory and consumption data.

[0007] The dynamic verification and early warning module is used to perform time-series comparison and analysis of the book outbound data stream and the physical movement data stream. When the analysis result exceeds the preset difference threshold, a first type of early warning signal is generated.

[0008] The predictive analysis and early warning module is used to generate a second type of early warning signal for inventory shortage risk or obsolescence risk based on the historical inventory and consumption data and through predictive model analysis.

[0009] The task fusion and scheduling module is used to receive the first type of early warning signal and the second type of early warning signal, fuse them and prioritize them according to preset rules, and generate a unified inventory task instruction.

[0010] The instruction execution and feedback module is used to respond to the inventory task instruction, perform on-site inventory of the target power materials specified in the instruction, and perform source analysis of discrepancies based on the inventory results and historical business data to complete the inventory account adjustment.

[0011] The feedback data management module is used to associate and store the results of the difference tracing analysis and the inventory account adjustment information to form optimization feedback data.

[0012] The model optimization module is used to optimize and adjust the parameters of the preset difference threshold in the dynamic verification and early warning module and the prediction model in the prediction analysis and early warning module based on the optimization feedback data.

[0013] Preferably, the data acquisition module performs the data acquisition process including the following steps:

[0014] The system connects to the warehouse management system and acquires the book-based outbound records of power materials according to a preset cycle or in real time, forming the book-based outbound data stream.

[0015] Access the network of IoT sensing devices deployed at warehouse entrances and exits, shelves and handling equipment, obtain the identification of power materials and movement events scanned or sensed by the IoT sensing devices, and form the physical movement data stream;

[0016] Access the historical database of the enterprise resource planning system or warehouse management system, and extract historical inbound records, outbound records and inventory balance records according to material category and time dimension to form the historical inventory and consumption data.

[0017] The book-based outbound data stream, the physical movement data stream, and the historical inventory and consumption data are standardized, cleaned, and synchronized with time, and the processed data is published to the data bus.

[0018] Preferably, the process of the dynamic verification and early warning module performing time-series comparison analysis and early warning includes the following steps:

[0019] Based on the same material identification and unified time benchmark, the book outbound data stream and the physical movement data stream are paired and aligned to generate synchronous time-series data pairs at the material level.

[0020] Based on a preset time window, for each of the synchronized time-series data pairs, calculate its cumulative book outbound volume and cumulative physical movement volume, and calculate the difference rate or absolute difference value between the two.

[0021] Compare the difference rate or absolute difference value with the preset difference threshold;

[0022] When the difference rate or absolute difference value continuously exceeds the preset difference threshold for a preset duration, or exceeds the preset emergency threshold in a single instance, it is determined that the account balance trend is abnormal, and the first type of early warning signal is generated.

[0023] Preferably, the predictive analysis and early warning module performs predictive analysis and early warning in the following steps:

[0024] The historical inventory and consumption data are preprocessed, and a data sample set containing historical consumption sequences, inventory balance sequences and related features is constructed using material identification as the dimension.

[0025] The data sample set is input into a pre-trained prediction model, and the prediction model outputs a predicted value of material consumption for a future preset period and a predicted inventory balance at the end of the future preset period.

[0026] Based on the predicted consumption of materials, the predicted inventory balance, and the preset safety stock threshold, calculate the inventory shortage risk level within the preset period in the future.

[0027] Calculate the degree of deviation between the historical consumption sequence of materials and the predicted consumption value of materials. Based on the degree of deviation and in combination with the inventory age, calculate the material stagnation risk level.

[0028] When the inventory shortage risk level or the material stagnation risk level exceeds its respective preset risk threshold, the second type of early warning signal is generated.

[0029] Preferably, the task fusion and scheduling module performs the task fusion and scheduling process including the following steps:

[0030] Receive and parse the first type of early warning signal and the second type of early warning signal, and extract the early warning type, material identification, risk level and urgency information from the signals;

[0031] Based on the preset material importance mapping table, determine the preset importance level of the material corresponding to the material identity identifier;

[0032] Based on the warning type, the risk level, the urgency information, and the preset importance level, the comprehensive priority score corresponding to each warning signal is calculated according to the preset fusion scoring rules;

[0033] The warning signals pointing to the same material identification are merged, and the highest comprehensive priority score is used as the final task priority of the material.

[0034] Based on the final task priority of all materials, a unified inventory task instruction is generated, which includes a list of material identification identifiers to be inventoried, corresponding task types, and priorities.

[0035] Preferably, the process of the instruction execution and feedback module performing on-site inventory and discrepancy tracing analysis includes the following steps:

[0036] Parse the unified inventory task instructions to obtain the list of target power materials to be inventoried and their corresponding task priorities;

[0037] Based on the identification of the target power materials, query their book storage location information from the warehouse management system, and generate an on-site inventory operation sequence based on the task priority;

[0038] Based on the aforementioned on-site inventory operation sequence, the target power materials are scanned, identified, and counted using mobile inventory equipment to obtain on-site inventory quantity information;

[0039] Retrieve historical business data of the target power materials within a preset traceability period;

[0040] Based on the physical inventory count information, perform forward and reverse tracing analysis on the historical business data based on the law of inventory conservation to locate one or more abnormal business records that cause discrepancies between the accounts and the physical inventory.

[0041] Based on the type and cause of the abnormal business record located, generate and execute the corresponding inventory account adjustment instruction to complete the inventory account adjustment.

[0042] Preferably, the step of locating one or more abnormal business records that cause discrepancies between the accounting records and the actual transactions specifically includes:

[0043] From the historical business data, determine the initial inventory quantity, the quantity of all inbound records, the quantity of all outbound records, and the quantity of all inventory adjustment records within the preset traceability period;

[0044] Based on the law of conservation of inventory, the theoretical inventory quantity at the end of the preset traceability period is calculated using the initial inventory quantity, the quantity of all inbound records, the quantity of all outbound records, and the quantity of all inventory adjustment records.

[0045] The calculated theoretical inventory quantity is compared with the actual inventory quantity information to obtain the discrepancy between the book and physical inventory values.

[0046] If the discrepancy between the account and the actual transaction is not zero, it is determined that there is at least one abnormal business record in the historical business data.

[0047] Using the discrepancy between the accounts and actual inventory as a clue, each business record in the historical business data is traversed in reverse and logically verified to locate one or more abnormal business records that have errors in quantity, material identification, or business time.

