An artificial intelligence-based unmanned warehouse management system

Abstract

The present application relates to unmanned warehouse management technical field, specifically to a kind of unmanned warehouse management system based on artificial intelligence, system includes obtaining material number, category name and batch mark and write into radio frequency tag, collect radio frequency position and passage information to generate track content, match track direction to call up storage location number and linkage indicating device, check personnel number and radio frequency content mark abnormal passage, associated abnormal position and instruction number to generate control instruction.The present application writes into radio frequency tag by splicing material number, category information and batch mark, and binds storage location, realizes identification structure unification.Collect radio frequency position and passage node, build material moving path.According to path direction, match target storage location and linkage indicating device to generate guide record.In passage, compare personnel number and radio frequency content, mark abnormal and combine operation instruction check, point to unmanned processing area to complete abnormal disposal.

Description

Technical Field

[0001] This invention relates to the field of unmanned warehouse management technology, and in particular to an unmanned warehouse management system based on artificial intelligence. Background Technology

[0002] The field of unmanned warehouse management technology involves the automated inbound, storage, picking, and outbound processes of goods. Core aspects include information identification, route planning, material handling execution, inventory management, and environmental monitoring. Through the acquisition and processing of multi-source data, it enables the automated scheduling and execution of work processes, improving the collaborative efficiency of warehouse space and task processing.

[0003] Traditional unmanned warehouse management systems utilize rules to control transport equipment for handling and sorting, identify goods by reading barcodes using fixed cameras, maintain inventory data using a database, record entry and exit information using RFID cards, and issue path and task instructions from a central dispatch system. During execution, transport equipment performs obstacle avoidance and navigation based on local sensor data, and the overall process relies on preset logic and static data configuration.

[0004] Traditional identification methods rely on fixed cameras to read barcodes, which have limited information capacity, are severely affected by viewing angle and occlusion, and lack dynamic tracking capabilities. Route instructions are statically issued by a central system, making it difficult to reflect real-time operational status and posing risks of route conflicts and scheduling delays. Inventory information is maintained using a static database, lacking support from transaction flow, resulting in poor timeliness. The lack of a one-to-one correspondence between materials and personnel behavior leads to ambiguity in responsibility, difficulty in identifying abnormal passage, and increased operational risks. The lack of a linkage mechanism between behavioral paths and actual storage location status makes it difficult to handle abnormal operating conditions in a closed-loop manner. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing an unmanned warehouse management system based on artificial intelligence.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an unmanned warehouse management system based on artificial intelligence, the system comprising:

[0007] The material identification binding module obtains the material number, material category name, and arrival batch mark. It concatenates the material number and category name fields and writes them together with the batch mark into the RFID tag content, and associates them with the storage location to ensure that the tag content is consistent with the physical item, thus obtaining the unique identification content of the material.

[0008] The warehouse access identification module calls the unique identification content of the material, collects the radio frequency reading position and entry / exit start and end information when the material passes through the warehouse passage, associates the reading position and identification content with the passage flow process, and continuously follows the material movement to complete the path association description, thereby obtaining the material warehouse passage trajectory content.

[0009] The location guidance execution module retrieves the specified location number in the operation task based on the flow direction in the passage trajectory content, maps it to the address of the in-warehouse indicator device, enters the guidance state after receiving the trigger command, and continues until the flow is completed, thus obtaining the location guidance execution record.

[0010] The entrance and exit behavior verification module calls the cargo location number in the guidance execution record, collects the personnel identification number and radio frequency reading content in the passage process, processes them one by one, and classifies the passage process that does not form a corresponding relationship as abnormal to obtain abnormal passage mark content.

[0011] The unmanned warehouse management module retrieves the corresponding material identification content and operation instruction number based on the access location associated with the abnormal access marker content, checks the relationship between the two item by item, includes the unassociated storage locations into the unmanned processing scope and issues control actions, and generates unmanned warehouse management execution instructions.

[0012] As a further aspect of the present invention, the unique identification content of the material includes the radio frequency tag encoding structure, the material category splicing field, and the storage location association information; the material warehouse passage trajectory content includes the path location sequence, passage time node, and material identification mapping; the storage location guidance execution record includes the target storage location number, the indicating device address, and the guidance status identifier; the abnormal passage marker content includes the unmatched personnel number, the abnormal identification tag, and the passage abnormality type; and the unmanned warehouse management execution instruction includes the pending storage location number, the abnormal operation instruction, and the automatic control strategy.

[0013] As a further aspect of the present invention, the material identification binding module includes:

[0014] The information splicing submodule obtains the material number, material category name, and arrival batch mark registered in the warehousing operation, performs character splicing operation according to the field order of material number and category name, connects the result with the batch mark, identifies whether the character position is repeated or offset according to the set order rules, and adjusts the character arrangement order to obtain the spliced ​​field content.

