Intelligent shelf location health state evaluation method and system

Abstract

The application discloses a smart shelf goods location health state evaluation method and system, and relates to the technical field of smart warehouse equipment state monitoring.The application realizes complete reservation and cross-goods location correlation of goods loading dynamic characteristics by collecting dynamic differential waveforms of goods in the migration process between goods locations and splicing the waveforms into data stream fragments along the migration path.On this basis, layered coding of the shape and time constant of each waveform can accurately depict the bearing response characteristics of different goods locations, and then through the step-by-step consistency comparison along the path, the interference of goods differences and environmental noise is effectively eliminated, and the sensitivity and specificity of goods location health state discrimination are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent warehousing equipment status monitoring technology, and in particular to a method and system for assessing the health status of intelligent shelving locations. Background Technology

[0002] In the field of intelligent warehousing and logistics management, the health status of storage locations, as the basic unit of goods storage, directly affects inventory accuracy, operational efficiency, and equipment safety. In recent years, with the development of the Industrial Internet of Things (IIoT) and edge computing technologies, storage location status monitoring technologies have gradually evolved from traditional periodic manual inspections to online automatic assessments based on multi-source sensors. Existing technical solutions mainly include: static weighing monitoring based on strain gauge or piezoelectric weight sensors, which determines the presence of abnormal loads by detecting the weight difference between empty and fully loaded storage locations; goods trajectory tracking based on radio frequency identification (RFID), which uses readers to read the time-series information of electronic tags to construct the movement path of goods within the warehouse; and comprehensive status assessment methods combining auxiliary sensors such as vibration and temperature sensors. Some advanced systems further employ machine learning methods to analyze sensor time-series data, attempting to identify early signs of failure such as storage location structural deformation, sensor drift, or wear on bearing surfaces. Furthermore, for dynamic monitoring of the goods migration process, existing research has proposed determining the standardization of handling operations by recording the transfer time windows of goods between different storage locations and combining this with weight change curves. Overall, existing technologies are developing towards multi-source data fusion, real-time online analysis, and predictive maintenance.

[0003] However, the aforementioned existing technologies still have significant shortcomings in practical applications. First, static weight monitoring can only reflect the load status of the storage location at a single moment and cannot capture the dynamic characteristics of weight changes during cargo placement, such as overshoot, oscillation, or response hysteresis. These dynamic characteristics are precisely important early indicators of aging of elastic components in the storage location, changes in damping characteristics, or sensor loosening. Second, there is a lack of an effective spatiotemporal correlation mechanism between RFID-based migration path records and storage location weight data. This leads to the isolated analysis of weight response curves of the same cargo in different storage locations, making it difficult to eliminate individual differences introduced by factors such as cargo weight and placement posture through cross-storage comparisons. Consequently, it is impossible to accurately identify the health deterioration of the storage location itself. Third, existing waveform analysis methods mostly use fixed thresholds or single feature extraction (such as peak value, steady-state value), without hierarchical encoding of waveform morphology types (such as exponential, overshoot, and stepwise) and time constants. Therefore, it is difficult to quantitatively describe the response consistency of different storage locations under the same cargo loading conditions. Fourth, the lack of a step-by-step consistency comparison strategy along the migration path makes it impossible to compare the response pattern of a normal storage location with the response pattern of a normal storage location when an anomaly occurs, resulting in a high false alarm rate or missed detection. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, as well as in the specification abstract and the title of the invention, to avoid obscuring the purpose of this section, the specification abstract, and the title of the invention. Such simplifications or omissions shall not be used to limit the scope of the invention.

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: Differential waveforms of each cargo location during the transition from empty to full load are collected, and the migration path of goods between different cargo locations is recorded simultaneously; differential waveforms of each relevant cargo location are extracted along the migration path and spliced ​​together in chronological order to form a segment of the cargo migration data stream; layered encoding is performed on the differential waveforms of each cargo location within the migration data stream segment to obtain a waveform morphology encoding layer and a time constant encoding layer; the consistency of the layered encoding of each cargo location within the migration data stream segment is compared, and when a mismatch in encoding layers occurs, the health status of the corresponding cargo location is determined to be deteriorated.