[0048] Preferably, the process of the feedback data management module performing associated storage and forming optimized feedback data includes the following steps:

[0049] Obtain the record attribute information of the abnormal business records located by the difference tracing analysis, as well as the adjustment content information of the inventory account adjustment instructions executed based on the abnormal business records;

[0050] The record attribute information of the abnormal business record is associated and bound with the adjustment content information of the corresponding inventory account adjustment instruction;

[0051] Associate the abnormal business records and adjustment information after they are linked with the original early warning signal information that triggered this difference source analysis;

[0052] The complete data records after the association and binding are stored in the optimization feedback database according to the preset data structure to form the optimization feedback data.

[0053] Preferably, the process of parameter optimization and adjustment performed by the model optimization module includes the following steps:

[0054] From the optimization feedback database, optimization feedback data within a preset time range is periodically extracted;

[0055] Based on the optimized feedback data, the verification result distribution of the first type of early warning signal and the verification result distribution of the second type of early warning signal are statistically analyzed respectively.

[0056] Based on the distribution of verification results of the first type of early warning signal, the preset difference threshold in the dynamic verification and early warning module is optimized and adjusted;

[0057] Based on the distribution of verification results for the second type of early warning signal, the prediction model in the prediction analysis and early warning module is optimized and adjusted.

[0058] Secondly, the present invention provides the following technical solution: a warehouse inventory identification method for power materials, the method comprising:

[0059] Collect the data stream of power materials issued from the books and the data stream of physical movement obtained through IoT sensing devices, and obtain historical inventory and consumption data;

[0060] A time-series comparison analysis is performed on the book outbound data stream and the physical movement data stream. When the analysis result exceeds a preset difference threshold, a first type of early warning signal is generated.

[0061] Based on the historical inventory and consumption data, a second type of early warning signal is generated through predictive model analysis to target the risk of inventory shortage or obsolescence.

[0062] Receive the first type of early warning signal and the second type of early warning signal, fuse them and prioritize them according to preset rules, and generate a unified inventory task instruction;

[0063] In response to the inventory task instruction, an on-site inventory is performed on the target power materials specified in the instruction, and a difference tracing analysis is conducted based on the inventory results and historical business data to complete the inventory account adjustment;

[0064] The results of the discrepancy tracing analysis and the inventory account adjustment information are linked and stored to form optimization feedback data;

[0065] Based on the optimized feedback data, the preset difference threshold in the dynamic verification and early warning module and the prediction model in the prediction analysis and early warning module are optimized and adjusted.

[0066] The present invention has the following beneficial effects:

[0067] 1. In this invention, by comparing the accounting data stream with the physical movement data stream sensed by the Internet of Things in real time, the system can immediately detect the deviation between the accounting and physical trends when the business occurs, significantly advancing the discovery point of the difference from the end of the inventory cycle to the process of the business, thus significantly improving the visualization and transparency of the inventory status.

[0068] 2. In this invention, the system proactively identifies inventory shortage and obsolescence risks through a predictive model, integrates real-time early warning signals, and uses an intelligent scheduling algorithm to prioritize massive early warnings, thereby accurately directing limited human and time resources to the highest-risk and highest-value materials, mitigating the greatest business risks with minimal verification costs, and greatly improving warehouse management efficiency.

[0069] 3. This invention not only identifies problems and schedules tasks, but also accurately pinpoints root causes through source analysis based on the law of inventory conservation, and continuously optimizes early warning thresholds and prediction models using disposal result data. This enables the system to have self-learning and self-evolution capabilities, realizing a shift in warehouse management from open-loop control relying on human experience to closed-loop optimization based on data intelligence. Attached Figure Description

[0070] Figure 1 This is a schematic diagram of the architecture of a warehouse inventory and identification system for power materials proposed in this invention.

[0071] Figure 2 This is a flowchart illustrating a warehouse inventory identification method for electrical materials proposed in this invention. Detailed Implementation

[0072] The technical solutions in 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.

[0073] Example 1:

[0074] In a first embodiment of the present invention, the present invention provides a warehouse inventory identification system for electrical materials, such as... Figure 1 As shown, it includes the following steps:

[0075] The data acquisition module is used to collect the book data stream of power materials leaving the warehouse and the physical movement data stream obtained through IoT sensing devices, and to acquire historical inventory and consumption data.

[0076] Furthermore, the data acquisition module performs the following steps during the data acquisition process:

[0077] Connect to the warehouse management system to obtain the book-based outbound records of power materials according to a preset cycle or in real time, forming a book-based outbound data stream;

[0078] Connect to the network of IoT sensing devices deployed at warehouse entrances and exits, shelves and handling equipment, obtain the identification of power materials and movement events scanned or sensed by the IoT sensing devices, and form a physical movement data stream;

[0079] Access the historical database of the enterprise resource planning system or warehouse management system, and extract historical inbound records, outbound records and inventory balance records according to material category and time dimension to form historical inventory and consumption data;

[0080] Standardize and clean the data streams of outbound shipments, physical movement data, and historical inventory and consumption data, and synchronize them with the time. Then publish the processed data to the data bus.

[0081] Specifically, the data acquisition module first connects to the warehouse management system. This warehouse management system is an existing system, such as one built using SAP WM or Oracle WMS. Through the application programming interface provided by the warehouse management system, the data acquisition module acquires each outbound operation record of electrical materials at a preset period, such as every minute or every five minutes, or by monitoring database changes in real time. Each outbound record includes at least the outbound order number, electrical material identification, outbound quantity, and outbound time. The data acquisition module then uses this sequence of outbound records acquired in chronological order to form a data stream of outbound inventory.

[0082] The data acquisition module also connects to a network of IoT sensing devices deployed in the warehouse. This network includes UHF RFID readers fixed at warehouse entrances and exits, QR code scanners installed in each shelf aisle, and mobile data terminals installed on forklifts and other handling equipment. The data acquisition module receives scanning or sensing events reported by these IoT sensing devices in real time through an IoT gateway. Each event includes at least the identification of the electrical equipment being read, the event type, the location of the event, and a timestamp. Event types include outbound scans, inbound scans, and location movement scans. The data acquisition module receives these event sequences in chronological order, forming a physical movement data stream.

[0083] The data acquisition module also connects to the historical database of the Enterprise Resource Planning (ERP) system or warehouse management system. This historical database is a relational database of the existing system, such as MySQL or Oracle. Using Structured Query Language (SCL), the module extracts all inbound records, all outbound records, and daily or weekly inventory snapshots within a preset historical time range, categorized by material type and time. The extracted inbound and outbound records include material identification, quantity, and time. The extracted inventory balance records include material identification, remaining quantity, and record time. The data acquisition module integrates these records to form historical inventory and consumption data.

[0084] The data acquisition module includes a data preprocessing unit. This unit performs standardized cleaning and time synchronization processing on the book-based outbound data stream, physical movement data stream, and historical inventory and consumption data. Standardized cleaning includes: uniformly mapping the identification of the same power materials from different sources to a standard internal code; uniformly converting all quantity units to standard units of measurement; and removing outlier records that clearly exceed reasonable ranges. Time synchronization processing includes: uniformly converting the time information in all records to the Unix timestamp format under the same time zone, accurate to the second.