[0015] The tag writing submodule calls the content of the splicing field, allocates the character content to the corresponding area of ​​the RFID tag according to the tag structure, performs character division and content writing operations in the order of segments, detects the offset between the content of each segment and the set range, rewrites after adjusting the order, collects the return status of each segment, and obtains the tag writing feedback set.

[0016] The location association submodule calls the tag writing feedback set, determines whether to perform the information binding action based on the returned status, obtains the location information in the operation, performs a pairing operation with the content in the RFID tag according to the field order, writes the corresponding relationship into the location data area, and obtains the unique identification content of the material.

[0017] As a further aspect of the present invention, the warehouse access identification module includes:

[0018] The identification information calling submodule calls the unique identification content of the material. When the material enters the warehouse channel entrance, the identification content field is read synchronously. The time sequence is compared according to the data collected at the channel entrance time period. It checks whether the identification field matches the channel passage record within the preset range. The fields are sorted and classified according to the time correspondence to obtain the identification matching mark set.

[0019] The location information acquisition submodule calls the identification and matching mark set to collect radio frequency reading location data and entry / exit start and end information of materials during the passage process. It combines and filters according to reading time and position order, performs channel direction screening on the collected items, removes duplicate records and dwell segments, and generates a set of location trajectory segments according to reading frequency.

[0020] The path association processing submodule calls the set of location trajectory segments, performs path sorting processing on each segment and the identified content according to the time order, calculates the time distance and jump order offset value between consecutive positions in the path, classifies and merges the jump situation between segments according to material identification, combines and describes the channel trajectory corresponding to each identification, and obtains the passage trajectory content in the material warehouse.

[0021] As a further aspect of the present invention, the open cargo location guidance execution module includes:

[0022] The direction extraction submodule calls the passage trajectory content in the material warehouse, identifies the continuous passage segments and directional order in the trajectory, extracts the flow number information under the corresponding direction, matches and filters it with the content of the task to be processed in the work task, performs number classification operation on the matching items, and obtains the task location number set.

[0023] The address mapping submodule calls the task location number set, extracts the address field information by referring to the instruction equipment binding table according to the number order, performs field mapping processing between the number content and the equipment address, identifies and removes duplicate numbers and missing address items during the retrieval process, and sorts the valid matching results into a sequence to obtain the location address mapping list.

[0024] The instruction triggering submodule calls the cargo location address mapping list, sequentially sends guidance instructions to the corresponding address devices, monitors the device receiving status and records the receiving feedback content, determines whether the device has entered the indication state based on the indication status code in the feedback, controls the duration of the state until the path passage is completed, and obtains the cargo location guidance execution record.

[0025] As a further aspect of the present invention, the entrance / exit behavior core module includes:

[0026] The storage location number extraction submodule calls the storage location guidance execution record, extracts the storage location numbers involved in each record, segments and organizes the numbers according to the time sequence, compares the organized numbers with the operation timeline in sequence, removes the number content that does not fall into the valid time segment, and obtains the set of passable storage location numbers.

[0027] The identification information collection submodule calls the set of access location numbers, collects personnel identification numbers and radio frequency readings in the passage of the corresponding location number during the passage period, performs joint matching of readings and personnel numbers in chronological order, removes duplicate personnel records and blank readings, and obtains a set of access identification information.

[0028] The access verification submodule calls the access identification information set, matches the personnel number with the radio frequency content according to the reading time, extracts the numbers and contents that cannot be matched and classifies them into abnormal access processes, groups them according to the cargo location number, outputs the number of unmatched reading identifiers in each group, and obtains the abnormal access marker content.

[0029] As a further aspect of the present invention, the unmanned warehouse management module includes:

[0030] The location extraction submodule calls the access location information in the abnormal access marker content, extracts the material identification content and operation instruction number corresponding to the location, merges the identification content and instruction number in chronological order, marks and filters out the location number that is missing any content, and obtains a set of unmatched location numbers.

[0031] The instruction verification submodule calls the set of unmatched storage location numbers, compares whether there is a corresponding item between the valid identification content and the operation number in the current task for each number, checks whether the number mapping result forms a one-to-one record, and summarizes the content that cannot be verified to obtain a list of unattended storage locations.

[0032] The control action issuance submodule calls the unmanned processing location list, matches it to the corresponding device address in the control device distribution table according to the number order, sends an entry control status command to the device address, records the device's received feedback content and continuous response action status, and obtains the unmanned warehouse management execution command.

[0033] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0034] In this invention, material identification is achieved by concatenating material numbers and category information with batch markers and writing them into RFID tags, giving material identifiers a unified structure and unique attributes that are consistent with the actual storage location. By continuously collecting RFID reading locations and access node information during material flow, a continuous in-warehouse movement path is constructed to reflect the flow relationship of materials between different channels. By matching the path direction with the target storage location in the work task and linking it with in-warehouse indicating devices, a guidance record synchronized with the flow process is formed. By introducing a correspondence between personnel identification numbers and material RFID content at the access points, access behaviors that cannot form a mapping are marked, and these are checked against abnormal locations and work instructions to form control directions for unattended areas, supporting centralized handling of abnormal states within the warehouse. Attached Figure Description

[0035] Figure 1 This is a flowchart of the method of the present invention;

[0036] Figure 2 This is a flowchart illustrating the acquisition process of the material identification binding module of the present invention.