[0007] As a preferred embodiment of the present invention, the acquisition of differential waveforms for each cargo location during the transition from empty to full load includes: when the weight sensor detects that the weight is continuously increasing from the empty reference value, and the rate of weight change is greater than the rate of increase threshold and remains no less than the shortest increase time, triggering the data acquisition window for the corresponding cargo location; recording the continuous weight change curve from the trigger start time to the end time when the weight first meets the stability judgment condition, with each sampling time and the corresponding weight value constituting a time-weight sequence, denoted as a differential waveform; during the period from the trigger start time to the end time, if the weight has exceeded twice the empty reference value... If, after an abnormal descent margin, a weight value appears that is less than the current maximum weight value minus the abnormal descent margin, the data acquisition window is terminated early and the differential waveform is marked as abnormally placed and not involved in subsequent processing. The stability determination condition is: within the stable duration window, the absolute value of the weight change is less than the stable fluctuation threshold. During the data acquisition window, the electronic tag reader reads the electronic tags attached to the goods placed in the storage location at preset reading intervals. If the unique identifier of the goods, i.e., the electronic product code, is successfully obtained at any reading time within the window, the electronic product code is appended to the header of the differential waveform file at the end of the window.

[0008] As a preferred embodiment of the present invention, the recording of the migration path of goods between different storage locations includes: maintaining a list sorted by waveform trigger start time for each electronic product code; whenever a non-abnormal differential waveform is generated, adding the header information to the list of the corresponding electronic product code and automatically sorting it by trigger start time; when the list length is not less than 2, checking adjacent differential waveform pairs in chronological order and generating migration path records: if the storage locations of two adjacent differential waveforms are the same, they are considered as multiple independent operations in the same storage location, and no migration path record is generated, only the waveform is retained for other analyses; if the storage locations are different, the migration interval is calculated, the migration interval being equal to the trigger start time of the later differential waveform minus the end time of the previous differential waveform; if the migration interval is greater than a preset maximum migration time threshold, the corresponding migration segment is marked as a transmission timeout; if the migration interval is less than zero, it is marked as a time overlap anomaly; generating migration path records in sequence; when marked as a time overlap anomaly, discarding the corresponding migration path or skipping the subsequent processing of the electronic product code.

[0009] As a preferred embodiment of the present invention, the step of extracting the differential waveforms of each relevant storage location includes: extracting the complete binary data of the differential waveforms of each relevant storage location sequentially according to the storage location order corresponding to the same electronic product code in the migration path record, wherein the binary data includes a file header and a weight data sequence; arranging the differential waveforms of each storage location in the order of the trigger start times in the file header; the separation mark includes: specifying a separation mark byte sequence, and specifying that if a byte value that is the same as or may conflict with the separation mark appears in the waveform binary data, an escape byte is inserted after it; the separation mark is inserted between the end time of the differential waveform of the previous storage location and the start time of the differential waveform of the next storage location.

[0010] As a preferred embodiment of the present invention, the step of splicing together the migration data stream fragment of the goods includes: performing escape encoding on the 16-byte original binary data of the electronic product code to obtain the escaped EPC field; placing the escaped EPC field at the beginning of the data string, followed by a three-byte separator, and then sequentially splicing the escaped waveform data block and the separator; defining the combined overall data string as the migration data stream fragment of the goods.

[0011] As a preferred embodiment of the present invention, the hierarchical encoding of the differential waveforms of each storage location within the migration data stream segment includes: segmenting the migration data stream segment into independent differential waveforms for each relevant storage location according to the position of the separator mark; performing escape restoration during segmentation; reading the trigger start time, end time, and weight data sequence from the file header of each independent differential waveform; calculating a stable value for each independent differential waveform, wherein the stable value is the arithmetic mean of the weight values ​​within the last preset duration of the window; if the maximum weight exceeds the stable value multiplied by a preset overshoot ratio threshold, it is determined to be an overshoot type and encoded as a preset morphological code; otherwise, curve fitting is performed on the rising segment data from the trigger start time to the weight first entering the range of the stable value plus or minus a preset stable fluctuation threshold, with the fitting using an exponential curve shape; if the goodness of fit is greater than a preset goodness of fit threshold, it is determined to be an exponential type and encoded as a preset morphological code; otherwise, it is determined to be a step type and encoded as a preset morphological code; the waveform morphological code is a fixed-length character sequence.

[0012] As a preferred embodiment of the present invention, the following steps are taken: from the trigger start time to the time interval between the weight reaching its first stable value and the difference between the weight and the empty load reference value, multiplied by a preset ratio and added to the empty load reference value; if the time interval exceeds the preset maximum coded time, the maximum value is taken; the rise time is divided by the unit time length and rounded to the nearest integer as the quantization value; the quantization value is converted into a fixed-digit string, and if it is insufficient, zeros are padded at the high bits to obtain the time constant code; if the quantization value exceeds the upper limit of the code, the upper limit value is taken as the code; the waveform shape code and the time constant code of each independent differential waveform are concatenated in sequence to form the layered code of the cargo location, with the waveform shape code first and the time constant code second, and the total length is a fixed number of characters.