[0085] After standardized cleaning and time synchronization, the data acquisition module publishes the processed book-based outbound data stream, physical movement data stream, and historical inventory and consumption data as independent message topics to the data bus via a message middleware. The message middleware is either Apache Kafka or RabbitMQ. The data bus is the publish-subscribe channel formed by this message middleware. Downstream dynamic verification and early warning modules, as well as predictive analysis and early warning modules, obtain the required data from the data bus by subscribing to the corresponding message topics.

[0086] The dynamic verification and early warning module is used to perform time-series comparison and analysis of the outbound data stream on the books and the physical movement data stream. When the analysis results exceed the preset difference threshold, a first-type early warning signal is generated.

[0087] Furthermore, the dynamic verification and early warning module performs time-series comparison analysis and early warning processes, including the following steps:

[0088] Based on the same material identification and unified time benchmark, the data stream of outbound shipment in the book and the data stream of physical movement are paired and aligned to generate synchronous time-series data pairs at the material level.

[0089] Based on a preset time window, for each pair of synchronized time-series data, calculate its cumulative book outbound volume and cumulative physical movement volume, and calculate the difference rate or absolute difference value between the two.

[0090] Compare the difference rate or absolute difference value with a preset difference threshold;

[0091] When the difference rate or absolute difference value continuously exceeds the preset difference threshold for a preset duration, or exceeds the preset emergency threshold in a single instance, it is determined to be an abnormal trend in the accounts and generates a first-class warning signal.

[0092] Specifically, the dynamic verification and early warning module subscribes to the book-based outbound data stream and the physical movement data stream processed by the data acquisition module from the data bus. This module is equipped with a stream processing engine, implemented using frameworks such as Apache Flink or Spark Streaming. The dynamic verification and early warning module first performs data pairing and alignment. The stream processing engine groups records from the two data streams according to the same material identification. For each material identification, the stream processing engine performs windowed alignment of records in the two data streams using a unified time base, i.e., system time. Specifically, an alignment time window is set, with a window length of, for example, 10 seconds. Within this window, the number of book-based outbound records and the number of physical movement records with the event type "outbound scan" are aggregated and paired. For each material identification, a synchronous time-series data pair is generated, which includes the start timestamp of the time window, the cumulative book-based outbound quantity within the window, and the cumulative physical movement outbound quantity within the window.

[0093] The dynamic verification and early warning module then performs discrepancy calculations. This module sets an analysis time window, the length of which is greater than the alignment time window, for example, 1 hour. For each material identification identifier, the stream processing engine performs aggregation calculations on multiple consecutively generated synchronous time-series data pairs within this analysis time window. The cumulative total amount of outbound shipments within this analysis time window is calculated and denoted as... Calculate the cumulative total of physical goods moving out of the warehouse within the analysis time window, and record it as . Subsequently, the dynamic verification and early warning module calculates the discrepancy index. The formula for calculating the discrepancy rate is:

[0094] ;

[0095] The formula for calculating the absolute difference is:

[0096] ;

[0097] The module can be configured to use either the difference rate or the absolute difference value as the judgment criterion. The dynamic verification and early warning module compares the calculated difference rate or absolute difference value with a preset difference threshold. This preset difference threshold is a configurable parameter. For example, when using the difference rate, the preset difference threshold can be set to 0.05, or 5%; when using the absolute difference value, the preset difference threshold can be set to a specific number, such as 2.

[0098] The dynamic verification and early warning module has early warning judgment logic. This logic includes two triggering conditions. The first condition is: when the calculated difference rate or absolute difference value continuously exceeds a preset difference threshold for a preset duration, such as two consecutive analysis time windows, i.e., two consecutive hours. The second condition is: when the calculated difference rate or absolute difference value exceeds a preset emergency threshold at a single time, and this emergency threshold is higher than the preset difference threshold, for example, the difference rate emergency threshold is set to 0.15, i.e., 15%. When either of the above triggering conditions is met, the dynamic verification and early warning module determines that the material identification has an abnormal trend in its accounting status. The module then generates a first type of early warning signal. This first type of early warning signal is a structured data object, containing at least the following fields: signal type identifier, triggering material identification, trigger time, type of difference indicator used, specific difference value, and triggering condition type. The generated first type of early warning signal is published to the data bus for the downstream task fusion and scheduling module to subscribe to and use.

[0099] The predictive analysis and early warning module is used to generate a second type of early warning signal for inventory shortage risk or obsolescence risk by analyzing historical inventory and consumption data through predictive models.

[0100] Furthermore, the predictive analysis and early warning module performs predictive analysis and early warning in the following steps:

[0101] Historical inventory and consumption data are preprocessed, and a data sample set containing historical consumption sequences, inventory balance sequences and related features is constructed using material identification as the dimension.

[0102] Input the data sample set into the pre-trained prediction model, and the prediction model outputs the predicted value of material consumption for the future preset period and the predicted inventory balance at the end of the future preset period.

[0103] Based on the predicted consumption of materials, the predicted inventory balance, and the preset safety stock threshold, calculate the inventory shortage risk level within the preset period in the future.

[0104] Calculate the degree of deviation between the historical consumption sequence of materials and the predicted consumption value of materials. Based on the degree of deviation and in combination with the inventory age, calculate the risk level of obsolescence of materials.

[0105] When the risk level of inventory shortage or the risk level of stagnant materials exceed their respective preset risk thresholds, a second type of early warning signal is generated.

[0106] Specifically, the predictive analysis and early warning module acquires historical inventory and consumption data processed by the data acquisition module from the data bus. This module includes a data preprocessing unit, a model inference unit, and a risk calculation unit. The data preprocessing unit preprocesses the historical inventory and consumption data. Preprocessing includes: extracting the daily consumption quantity of each material identifier as an independent dimension, in chronological order, to form a historical consumption sequence for that material. Simultaneously, it extracts the daily inventory balance quantity at the corresponding time point, forming a historical inventory balance sequence for that material. Furthermore, it extracts potentially related features to consumption, including but not limited to the day of the week, month information, and project type codes associated with the consumption obtained from the enterprise resource planning system. The data preprocessing unit combines the historical consumption sequence, historical inventory balance sequence, and related features for each material identifier to form a data sample. The collection of data samples for all materials constitutes the data sample set.