[0037] Figure 3 This is a flowchart illustrating the acquisition process of the warehouse access identification module of the present invention.

[0038] Figure 4 This is a flowchart illustrating the acquisition process of the cargo location guidance execution module of the present invention.

[0039] Figure 5 This is a flowchart illustrating the acquisition process of the entrance / exit behavior verification module of the present invention.

[0040] Figure 6 This is a flowchart illustrating the acquisition process of the unmanned warehouse management module of the present invention. Detailed Implementation

[0041] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0042] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0043] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0044] Please see Figure 1 This invention provides a technical solution: an unmanned warehouse management system based on artificial intelligence, the system comprising:

[0045] The material identification binding module obtains the material number, material category name, and arrival batch mark registered in the warehousing operation, concatenates the material number and category name, and then writes them together with the batch mark into the RFID tag content. After writing, the tag content is associated with the current storage location to maintain the consistency between the tag content and the physical item, and generates unique identification content for the material.

[0046] The warehouse access identification module calls the unique identification content of the material and collects the radio frequency reading position and entry / exit start and end information when the material passes through the warehouse passage. It associates the reading position and identification content with the passage flow process, continuously follows the material movement status to complete the path association description, and generates the material warehouse passage trajectory content.

[0047] The location guidance execution module retrieves the location number specified in the corresponding work task based on the flow direction reflected in the material warehouse passage trajectory. It maps the location number to the address of the indicator device in the warehouse. After receiving the trigger command, the corresponding address device enters the guidance state and maintains the indicator state until the flow is completed, generating a location guidance execution record.

[0048] The entrance and exit behavior verification module calls the location number involved in the location guidance execution record, collects personnel identification number and radio frequency reading content during the passage process, performs item-by-item matching of personnel number and radio frequency content, classifies passage processes that cannot form a matching relationship as abnormal, and generates abnormal passage mark content.

[0049] The unmanned warehouse management module retrieves the corresponding material identification information and operation instruction number based on the access location associated with the abnormal access marker content, checks the relationship between the two item by item, includes the unassociated storage locations into the unmanned processing scope and issues control actions, and generates unmanned warehouse management execution instructions.

[0050] The unique identification content of materials includes the RFID tag encoding structure, material category splicing field, and storage location association information. The material warehouse passage trajectory content includes the path location sequence, passage time node, and material identification mapping. The storage location guidance execution record includes the target storage location number, the address of the indicating device, and the guidance status indicator. The abnormal passage marking content includes the unmatched personnel number, the abnormal identification tag, and the passage abnormality type. The unmanned warehouse management execution instructions include the pending storage location number, the abnormal operation instruction, and the automatic control strategy.

[0051] Please see Figure 2 The material identification binding module includes:

[0052] The information splicing submodule obtains the material number, material category name, and arrival batch mark registered in the warehousing operation, performs character splicing operation according to the field order of material number and category name, connects the result with the batch mark, identifies whether the character position is repeated or offset according to the set order rules, and adjusts the character arrangement order to obtain the spliced ​​field content.

[0053] A highly reliable bidirectional communication connection is established with the underlying database of the Warehouse Management System (WMS) via an industrial Ethernet interface (based on the TCP / IP protocol stack). Upon receiving an inbound trigger signal, the system locks the highest-priority inbound operation document in the current operation queue, with document number IN-2024-0520-X. The system then activates the field parsing engine to accurately extract key attribute data defining the material's identity from the complex document message. This includes a fixed-length 6-byte material number M-8802, an 8-byte material category name ELEC-CAP, and a 9-byte arrival batch marker BATCH-24A. This basic data enters the encoding conversion process. The processor, according to a preset concatenation rule—material number first, category name in the middle, and batch marker last—calls the ASCII code table to convert each character into a hexadecimal encoding sequence recognizable by the underlying machine. To ensure the orderly arrangement of data in memory, the system allocates space in RAM... A virtual memory-mapped cache with a length of 32 bytes (i.e., 256 bits) is initialized with all zeros. The converted hexadecimal data stream is sequentially filled into the cache according to the predetermined address pointers. During the filling process, the built-in conflict detection algorithm monitors the occupancy of the current write address bits in real time to identify whether there is a risk of duplicate overwriting or logical offset of character positions. For example, during the execution of this embodiment, when the system completes the writing of the first field and attempts to write the starting position of the second field, it detects a data residual conflict at the predicted address. At this time, the system immediately triggers the automatic avoidance mechanism, forcibly offsets the starting write address of the second field by 16 bits (i.e., 2 bytes) and automatically fills in the original conflicting address range with hexadecimal null characters to maintain the alignment and integrity of the entire data frame. After this series of verification and adjustment operations, the independent data segments are integrated into a continuous, conflict-free hexadecimal data stream, resulting in a concatenated field content that can accurately map the physical storage bit order.