[0013] As a preferred embodiment of the present invention, the determination of the health status of the storage location includes: sequentially traversing each storage location along the migration path: if the current storage location is neither the first nor the last, then comparing the morphological code with the morphological codes of the previous and next storage locations; if the current morphological code is different from the morphological code of the previous storage location and also different from the morphological code of the next storage location, then marking the storage location as a morphological mismatch storage location; if the current storage location is the first storage location, then comparing the morphological code with the morphological code of the second storage location; if they are different, then marking it as a morphological mismatch; if the current storage location is the last storage location, then comparing the morphological code with the morphological code of the second-to-last storage location; if they are different, then marking it as a morphological mismatch.

[0014] As a preferred embodiment of the present invention, the determination of the health status of the storage location further includes: for each storage location not marked as morphological mismatch, comparing the time constant code with the time constant codes of adjacent storage locations; if it is different from the time constant codes of all adjacent storage locations, then marking the storage location as a constant mismatch storage location; recording the electronic product code corresponding to the storage location marked as morphological mismatch or constant mismatch and the location of the storage location in the migration path as a storage location with deteriorated health status.

[0015] This invention also includes an intelligent shelf location health status assessment system, comprising a data acquisition and recording module that acquires the differential waveform of each location during the transition from empty to full load, and simultaneously records the migration path of goods between different locations; a segmentation generation module that extracts the differential waveforms of each relevant location along the migration path and splices them in chronological order to form a migration data stream segment of the goods; a layered encoding module that performs layered encoding on the differential waveform of each location within the migration data stream segment to obtain a waveform morphology encoding layer and a time constant encoding layer; and a degradation judgment module that compares the consistency of the layered encoding of each location within the migration data stream segment, and determines that the health status of the corresponding location is degraded when a mismatch in encoding layers occurs.

[0016] The beneficial effects of this invention are as follows: By collecting dynamic differential waveforms of goods during their migration between storage locations and stitching the waveforms into data stream fragments along the migration path, this invention achieves complete preservation of the dynamic characteristics of goods loading and cross-storage location association. On this basis, by performing hierarchical encoding of the morphology and time constant of each waveform, the load-bearing response characteristics of different storage locations can be accurately characterized. Furthermore, by comparing the consistency along the path step by step, the interference of differences in the goods themselves and environmental noise is effectively eliminated, significantly improving the sensitivity and specificity of the health status judgment of storage locations.

[0017] This invention can identify latent faults such as fatigue of elastic components in storage locations, changes in damping characteristics, or sensor drift at an early stage, reducing the frequency of manual inspections and extending equipment life. At the same time, it avoids false alarms and missed detections caused by isolated threshold judgments, providing reliable technical support for predictive maintenance of intelligent warehousing systems. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart of the intelligent shelf location health status assessment method shown in this invention.

[0019] Figure 2This is a structural diagram of the intelligent shelf location health status assessment system shown in this invention. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort should fall within the scope of protection of this invention.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] According to an embodiment of the present invention, in combination Figure 1 The flowchart shown illustrates the method for assessing the health status of smart shelving locations, including: S1: Collect the differential waveform of each storage location during the process of switching from empty to full load, and record the migration path of goods between different storage locations.

[0024] First, during the system initialization phase, a static calibration process is performed for each storage location, including the following steps: With the cargo location in an unloaded state, the output signal of the weight sensor is continuously collected at a sampling frequency of not less than 20 Hz. After median filtering, the arithmetic mean is calculated and stored as the empty load reference value of the cargo location. The value is recalibrated every 24 hours or after each cargo location maintenance.

[0025] Simultaneously, a preset ascent rate threshold (range 0.05~0.20 kg / s) and a minimum ascent time are defined. (Value range 0.05~0.20 s), stable duration window (value range 0.3~1.0 s), stable fluctuation threshold (Value range 0.02~0.10 kg) and abnormal drop margin (value range 0.15~0.30 kg).

[0026] The above parameters are stored in the edge computing nodes in the form of configuration files and can be dynamically adjusted remotely.

[0027] Furthermore, the weight sensor uses a fixed sampling frequency. ( The system continuously collects the instantaneous weight values ​​of the current storage location (≥ 20 Hz). A sliding window of length L is maintained, where... .