[0107] The predictive analysis and early warning module uses a pre-trained predictive model. This predictive model is a time series prediction model, specifically employing a Long Short-Term Memory (LSTM) network model. This LSTM network model consists of one input layer, two LSTM hidden layers, and one fully connected output layer. The number of neurons in the input layer corresponds to the dimension of the input features. The first LSTM hidden layer has 64 neurons, and the second LSTM hidden layer has 32 neurons. The fully connected output layer has two neurons, outputting the predicted consumption value for a preset period, such as 7 days, and the predicted inventory balance at the end of that period. The model is trained using historical data, predicting values ​​for the next 7 days from the previous N-7 days' sequences, using mean squared error as the loss function, and employing the Adam optimizer for training.

[0108] The model inference unit inputs the constructed data sample set into the pre-trained Long Short-Term Memory (LSTM) network model. The model processes the data sample for each material identification identifier and outputs two values: a predicted consumption value for the next 7 days, denoted as... And the projected inventory balance at the end of day 7, denoted as The risk calculation unit first calculates the inventory shortage risk level. This unit obtains the preset safety stock threshold for the material, denoted as... Calculate the net consumption forecast for the future period as Calculate risk indicators :

[0109] = ;

[0110] in, This is the latest inventory. The risk calculation unit will... Mapped to a preset level, for example: if ≤0, the grade is 0; if 0 < ≤0.3, level 1; if 0.3 < ≤0.7, level 2; if A value greater than 0.7 indicates a level of 3. A higher level signifies a greater risk of shortage.

[0111] The risk calculation unit then calculates the risk level of material stagnation. This unit calculates the deviation between the historical consumption sequence and the predicted material consumption value. Specifically, it calculates the standard deviation of the historical consumption sequence of the material over the past M days, for example, 30 days, denoted as [the standard deviation is missing from the original text]. Calculate the mean of the sequence, denoted as . Calculate the degree of deviation index :

[0112] ;

[0113] At the same time, obtain the current inventory age of the material, that is, the number of days since the earliest batch was put into storage, denoted as . stagnation risk level The value is determined by both the degree of deviation and the age of the stock, and the calculation formula is as follows:

[0114] T);

[0115] in, and Here, represents the weighting coefficient, for example, all set to 0.5, and T is the normalized coefficient for inventory age, for example, 180 days. Mapped to a preset level, for example: if <0.2, grade is 0; if 0.2≤ <0.5, level 1; if 0.5≤ <0.8, level 2; if A value ≥0.8 corresponds to a grade of 3. A higher grade indicates a greater risk of stagnation.

[0116] The early warning generation unit compares the calculated inventory shortage risk level and material stagnation risk level with their respective preset risk thresholds. For example, the preset risk threshold is set to level 2. When either risk level reaches or exceeds level 2, the early warning generation unit generates a second type of early warning signal. This second type of early warning signal is a structured data object containing at least the following fields: signal type identifier, triggering material identifier, trigger time, risk type, risk level, predicted consumption, and predicted inventory balance. The generated second type of early warning signal is then published to the data bus.

[0117] The task fusion and scheduling module is used to receive the first type of early warning signal and the second type of early warning signal, fuse them and prioritize them according to preset rules, and generate a unified inventory task instruction.

[0118] Furthermore, the task fusion and scheduling module performs the task fusion and scheduling process, which includes the following steps:

[0119] Receive and analyze the first and second type of early warning signals, and extract information such as the warning type, material identification, risk level, and urgency from the signals;

[0120] Based on the preset material importance mapping table, determine the preset importance level of the material corresponding to the material identification mark;

[0121] Based on the warning type, risk level, urgency information, and preset importance level, the comprehensive priority score corresponding to each warning signal is calculated according to the preset fusion scoring rules.

[0122] Warning signals pointing to the same material identification will be merged, and the highest comprehensive priority score among them will be used as the final task priority for that material.

[0123] All materials are sorted according to their final task priority, and a unified inventory task instruction is generated, which includes a list of material identification identifiers, corresponding task types, and priorities.

[0124] Specifically, the task fusion and scheduling module subscribes to the first and second types of early warning signals from the data bus. This module includes a signal parsing unit, a rule engine, and a task scheduling unit. The signal parsing unit receives the early warning signals and parses their structured fields. It extracts the early warning type, material identification, risk level, and urgency information from the first type of early warning signal. The early warning type is fixed as "account anomaly," and the risk level is dynamically determined based on the multiple by which the difference exceeds a threshold or the duration of the difference. For example, if the difference exceeds a preset threshold by more than twice or persists for more than three analysis windows, the risk level is set to "high"; otherwise, it is "medium." The urgency information is recorded as the trigger condition type. The module also extracts the early warning type, material identification, risk level, and urgency information from the second type of early warning signal. The early warning type is either "shortage risk" or "stagnant risk," the risk level is the numerical level directly carried in the signal, and the urgency information is uniformly set to "predictive."

[0125] The task fusion and scheduling module internally stores a preset material importance mapping table. This table uses the material's identity identifier as the primary key, associated with a preset importance level. The preset importance level is pre-defined based on factors such as the voltage level of the power material, its criticality in the power grid, and its unit price. For example, it is divided into three levels: "critical," "important," and "general," represented by the numbers 3, 2, and 1, respectively. The rule engine queries this mapping table based on the material's identity identifier extracted by the signal parsing unit to determine the preset importance level of the material.

[0126] The rule engine calculates the comprehensive priority score for each warning signal based on preset fusion scoring rules. This rule uses a weighted scoring formula: Comprehensive Priority Score = ;in, , , , The weighting coefficients are respectively for the warning type, risk level, urgency level, and importance of materials, for example, set to 0.2, 0.3, 0.3, and 0.2 respectively. To assign scores to different warning types, “abnormal accounts and physical inventory” is mapped to 90 points, “shortage risk” to 80 points, and “stagnant inventory risk” to 70 points. To assign risk level scores, “high,” “medium,” and “low” risk levels are mapped to 100, 70, and 40 points, respectively. For numerical levels, the level is multiplied by a fixed score, for example, level 3 is 100 points, level 2 is 70 points, and level 1 is 40 points. To score the urgency level, trigger conditions of type "single over-urgency threshold" or "predictive" in the signal are marked as high urgency and mapped to 100 points; "continuous over-threshold" is mapped to 70 points. To assign importance scores to materials, "critical," "important," and "general" are mapped to 100, 70, and 40 points respectively. The rule engine calculates a comprehensive priority score between 0 and 100 based on the extracted information from each signal.

[0127] The task scheduling unit performs signal merging. It groups all warning signals according to material identification. For multiple warning signals pointing to the same material identification, the task scheduling unit compares their overall priority scores and selects the highest score as the final task priority for that material identification.