[0054] The tag writing submodule calls the concatenation field content, allocates the character content to the corresponding area of ​​the RFID tag according to the tag structure, performs character division and content writing operations in the order of the segments, detects the offset between the content of each segment and the set range, rewrites after adjusting the execution order, collects the return status of each segment, and obtains the tag writing feedback set.

[0055] The system retrieves the formatted concatenated field content and simultaneously activates an industrial-grade RFID reader installed at a key node in the warehousing pipeline. The reader's operating frequency is strictly set in the ultra-high frequency band of 920 to 925 MHz to ensure penetration and reading distance. Based on the EPC Global Gen2 V2 international standard protocol, the system logically partitions the target RFID tag's internal storage space into a 96-bit EPC area for storing a unique index and a 512-bit User area for storing detailed business data. The processor initiates the segmented writing logic, dividing the first 96 highly identifiable bits of data from the concatenated field into an index data stream and allocating it to the EPC area, while allocating the remaining descriptive data stream to the User area. Before formally executing the electromagnetic wave writing operation, the system performs a pre-test to check the data length and storage capacity matching, detecting any offset or overflow between the length of the content to be written in each segment and the set range of the tag's corresponding storage area. In this embodiment, the system calculates that the total remaining data length to be written to the User area is 600 bits, while the physical capacity limit of the User area is... With only 512 bits, a data overflow value of 88 bits was determined through numerical comparison. Faced with this physical limitation, the system automatically executes a preset order adjustment strategy, identifies non-critical category name fields in the data stream, and performs a truncation operation on them, accurately removing the 88 bits of data exceeding the capacity limit. The CRC-16 cyclic redundancy check code is recalculated and appended to the end of the data to ensure data integrity. After processing, the reader transmits a modulated radio frequency signal to modify the state of the tag's internal storage unit. Immediately after writing, a query command is sent to collect the return status words of each segment. Only when the status codes returned by the EPC area and the User area both show 0x00 (i.e., the write was successful) will the system mark the operation as passed and summarize the tag write feedback set containing the success status and verification information of each segment.

[0056] The location association submodule calls the tag writing feedback set, determines whether to perform the information binding action based on the returned status, obtains the location information in the operation, performs a pairing operation with the content in the RFID tag according to the field order, writes the corresponding relationship into the location data area, and obtains the unique identification content of the material.

[0057] The system calls the aforementioned tag to write to the feedback set. The internal logic judgment unit scans the status bits in the feedback set one by one and performs strict conditional judgments based on the returned status. Only when the flag bits of all key sectors in the set are confirmed to be in a successful state is the current tag determined to be qualified for association and the subsequent information binding action is triggered. At this time, the system synchronously calls the high-precision laser rangefinder and vision recognition module deployed above the warehouse entrance to obtain the physical space coordinates of the pallet currently carrying the material in real time, namely the storage location information RACK-05-L2-10. This coordinate precisely corresponds to the second layer of the fifth row of shelves. At position 10, the processor performs a data pairing operation, extracting the unique EPC code 30396062 that has been successfully written and verified from the RFID tag according to field order. This code is then logically associated with the acquired cargo location coordinates. The system creates a new key-value index in the association table of the core database, using the EPC code as the primary key and the cargo location coordinates as the value. To ensure data traceability, the system also appends the current timestamp information and writes this confirmed correspondence into the cargo location data area for persistent storage. This results in a unique material identification content that strongly binds the physical location to digital identity information.

[0058] Please see Figure 3 The warehouse access identification module includes:

[0059] The identification information call submodule calls the unique identification content of the material. When the material enters the warehouse channel entrance, the identification content field is read synchronously. The time sequence is compared according to the data collected at the channel entrance time period. It checks whether the identification field matches the channel passage record within the preset range. The fields are sorted and classified according to the time correspondence to obtain the identification matching mark set.

[0060] The system uses the unique material identification information as a comparison benchmark. Simultaneously, fixed RFID readers deployed at the channel entrance begin reading the identification information fields of passing tags. To eliminate environmental noise and interference from other non-target tags, the system collects data within a 1000-millisecond time window set at the channel entrance. During this period, all captured signals are buffered and compared in chronological order. The collected real-time identification fields are matched against the pre-stored daily inbound plan list in the system to check if the identification fields fall within the preset valid passage range. This process effectively filters out misread or unauthorized tag signals. Once a match is successful, the system categorizes and merges multiple consecutive signal records (including multiple reads of different RSSI values) collected within the time window into a single inbound passage event based on the timestamp correspondence. A unique internal system ID is assigned to this event, resulting in an identification matching tag set containing the event ID, original EPC code, and precise trigger time.

[0061] The location information acquisition submodule calls the identification and matching tag set to collect radio frequency reading location data and entry / exit start and end information of materials during the passage process. It combines and filters according to reading time and position order, performs channel direction screening on the collected items, removes duplicate records and dwell segments, and generates a set of location trajectory segments according to reading frequency.

[0062] The system invokes a matching tag set to initiate full-process tracking of the material's movement within the channel. A linear antenna array (Ant-01 to Ant-05) with a strictly 2.0-meter physical spacing is deployed along the channel. The system continuously collects radio frequency (RF) position data fed back by each antenna as the material passes through this area, as well as start and end information for entering and leaving the sensing zone. The processor combines and filters the massive amount of raw data according to the order of reading time and the arrangement of the antenna physical positions. It also uses the increasing or decreasing trend of the antenna numbers to perform channel-direction filtering of the collected items. In this embodiment... The system detected that the reading sequence was from antenna 01 to antenna 04, and based on this, it determined that the material movement direction was forward entry into the warehouse. In order to improve the purity of the trajectory data, the system set a single-point dwell threshold of 3.0 seconds. Redundant data was eliminated by calculating and comparing the time difference between adjacent reading records. If the continuous reading time difference at a certain antenna exceeds the threshold, it is considered a dwell segment and is eliminated. Conversely, if the time difference is within the threshold range, it is considered a continuous passage. The system only retains key data points that can characterize the continuous movement state and generates a set of location trajectory segments consisting of a series of discrete points containing timestamps and spatial coordinates (such as Ant-ID, Time) according to the reading frequency.

[0063] The path association processing submodule calls the location trajectory fragment set, performs path sorting processing on each fragment and the identified content according to the time order, calculates the time distance and jump order offset between consecutive positions in the path, classifies and merges the jump situation between fragments according to material identification, combines and describes the channel trajectory corresponding to each identification, and obtains the passage trajectory content in the material warehouse.

[0064] The system calls a cleaned set of position trajectory segments to restore discrete coordinate points into continuous motion vectors. Based on time sequence, the system performs path sorting processing on each segment and the identified content. To verify the physical rationality of the path, the system introduces a motion verification model, setting the standard operating speed of the AGV to 1.0 m / s and the allowable system error offset threshold to 0.5 seconds. The processor calculates the actual time distance between adjacent consecutive positions in the path and compares it with the theoretical time distance calculated based on physical distance and standard speed to obtain the jump sequence offset value. In this embodiment, when the calculated offset value is less than the set 0.5-second threshold, the system determines that the position jump conforms to physical laws and belongs to normal continuous passage behavior. The system classifies and merges the jump situations between all verified segments according to material identification, eliminates any abnormal jump points that do not conform to motion logic, and performs vectorized connection and combination description on the selected point series, thereby constructing a smooth material warehouse passage trajectory that includes the complete spatiotemporal curve of the material from the channel entrance Ant-01 to the end Ant-05.

[0065] Please see Figure 4 The open cargo location guidance execution module includes:

[0066] The direction extraction submodule calls the content of the passage trajectory in the material warehouse, identifies the continuous passage segments and directional order in the trajectory, extracts the flow number information under the corresponding direction, matches and filters it with the content of the task to be processed in the work task, performs number classification operation on the matching items, and obtains the task location number set.

[0067] The system retrieves the material warehouse's internal movement trajectory data, analyzes the starting and ending coordinates of the trajectory, constructs a displacement vector, and calculates its direction. In this embodiment, the system identifies that the overall trajectory shows an increasing trend in the Y-axis coordinate. Based on this, it determines that the macroscopic flow direction of the materials is the north area in the positive Y-axis direction. The system extracts the flow number information under the corresponding north area direction and uses it as a query condition to interact with the job task list in the WMS system. The system matches and filters the content of the tasks to be processed. The system traverses the task pool and retrieves all job tasks whose target storage area is located in the "north area" and whose current status is marked as "pending shelving". By comparing the flow number with the associated ID in the task document, the system filters out specific tasks that match the current flow direction. The system performs a number classification operation on all matching items, centralizing the scattered task data to obtain a set of task location numbers containing specific target location numbers (such as RACK-05-L2-10).

[0068] The address mapping submodule calls the task location number set, extracts the address field information by referring to the instruction equipment binding table according to the number order, performs field mapping processing between the number content and the equipment address, identifies and removes duplicate numbers and missing address items during the retrieval process, and sorts the valid matching results into a sequence to obtain the location address mapping list.

[0069] The system calls the task location number set to convert logical location numbers into control addresses for physical devices. It retrieves address field information by comparing the numbers with a pre-stored device binding table in the controller. This binding table details the hardware ID and register address of the electronic tag (PTL) device corresponding to each location number. The system performs field mapping between the number RACK-05-L2-10 in the set and the records in the table, accurately extracting the underlying device ID 0x0B and the function register address for that location. During the retrieval process, the system's built-in exception handling logic identifies and removes any duplicate number requests or missing address entries not found in the table, preventing invalid instructions from blocking the bus. The cleaned and valid matching results are then sorted according to the bus topology to obtain a location address mapping list containing ordered device addresses, register information, and communication protocol parameters.