[0128] The task scheduling unit sorts all material identifications in descending order of their final task priority. The task scheduling unit generates a unified inventory task instruction. This instruction is a structured data object containing a list of material identifications to be inventoried. Each record in the list includes: material identification, corresponding final task priority, and the type of task to be executed. The task type is determined by the type of warning that triggers the task: if triggered by a first-type warning signal, the task type is "emergency inventory"; if triggered by a second-type warning signal ("shortage risk"), the task type is "preventative inventory (shortage)"; if triggered by a "stagnation risk," the task type is "preventative inventory (stagnation)." The generated unified inventory task instruction is published to the data bus for downstream instruction execution and feedback modules to subscribe to and use.

[0129] The instruction execution and feedback module is used to respond to inventory task instructions, perform on-site inventory of the target power materials specified in the instructions, and conduct differential analysis based on the inventory results and historical business data to complete inventory account adjustments.

[0130] Furthermore, the process of performing on-site inventory checks and discrepancy analysis by the instruction execution and feedback module includes the following steps:

[0131] Parse the unified inventory task instructions to obtain the list of target power materials to be inventoried and their corresponding task priorities;

[0132] Based on the identification of the target power materials, query their book storage location information from the warehouse management system, and generate a field inventory operation sequence based on task priority;

[0133] Based on the on-site inventory operation sequence, the target power materials are scanned, identified, and counted using mobile inventory equipment to obtain on-site inventory quantity information;

[0134] Retrieve historical business data of the target power materials within a preset traceability period;

[0135] Based on the physical inventory count information, perform forward and reverse tracing analysis on historical business data based on the law of inventory conservation to locate one or more abnormal business records that cause discrepancies between the accounts and the actual inventory.

[0136] Based on the type and reason of the abnormal business records located, generate and execute the corresponding inventory account adjustment instructions to complete the inventory account adjustment.

[0137] Furthermore, the steps for locating one or more abnormal business records that cause the discrepancy between the accounting records and the actual transactions specifically include:

[0138] From historical business data, determine the initial inventory quantity, the quantity of all inbound records, the quantity of all outbound records, and the quantity of all inventory adjustment records within the preset traceability period;

[0139] Based on the law of conservation of inventory, the theoretical inventory quantity at the end of the preset traceability period is calculated using the beginning inventory quantity, the quantity of all inbound records, the quantity of all outbound records, and the quantity of all inventory adjustment records.

[0140] The calculated theoretical inventory quantity is compared with the actual inventory quantity to obtain the discrepancy between the book and physical inventory.

[0141] If the discrepancy between the accounting records and the actual transaction is not zero, it is determined that there is at least one abnormal business record in the historical business data.

[0142] Using discrepancies between accounts and actual inventory as clues, we reverse-traverse and logically verify each business record in historical business data to locate one or more abnormal business records with errors in quantity, material identification, or business time.

[0143] Specifically, the instruction execution and feedback module subscribes to unified inventory task instructions from the data bus. This module includes a task parsing unit, a job planning unit, a mobile inventory interface, a data traceability unit, and an accounting adjustment unit. The task parsing unit parses the unified inventory task instructions to obtain the list of identification identifiers of the target power materials to be inventoried and the task priority corresponding to each identification identifier in the list.

[0144] The operation planning unit, based on the target power materials' identification identifiers, queries the warehouse management system via an application programming interface to obtain the system's book storage location information for each material's identification identifier, namely the theoretical warehouse number, zone number, row number, layer number, and location number. Based on the task priority of the materials and their physical location in the warehouse, the operation planning unit uses a shortest path optimization algorithm to generate a physical inventory operation sequence. This sequence is an ordered list of storage locations, indicating the movement routes that inventory personnel should follow and the identification identifiers of the materials to be inventoried at each location.

[0145] The mobile inventory interface is responsible for interacting with the mobile inventory devices. These devices are industrial PDAs or smartphones with a customized inventory application installed. Following the on-site inventory sequence, after arriving at the designated storage location, inventory personnel use the mobile inventory device to scan the barcode or RFID tag of the materials to identify their identity, and manually count the actual quantity of the materials at that location, entering the data into the device. The mobile inventory device binds the scanned and identified material identity with the entered count quantity, forming an on-site inventory record, which is uploaded to the instruction execution and feedback module in real-time or in batches. This module aggregates all records to form on-site inventory quantity information for each target power material identity. For a single material distributed across multiple storage locations, the count quantities from all locations are aggregated to obtain the total on-site inventory quantity for that material.

[0146] The data traceability unit retrieves historical business data. The data traceability unit sets a preset traceability period, for example, 30 days prior to the current inventory count date. The data traceability unit extracts all historical business data for the target power materials within this 30-day period from the historical database of the enterprise resource planning system or warehouse management system. Historical business data includes: the initial inventory quantity at midnight on the start date of the period, all inbound records and their quantities within the period, all outbound records and their quantities within the period, and all inventory adjustment records and their quantities within the period. Inbound records, outbound records, and adjustment records all include document number, material identification, quantity, and transaction time.

[0147] The data traceability unit performs traceability analysis based on the law of conservation of inventory. The law of conservation of inventory is stated as: Beginning Inventory + Total Inbound Inventory - Total Outbound Inventory ± Total Adjustments = Ending Inventory. The data traceability unit first performs forward calculations: using the extracted beginning inventory quantity, summing the quantities of all inbound records to obtain the total inbound inventory, summing the quantities of all outbound records to obtain the total outbound inventory, and summing the quantities of all adjustment records to obtain the total adjustment, these are then substituted into the right side of the formula to calculate the theoretical inventory quantity at the end of the preset traceability period, denoted as... ;

[0148] The data traceability unit then calculates the theoretical inventory quantity. The quantity information obtained from the actual inventory count is recorded as follows: Compare and calculate the difference between the book value and the actual value. :

[0149] ;

[0150] If the difference between the accounting records and the actual inventory is zero, the records are considered to match, and the process ends. If the difference between the accounting records and the actual inventory is not zero, it is determined that there is at least one abnormal business record in the historical business data.

[0151] The data traceability unit uses the difference between the recorded and actual inventory values ​​(Difference) as a clue for reverse traversal and logical verification. The unit iterates backward through each historical business record within a preset traceability period, from most recent to oldest business time. For each record, the system simulates removing or restoring its quantity from the current theoretical calculation, observing the change in the remaining difference. Specific verification logic includes: checking if the material identification of the record matches the target material; checking if there are any input errors in the quantity of the record; and checking if an error in the business time of the record caused it to be included in an incorrect period. Through iterative comparison, one or more abnormal business records with errors in quantity, material identification, or business time are identified. For example, if an outbound record is found to have a quantity of 10 but should actually have 1, and correcting this error would bring the Difference close to zero, then this record is identified as an abnormal business record.