[0070] The instruction triggering submodule calls the cargo location address mapping list, sends guidance instructions to the corresponding address devices in sequence, monitors the device receiving status and records the receiving feedback content, determines whether the device has entered the indication state based on the indication status code in the feedback, controls the duration of the state until the path passage is completed, and obtains the cargo location guidance execution record.

[0071] The system invokes the location address mapping list, acting as a bridge between the information system and the physical actuators. It sequentially sends guidance instructions in Modbus RTU standard format to the corresponding address devices in the list. These instructions explicitly include the device address, write coil function code, and the action parameters for setting and illuminating the indicator light. The aim is to control the target electronic tag to perform the setting and illuminating action to guide the operator. After the instruction is sent, the system immediately enters listening mode, monitoring the device's reception status on the bus and recording the received feedback. The processor compares the consistency between the feedback frame and the sent instruction frame to determine if the device has successfully entered the indication state. If the feedback matches, the lighting is confirmed to be successful. To save energy and adapt to the work rhythm, the system controls the duration of the state, setting an automatic extinguishing mechanism with a 30-second countdown, or terminating early when the sensor detects a pickup / release confirmation signal. The system records the start and end times, device response status, and results of each operation in detail, obtaining a location guidance execution record containing location coordinates, illumination status, and precise operation time.

[0072] Please see Figure 5 The access control module includes:

[0073] The location number extraction submodule calls the location guidance execution record, extracts the location numbers involved in each record, sorts the numbers into segments according to time sequence, compares the sorted numbers with the operation timeline, and removes the number content that does not fall into the valid time segment to obtain the set of accessible location numbers.

[0074] The system retrieves the accumulated location guidance execution records and extracts the target location number (e.g., RACK-05-L2-10) from each record. To ensure the real-time nature and effectiveness of the verification, the system segments these numbers according to time sequence and introduces a time window filtering mechanism. The processed numbers are compared sequentially with the actual operation timeline. The system sets an effective operation window based on the time the guidance command is issued (e.g., within 60 seconds after the command is issued). The processor checks the operation timestamp of each record and removes those records whose times do not fall within the effective operation window. These removed records represent timeout operations or historical data interference. After this round of fine filtering, a set of accessible location numbers that accurately reflects the current active operation status is obtained.

[0075] The identification information collection submodule calls the set of access location numbers, collects the personnel identification number and radio frequency reading content in the channel where the corresponding location number is located during the passage period, performs joint matching of the reading content and personnel number in chronological order, removes duplicate personnel records and blank reading items, and obtains the access identification information set.

[0076] The system calls the aforementioned set of access location numbers. For each valid location number in the set, the system automatically activates the monitoring equipment of its physical channel (such as RACK-05 lane). During the defined access period, the system synchronously collects personnel identification numbers (via facial recognition or RFID tags) and material radio frequency readings. In this embodiment, the system collects personnel number USER-OP-03 and material radio frequency content EPC-30396062. To determine whether these two belong to the same operation, the system sets a joint matching threshold of 1.0 second. The processor performs joint matching of the timestamp of the reading content and the timestamp of the personnel identification number in chronological order. When the calculated time difference between the two is less than the set threshold of 1.0 second, the system determines that the two are highly overlapping in time and space and belong to the same related access process. During this process, the system will also automatically remove repeated personnel identification records in a short period of time and blank reading items that only contain personnel and no materials (or vice versa), thus obtaining a set of access identification information containing the relationship between the determined operator and the corresponding operated material.

[0077] The access verification submodule calls the access identification information set, matches the personnel number with the radio frequency content according to the reading time, extracts the numbers and contents that cannot be matched and classifies them into abnormal access processes, processes them by cargo location number, outputs the number of unmatched reading tags in each group, and obtains the abnormal access marker content.

[0078] The system invokes the access identification information set and performs compliance verification. It matches personnel numbers and RFID content in the set with each read time and retrieves the currently active task authorization table from the access control database for comparison. This authorization table specifies that specific tasks must be performed by specific personnel on specific materials. The system checks each collected "person-material" relationship against the authorization table. If a discrepancy is found between the collected data and the authorized task (e.g., personnel are not authorized to operate the material, or the material is in the wrong area), the system extracts these unmatched numbers and contents and categorizes them as abnormal access processes. The system groups these abnormal records by storage location number, calculates the nature of the violation in each group, and outputs the number of unmatched read tags in each group. For example, in this embodiment, the number of unauthorized personnel entering storage location RACK-05-L2-10 is counted as 1, and abnormal access tag content containing detailed abnormality type, location, and quantity statistics is obtained.

[0079] Please see Figure 6 The unmanned warehouse management module includes:

[0080] The location extraction submodule calls the access location information in the abnormal access marker content, extracts the material identification content and operation instruction number corresponding to the location, merges the identification content and instruction number in chronological order, marks and filters out the location numbers that are missing any content, and obtains a set of unmatched location numbers.

[0081] The system retrieves the abnormal access marker content and extracts the specific access location information RACK-05-L2-10 from it. It then further traces back to extract the material identification content and operation instruction number associated with this location. The system merges and organizes the identified material content and the corresponding operation instruction number in chronological order, attempting to reconstruct the context of the abnormality. During this process, if the system finds location information but cannot associate it with any corresponding valid instruction number within the current time window (i.e., there should theoretically be no operation activity at this location), this usually means that an unplanned intrusion or misoperation has occurred. The system then marks and filters out the location number and adds it to the set of unmatched location numbers.

[0082] The instruction verification submodule calls the set of unmatched storage location numbers, compares whether there is a corresponding item between the valid identification content and the operation number in the current task for each number, checks whether the number mapping result forms a one-to-one record, and summarizes the content that cannot be verified to obtain a list of unattended storage locations.

[0083] When calling the set of unmatched location numbers, to prevent misjudgments due to network latency or data synchronization delays, the system performs a second deep verification logic. The processor compares whether there is a valid correspondence between the identification content and the job number for each number in the set in the current global task pool. For this purpose, the system sets a verification waiting period of 500 milliseconds. During this period, the system will suspend the judgment and actively refresh the task status to check whether the number mapping result can form a valid one-to-one record after the delay. If, after the 500-millisecond verification delay, the system still cannot find a valid task record for that location in the full database task list, the exception is determined to be a definite unverifiable state. The system then aggregates all unverifiable content to generate a detailed list of unattended locations.

[0084] The control action issuance submodule calls up the list of unmanned handling locations, matches them to the corresponding device addresses in the control device distribution table according to the number sequence, sends an entry control status command to the device address, records the device's received feedback content and continuous response action status, and obtains unmanned warehouse management execution commands.

[0085] The system invokes the unmanned storage location list and executes a physical blocking strategy. Following the location number order in the list, the system searches for and matches the corresponding device address (such as a security door controller or audible and visual alarm) in the control equipment distribution table. To achieve tiered control, the system has a built-in risk assessment model, setting the risk weight for personnel intrusion to 3.0, the risk weight for material misplacement to 1.0, and the baseline risk threshold for triggering a blockade to 2.0. The system performs a weighted calculation based on the number of detected anomalies and their corresponding weights, resulting in a current risk coefficient of 3.0. The system compares this value with the baseline risk threshold of 2.0. Since the calculated result of 3.0 is greater than the threshold of 2.0, it indicates that the current anomaly has exceeded the system's safety tolerance limit. Based on this, the system immediately sends a high-priority entry control command to the corresponding device address, forcibly locking the entrance and exit of the relevant area. After the command is issued, the system continuously records the device's received feedback and continuous response action status (such as door lock closing signals), obtaining unmanned warehouse management execution commands containing detailed physical blocking operation logs and status confirmation information.

[0086] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An unmanned warehouse management system based on artificial intelligence, characterized in that, The system includes: The material identification binding module obtains the material number, material category name, and arrival batch mark. It concatenates the material number and category name fields and writes them together with the batch mark into the RFID tag content, and associates them with the storage location to ensure that the tag content is consistent with the physical item, thus obtaining the unique identification content of the material. The warehouse access identification module calls the unique identification content of the material, collects the radio frequency reading position and entry / exit start and end information when the material passes through the warehouse passage, associates the reading position and identification content with the passage flow process, and continuously follows the material movement to complete the path association description, thereby obtaining the material warehouse passage trajectory content. The location guidance execution module retrieves the specified location number in the operation task based on the flow direction in the passage trajectory content, maps it to the address of the in-warehouse indicator device, enters the guidance state after receiving the trigger command, and continues until the flow is completed, thus obtaining the location guidance execution record. The entrance and exit behavior verification module calls the cargo location number in the guidance execution record, collects the personnel identification number and radio frequency reading content in the passage process, processes them one by one, and classifies the passage process that does not form a corresponding relationship as abnormal to obtain abnormal passage mark content.

2. The unmanned warehouse management system based on artificial intelligence according to claim 1, characterized in that: The unique identification content of the material includes the radio frequency tag encoding structure, the material category splicing field, and the storage location association information. The material warehouse passage trajectory content includes the path location sequence, passage time node, and material identification mapping. The storage location guidance execution record includes the target storage location number, the indicating device address, and the guidance status identifier. The abnormal passage marker content includes the unmatched personnel number, the abnormal identification tag, and the passage abnormality type.