[0152] The accounting adjustment unit generates corresponding inventory adjustment instructions based on the type and cause of the abnormal business records located by the data traceability unit. For example, if the abnormality is a missing entry on an inbound order, an instruction to supplement the inbound entry is generated; if the abnormality is an error in the quantity entered for an outbound order, an instruction to reverse the incorrect record and enter the correct record is generated. The accounting adjustment unit calls relevant services of the warehouse management system or enterprise resource planning system through the application programming interface to execute the generated inventory adjustment instructions and complete the inventory adjustment within the system. After the adjustment is completed, the book inventory of the material is consistent with the physical inventory count.

[0153] The feedback data management module is used to link and store the results of the discrepancy tracing analysis with inventory account adjustment information to form optimization feedback data.

[0154] Furthermore, the feedback data management module performs the process of associated storage and forming optimized feedback data, which includes the following steps:

[0155] Obtain the record attribute information of abnormal business records identified by the difference tracing analysis, as well as the adjustment content information of the inventory account adjustment instructions executed based on the abnormal business records;

[0156] Associate and bind the record attribute information of abnormal business records with the adjustment content information of the corresponding inventory account adjustment instructions;

[0157] Associate the abnormal business records and adjustment information after they are linked with the original early warning signal information that triggered this difference source analysis;

[0158] The complete data records after association and binding are stored in the optimization feedback database according to the preset data structure to form optimization feedback data.

[0159] Specifically, the feedback data management module receives input data from the instruction execution and feedback module. This input data includes record attribute information of abnormal business records located by the discrepancy tracing analysis and adjustment content information of inventory account adjustment instructions executed based on the abnormal business records. Simultaneously, this module obtains the original early warning signal information that triggered this discrepancy tracing analysis from the data bus; this information is recorded and transmitted by the instruction execution and feedback module when the tracing analysis is initiated.

[0160] The feedback data management module includes a data association unit. This unit first processes record attribute information and adjustment content information. Record attribute information includes at least: the unique document number of the abnormal business record, the business type of the record, the material identification in the record, the incorrect quantity value in the record, the business time of the record, and the reason for the error. Adjustment content information includes at least: the unique number of the corresponding adjustment instruction, the adjustment type, the correct quantity value to be adjusted, and the execution time of the adjustment. The data association unit associates and binds the record attribute information of an abnormal business record with the adjustment content information of the inventory account adjustment instruction it directly triggers, generating a "problem-solution" association record.

[0161] The data association unit then associates the generated "problem-solution" association record with the source of this verification, namely the original warning signal information. The original warning signal information includes at least: warning signal ID, signal type, signal generation time, and signal source module. The data association unit associates these three parts of information to form a complete source tracing feedback record.

[0162] The feedback data management module stores complete traceability feedback records in the optimization feedback database according to a preset data structure. The optimization feedback database is a relational database. The preset data structure is defined as a single data table containing the following fields: Record ID: unique primary key; associated original warning signal ID; warning signal type: record is "Category 1" or "Category 2" and subtype; warning generation time; target material identification; abnormal business record document number; abnormal business type; error quantity value; error reason; adjustment instruction number; adjustment type; correct adjustment quantity value; adjustment execution time; record entry time;

[0163] The feedback data management module maps each complete source tracing feedback record to a row in the aforementioned data table and performs an insertion operation to store it in the optimization feedback database. Thus, the structured data continuously accumulated in the optimization feedback database constitutes the optimization feedback data. This data clearly records the entire chain of information from early warning triggering to problem localization and resolution, with each part interconnected through key fields, providing traceable and statistically verifiable training and evaluation samples for subsequent model optimization.

[0164] The model optimization module is used to optimize and adjust the parameters of the preset difference threshold in the dynamic verification and early warning module and the prediction model in the predictive analysis and early warning module based on the optimization feedback data.

[0165] Furthermore, the process of parameter optimization and adjustment performed by the model optimization module includes the following steps:

[0166] Periodically extract optimization feedback data within a preset time range from the optimization feedback database;

[0167] Based on the optimized feedback data, the distribution of verification results for the first type of early warning signal and the distribution of verification results for the second type of early warning signal were statistically analyzed respectively.

[0168] Based on the distribution of verification results of the first type of early warning signal, the preset difference threshold in the dynamic verification and early warning module is optimized and adjusted;

[0169] Based on the distribution of verification results for the second type of early warning signal, the prediction model in the prediction analysis and early warning module is optimized and adjusted.

[0170] Specifically, the model optimization module runs optimization tasks periodically, for example, once every day at midnight. This module includes a data extraction unit, a performance statistics unit, a threshold optimization unit, and a model retraining unit. The data extraction unit retrieves optimization feedback data from the optimization feedback database within a preset time range. This preset time range is, for example, the most recent 90 days. The extracted data includes at least: warning signal type, warning signal ID, warning generation time, target material identification, abnormal business record document number, and error reason.

[0171] Based on the extracted optimization feedback data, the performance statistics unit separately calculates the distribution of verification results for the first and second types of early warning signals. The statistical method is as follows: for each early warning signal record, it is judged based on its associated abnormal business records and adjustment information. If the early warning signal successfully locates and leads to a valid inventory adjustment, the early warning is marked as "accurate"; if the early warning signal is not associated with any abnormal business record, i.e., the inventory records match the physical inventory after verification, the early warning is marked as "false alarm"; if, during the period or for the materials covered by the early warning signal, a discrepancy between the physical inventory and the physical inventory is subsequently discovered through other means but was not covered by this early warning, the early warning is marked as "missed" based on cross-validation of other records in the optimization feedback data. The performance statistics unit then summarizes the respective quantities of "accurate," "false alarm," and "missed" early warning signals for both the first and second types.

[0172] The threshold optimization unit optimizes and adjusts the preset difference thresholds in the dynamic verification and early warning module based on the verification result distribution of the first type of early warning signal. The threshold optimization unit calculates the current early warning accuracy rate. and false alarm rate accuracy =Number of accurate alarms / Number of accurate alarms plus number of false alarms; False alarm rate =Number of false alarms / Number of accurate predictions plus number of false alarms. The threshold optimization unit sets the optimization objective as: minimizing the false alarm rate while ensuring the accuracy is not lower than the target value A, for example, 0.85. The optimization algorithm uses a heuristic search: based on the current preset difference threshold, the threshold is finely adjusted upwards and downwards in a fixed step size, for example, 0.01, using historical data to simulate and calculate the accuracy and false alarm rate under different thresholds. The threshold that meets the accuracy requirement and has the lowest false alarm rate is selected as the new preset difference threshold. If neither upward nor downward adjustment can obtain a better solution, the original threshold is maintained.