3. The unmanned warehouse management system based on artificial intelligence according to claim 1, characterized in that, The material identification binding module includes: The information splicing submodule obtains the material number, material category name, and arrival batch mark registered in the warehousing operation, performs character splicing operation according to the field order of material number and category name, connects the result with the batch mark, identifies whether the character position is repeated or offset according to the set order rules, and adjusts the character arrangement order to obtain the spliced ​​field content. The tag writing submodule calls the content of the splicing field, allocates the character content to the corresponding area of ​​the RFID tag according to the tag structure, performs character division and content writing operations in the order of segments, detects the offset between the content of each segment and the set range, rewrites after adjusting the order, collects the return status of each segment, and obtains the tag writing feedback set. The location association submodule calls the tag writing feedback set, determines whether to perform the information binding action based on the returned status, obtains the location information in the operation, performs a pairing operation with the content in the RFID tag according to the field order, writes the corresponding relationship into the location data area, and obtains the unique identification content of the material.

4. The unmanned warehouse management system based on artificial intelligence according to claim 1, characterized in that, The warehouse access identification module includes: The identification information calling submodule calls the unique identification content of the material. When the material enters the warehouse channel entrance, the identification content field is read synchronously. The time sequence is compared according to the data collected at the channel entrance time period. It checks whether the identification field matches the channel passage record within the preset range. The fields are sorted and classified according to the time correspondence to obtain the identification matching mark set. The location information acquisition submodule calls the identification and matching mark set to collect radio frequency reading location data and entry / exit start and end information of materials during the passage process. It combines and filters according to reading time and position order, performs channel direction screening on the collected items, removes duplicate records and dwell segments, and generates a set of location trajectory segments according to reading frequency. The path association processing submodule calls the set of location trajectory segments, performs path sorting processing on each segment and the identified content according to the time order, calculates the time distance and jump order offset value between consecutive positions in the path, classifies and merges the jump situation between segments according to material identification, combines and describes the channel trajectory corresponding to each identification, and obtains the passage trajectory content in the material warehouse.

5. The unmanned warehouse management system based on artificial intelligence according to claim 1, characterized in that, The open cargo location guidance execution module includes: The direction extraction submodule calls the passage trajectory content in the material warehouse, identifies the continuous passage segments and directional order in the trajectory, extracts the flow number information under the corresponding direction, matches and filters it with the content of the task to be processed in the work task, performs number classification operation on the matching items, and obtains the task location number set. The address mapping submodule calls the task location number set, extracts the address field information by referring to the instruction equipment binding table according to the number order, performs field mapping processing between the number content and the equipment address, identifies and removes duplicate numbers and missing address items during the retrieval process, and sorts the valid matching results into a sequence to obtain the location address mapping list. The instruction triggering submodule calls the cargo location address mapping list, sequentially sends guidance instructions to the corresponding address devices, monitors the device receiving status and records the receiving feedback content, determines whether the device has entered the indication state based on the indication status code in the feedback, controls the duration of the state until the path passage is completed, and obtains the cargo location guidance execution record.

6. The unmanned warehouse management system based on artificial intelligence according to claim 1, characterized in that, The specified entrance / exit behavior core module includes: The storage location number extraction submodule calls the storage location guidance execution record, extracts the storage location numbers involved in each record, segments and organizes the numbers according to the time sequence, compares the organized numbers with the operation timeline in sequence, removes the number content that does not fall into the valid time segment, and obtains the set of passable storage location numbers. The identification information collection submodule calls the set of access location numbers, collects personnel identification numbers and radio frequency readings in the passage of the corresponding location number during the passage period, performs joint matching of readings and personnel numbers in chronological order, removes duplicate personnel records and blank readings, and obtains a set of access identification information. The access verification submodule calls the access identification information set, matches the personnel number with the radio frequency content according to the reading time, extracts the numbers and contents that cannot be matched and classifies them into abnormal access processes, groups them according to the cargo location number, outputs the number of unmatched reading identifiers in each group, and obtains the abnormal access marker content.

7. The unmanned warehouse management system based on artificial intelligence according to claim 1, characterized in that, The system also includes The unmanned warehouse management module retrieves the corresponding material identification content and operation instruction number based on the access location associated with the abnormal access marker content, checks the relationship between the two item by item, includes the unassociated storage locations into the unmanned processing scope and issues control actions, and obtains unmanned warehouse management execution instructions. The unmanned warehouse management execution instructions include the pending storage location number, abnormal operation instructions, and automatic control strategies.

8. The unmanned warehouse management system based on artificial intelligence according to claim 7, characterized in that, The unmanned warehouse management module includes: The location extraction submodule calls the access location information in the abnormal access marker content, extracts the material identification content and operation instruction number corresponding to the location, merges the identification content and instruction number in chronological order, marks and filters out the location number that is missing any content, and obtains a set of unmatched location numbers. The instruction verification submodule calls the set of unmatched storage location numbers, compares whether there is a corresponding item between the valid identification content and the operation number in the current task for each number, checks whether the number mapping result forms a one-to-one record, and summarizes the content that cannot be verified to obtain a list of unattended storage locations. The control action issuance submodule calls the unmanned processing location list, matches it to the corresponding device address in the control device distribution table according to the number order, sends an entry control status command to the device address, records the device's received feedback content and continuous response action status, and obtains the unmanned warehouse management execution command.

Images