[0173] The model retraining unit optimizes and adjusts the prediction model in the prediction analysis and early warning module based on the distribution of validation results for the second type of early warning signal. The model retraining unit first evaluates model performance. For shortage risk early warning, its accuracy is calculated. For stagnation risk early warning, its accuracy is also calculated. The model retraining unit sets a performance threshold; for example, if the accuracy is below 0.7, model retraining is triggered.

[0174] If retraining is triggered, the model retraining unit obtains the latest historical inventory and consumption data with a longer time span from the data acquisition module, such as data from the past 365 days. It uses the same Long Short-Term Memory (LSTM) network structure as the original training model, containing an input layer, two LSTM hidden layers, and a fully connected output layer. The model retraining unit constructs a training sample set from the latest historical data according to the same rules and retrains the model parameters using the mean squared error loss function and the Adam optimizer. After training, a validation set is used to evaluate the prediction accuracy of the new model. If the new model outperforms the old model on the validation set, the new model replaces the prediction model currently being used in the prediction analysis and early warning module.

[0175] The model optimization module will configure the optimized and adjusted preset difference threshold and / or updated prediction model parameters to the dynamic verification and early warning module and the prediction analysis and early warning module, respectively, to complete the closed-loop optimization.

[0176] Example 2:

[0177] While existing power material warehouse management systems utilize information systems and IoT technology, the core process of verifying inventory records still relies on manual, periodic, or random static inventory checks. This approach cannot detect and warn of discrepancies between records and actual inventory in real time when material inflows and outflows occur, leading to significant delays in problem detection. Furthermore, limited inventory resources cannot be intelligently focused on high-risk materials, creating a fundamental contradiction between "static, lagging inventory checks" and "dynamic, real-time operations," thus hindering the development of warehouse management towards a proactive, precise, and efficient intelligent stage. To address these issues, this invention provides a warehouse inventory identification method for power materials, the structure of which is as follows: Figure 2 As shown. The specific implementation process of this method is as follows:

[0178] Collect the data stream of power materials issued from the books and the data stream of physical movement obtained through IoT sensing devices, and obtain historical inventory and consumption data;

[0179] A time-series comparison analysis is performed on the outbound data stream and the physical movement data stream. When the analysis results exceed the preset difference threshold, a first-type early warning signal is generated.

[0180] Based on historical inventory and consumption data, a second type of early warning signal is generated through predictive model analysis to target the risk of inventory shortage or obsolescence.

[0181] Receive the first type of early warning signal and the second type of early warning signal, merge them and prioritize them according to preset rules, and generate a unified inventory task instruction;

[0182] In response to the inventory task instruction, perform on-site inventory of the target power materials specified in the instruction, and conduct a source-tracing analysis of the discrepancies based on the inventory results and historical business data to complete the inventory account adjustment;

[0183] The results of the discrepancy tracing analysis and inventory account adjustment information are linked and stored to form optimization feedback data;

[0184] Based on the optimized feedback data, the parameters of the preset difference threshold in the dynamic verification and early warning module and the prediction model in the predictive analysis and early warning module are optimized and adjusted.

[0185] Specifically, firstly, the system connects to the warehouse management system and the IoT sensing device network to collect both the book-based outbound data stream and the physical movement data stream of power materials, and obtains historical inventory and consumption data from the enterprise resource planning system. Secondly, the system performs time-series alignment and window cumulative calculation on the streaming data. When the discrepancy between the book-based and physical flow rates consistently exceeds 5% for 2 hours or exceeds 15% in a single instance, an anomaly warning signal is generated. Simultaneously, a Long Short-Term Memory (LSTM) network model is used to train and predict historical data, generating a shortage and obsolescence risk warning signal for the next 7 days. Then, based on the material importance, warning type, and risk... The system uses a weighted scoring algorithm to fuse and prioritize two types of early warning signals based on their level and urgency, generating task instructions that include a list of materials to be inventoried and their priorities. Then, in response to these instructions, an on-site inventory check is conducted, and based on the law of conservation of inventory, the business flow within the inventory period is calculated forward and traced backward to locate and correct any abnormal business records causing discrepancies. Subsequently, the early warnings, abnormal records, and adjustment instructions are stored together to form structured feedback data. Finally, based on the early warning verification results in the feedback data, the discrepancy threshold is dynamically adjusted and the prediction model is retrained periodically to complete the system's closed-loop optimization.

[0186] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A warehouse inventory and identification system for electrical materials, characterized in that, Includes the following modules: The data acquisition module is used to collect the book data stream of power materials leaving the warehouse and the physical movement data stream obtained through IoT sensing devices, and to acquire historical inventory and consumption data. The dynamic verification and early warning module is used to perform time-series comparison and analysis of the book outbound data stream and the physical movement data stream. When the analysis result exceeds the preset difference threshold, a first type of early warning signal is generated. The predictive analysis and early warning module is used to generate a second type of early warning signal for inventory shortage risk or obsolescence risk based on the historical inventory and consumption data and through predictive model analysis. The task fusion and scheduling module is used to receive the first type of early warning signal and the second type of early warning signal, fuse them and prioritize them according to preset rules, and generate a unified inventory task instruction. The instruction execution and feedback module is used to respond to the inventory task instruction, perform on-site inventory of the target power materials specified in the instruction, and perform source analysis of discrepancies based on the inventory results and historical business data to complete the inventory account adjustment. The feedback data management module is used to associate and store the results of the difference tracing analysis and the inventory account adjustment information to form optimization feedback data. The model optimization module is used to optimize and adjust the parameters of the preset difference threshold in the dynamic verification and early warning module and the prediction model in the prediction analysis and early warning module based on the optimization feedback data.

2. The warehouse inventory and identification system for electrical materials according to claim 1, characterized in that, The data acquisition module performs the data acquisition process including the following steps: The system connects to the warehouse management system and acquires the book-based outbound records of power materials according to a preset cycle or in real time, forming the book-based outbound data stream. Access the network of IoT sensing devices deployed at warehouse entrances and exits, shelves and handling equipment, obtain the identification of power materials and movement events scanned or sensed by the IoT sensing devices, and form the physical movement data stream; Access the historical database of the enterprise resource planning system or warehouse management system, and extract historical inbound records, outbound records and inventory balance records according to material category and time dimension to form the historical inventory and consumption data. The book-based outbound data stream, the physical movement data stream, and the historical inventory and consumption data are standardized, cleaned, and synchronized with time, and the processed data is published to the data bus.

3. A warehouse inventory and identification system for electrical materials according to claim 1, characterized in that, The dynamic verification and early warning module performs time-series comparison analysis and early warning processes, including the following steps: Based on the same material identification and unified time benchmark, the book outbound data stream and the physical movement data stream are paired and aligned to generate synchronous time-series data pairs at the material level. Based on a preset time window, for each of the synchronized time-series data pairs, calculate its cumulative book outbound volume and cumulative physical movement volume, and calculate the difference rate or absolute difference value between the two. Compare the difference rate or absolute difference value with the preset difference threshold; When the difference rate or absolute difference value continuously exceeds the preset difference threshold for a preset duration, or exceeds the preset emergency threshold in a single instance, it is determined that the account balance trend is abnormal, and the first type of early warning signal is generated.

4. A warehouse inventory and identification system for electrical materials according to claim 1, characterized in that, The predictive analysis and early warning module performs predictive analysis and early warning in the following steps: The historical inventory and consumption data are preprocessed, and a data sample set containing historical consumption sequences, inventory balance sequences and related features is constructed using material identification as the dimension. The data sample set is input into a pre-trained prediction model, and the prediction model outputs a predicted value of material consumption for a future preset period and a predicted inventory balance at the end of the future preset period. Based on the predicted consumption of materials, the predicted inventory balance, and the preset safety stock threshold, calculate the inventory shortage risk level within the preset period in the future. Calculate the degree of deviation between the historical consumption sequence of materials and the predicted consumption value of materials. Based on the degree of deviation and in combination with the inventory age, calculate the material stagnation risk level. When the inventory shortage risk level or the material stagnation risk level exceeds its respective preset risk threshold, the second type of early warning signal is generated.

5. A warehouse inventory and identification system for electrical materials according to claim 1, characterized in that, The task fusion and scheduling module performs the task fusion and scheduling process including the following steps: Receive and parse the first type of early warning signal and the second type of early warning signal, and extract the early warning type, material identification, risk level and urgency information from the signals; Based on the preset material importance mapping table, determine the preset importance level of the material corresponding to the material identity identifier; Based on the warning type, the risk level, the urgency information, and the preset importance level, the comprehensive priority score corresponding to each warning signal is calculated according to the preset fusion scoring rules; The warning signals pointing to the same material identification are merged, and the highest comprehensive priority score is used as the final task priority of the material. Based on the final task priority of all materials, a unified inventory task instruction is generated, which includes a list of material identification identifiers to be inventoried, corresponding task types, and priorities.

6. A warehouse inventory and identification system for electrical materials according to claim 1, characterized in that, The process of the instruction execution and feedback module performing on-site inventory and discrepancy tracing analysis includes the following steps: Parse the unified inventory task instructions to obtain the list of target power materials to be inventoried and their corresponding task priorities; Based on the identification of the target power materials, query their book storage location information from the warehouse management system, and generate an on-site inventory operation sequence based on the task priority; Based on the aforementioned on-site inventory operation sequence, the target power materials are scanned, identified, and counted using mobile inventory equipment to obtain on-site inventory quantity information; Retrieve historical business data of the target power materials within a preset traceability period; Based on the physical inventory count information, perform forward and reverse tracing analysis on the historical business data based on the law of inventory conservation to locate one or more abnormal business records that cause discrepancies between the accounts and the physical inventory. Based on the type and cause of the abnormal business record located, generate and execute the corresponding inventory account adjustment instruction to complete the inventory account adjustment.

7. A warehouse inventory and identification system for electrical materials according to claim 6, characterized in that, The steps for locating one or more abnormal business records that cause discrepancies between accounting records and actual transactions specifically include: From the historical business data, determine the initial inventory quantity, the quantity of all inbound records, the quantity of all outbound records, and the quantity of all inventory adjustment records within the preset traceability period; Based on the law of conservation of inventory, the theoretical inventory quantity at the end of the preset traceability period is calculated using the initial inventory quantity, the quantity of all inbound records, the quantity of all outbound records, and the quantity of all inventory adjustment records. The calculated theoretical inventory quantity is compared with the actual inventory quantity information to obtain the discrepancy between the book and physical inventory values. If the discrepancy between the account and the actual transaction is not zero, it is determined that there is at least one abnormal business record in the historical business data. Using the discrepancy between the accounts and actual inventory as a clue, each business record in the historical business data is traversed in reverse and logically verified to locate one or more abnormal business records that have errors in quantity, material identification, or business time.

8. A warehouse inventory and identification system for electrical materials according to claim 1, characterized in that, The process by which the feedback data management module performs associated storage and generates optimized feedback data includes the following steps: Obtain the record attribute information of the abnormal business records located by the difference tracing analysis, as well as the adjustment content information of the inventory account adjustment instructions executed based on the abnormal business records; The record attribute information of the abnormal business record is associated and bound with the adjustment content information of the corresponding inventory account adjustment instruction; Associate the abnormal business records and adjustment information after they are linked with the original early warning signal information that triggered this difference source analysis; The complete data records after the association and binding are stored in the optimization feedback database according to the preset data structure to form the optimization feedback data.

9. A warehouse inventory and identification system for electrical materials according to claim 8, characterized in that, The process of parameter optimization and adjustment performed by the model optimization module includes the following steps: From the optimization feedback database, optimization feedback data within a preset time range is periodically extracted; Based on the optimized feedback data, the verification result distribution of the first type of early warning signal and the verification result distribution of the second type of early warning signal are statistically analyzed respectively. Based on the distribution of verification results of the first type of early warning signal, the preset difference threshold in the dynamic verification and early warning module is optimized and adjusted; Based on the distribution of verification results for the second type of early warning signal, the prediction model in the prediction analysis and early warning module is optimized and adjusted.

10. A warehouse inventory identification method for electrical materials, characterized in that, A warehouse inventory identification system for electrical materials according to any one of claims 1-9, the method comprising: Collect the data stream of power materials issued from the books and the data stream of physical movement obtained through IoT sensing devices, and obtain historical inventory and consumption data; A time-series comparison analysis is performed on the book outbound data stream and the physical movement data stream. When the analysis result exceeds a preset difference threshold, a first type of early warning signal is generated. Based on the historical inventory and consumption data, a second type of early warning signal is generated through predictive model analysis to target the risk of inventory shortage or obsolescence. Receive the first type of early warning signal and the second type of early warning signal, fuse them and prioritize them according to preset rules, and generate a unified inventory task instruction; In response to the inventory task instruction, an on-site inventory is performed on the target power materials specified in the instruction, and a difference tracing analysis is conducted based on the inventory results and historical business data to complete the inventory account adjustment; The results of the discrepancy tracing analysis and the inventory account adjustment information are linked and stored to form optimization feedback data; Based on the optimized feedback data, the preset difference threshold in the dynamic verification and early warning module and the prediction model in the prediction analysis and early warning module are optimized and adjusted.

Images