[0028] Real-time calculation of weight change rate within the window: The slope of the time-weight data points within the window can be fitted using the least squares method, or the difference between the endpoints can be used. A valid ascent trigger is determined when the following conditions are met simultaneously: the current weight is greater than or equal to the no-load baseline value plus a preset small positive number (e.g., 0.01 kg); the weight change rate within the window is greater than the ascent rate threshold and remains no less than the shortest ascent time. Once the trigger conditions are met, the current time is recorded as the trigger start time, and the data acquisition window is opened.

[0029] During the acquisition window, each sampling moment and its corresponding weight value are continuously recorded to form the original time-weight sequence. Simultaneously, real-time monitoring is conducted to detect any abnormal weight decreases: the abnormal decrease criterion is defined as follows: after the weight has exceeded the no-load baseline value plus twice the abnormal decrease margin, there exists at any sampling moment where the weight value is less than the currently observed maximum weight value minus the abnormal decrease margin.

[0030] Once the condition is met, the acquisition window is immediately terminated, the recorded incomplete waveform is marked as an abnormal placement, the waveform data is discarded, it is not written to the subsequent processing queue, and the location identifier, trigger time and abnormal reason are recorded in the system log.

[0031] If the acquisition window ends normally, the moment when the weight first meets the stability criterion is recorded as the end moment. The stability criterion is implemented by: tracing back the stability duration from the current moment, calculating the range of weight values ​​within that time interval; if the range is less than the stability fluctuation threshold, the weight is considered stable. All time-weight data points from the start to the end moment are resampled at equal intervals (using linear interpolation, with the resampling frequency consistent with the original sampling frequency) to form a standardized differential waveform data volume.

[0032] Furthermore, UHF RFID readers are deployed above or to the side of each storage location, with the reading coverage area overlapping the storage location's carrying area. The readers and weight sensors share the same real-time clock source (such as a network clock synchronized via the IEEE 1588 PTP protocol or a GPS timing module), with a clock synchronization error not exceeding ±1 ms. The readers perform inventory operations cyclically at preset reading intervals (typically 0.05~0.20 s), returning the Electronic Product Code (EPC, 96-bit or 128-bit standard encoding) and corresponding reading timestamps for all electronic tags currently located in the storage location.

[0033] During the data acquisition window, the system maintains a temporary cache set to store EPCs successfully read within the window. Whenever the reader returns the inventory results, it iterates through each EPC. If the EPC is not yet in the temporary cache set, it is added, and the first occurrence time (EPC) is recorded. At the end of the window, if the temporary cache set is not empty, a master EPC is selected: if there is only one EPC in the temporary cache set, it is directly used as the cargo identifier for that waveform; if there are multiple EPCs in the temporary cache set, the master EPC is selected according to preset rules (such as selecting the earliest first occurrence time, the highest signal strength, or matching the EPC of the previous window), and the remaining EPCs are recorded as associated cargoes in the extended field of the file header.

[0034] The file header uses a compact binary structure and contains the following fields in sequence: EPC (fixed 16 bytes, padded with 0x00 on the right if insufficient, and truncated and marked if excessive), trigger start time (8-byte unsigned integer, representing the number of milliseconds since UTC 2000-01-01 00:00:00.000), end time (also 8 bytes), sampling frequency (4-byte single-precision floating-point number, in Hz), weight unit encoding (1 byte, 0x00 for kilogram, 0x01 for gram, 0x02 for newton), waveform data length (4-byte unsigned integer, representing the number of floating-point numbers in the subsequent weight data sequence), and reserved fields (4 bytes for expansion).

[0035] Following the file header, the weight values ​​for each sampling point are stored sequentially. Each weight value is a 4-byte single-precision floating-point number, arranged continuously in chronological order. The file header and the weight data sequence are written together into a binary file or memory block to complete the encapsulation of the differential waveform.

[0036] Ideally, the system maintains a global hash map table with the key being the string "EPC" and the value being a doubly linked list structure. Each node in the list stores a pointer to the generated waveform file (or file path) and the waveform's metadata (location ID, start time, end time, and anomaly flag). When a new non-abnormal differential waveform (i.e., a waveform not marked as abnormal placement) is encapsulated, the following operations are performed: extract the EPC from its file header; if the key for this EPC does not exist in the hash map table, create a new empty linked list; insert the metadata of the current waveform into the linked list structure corresponding to this EPC in ascending order of start time—specifically, since the waveforms are generated chronologically, they can usually be directly appended to the end of the linked list, but to prevent system clock rollback or out-of-order reporting, a binary search is still required to determine the insertion position, ensuring that the linked list is strictly arranged in ascending order of start time.

[0037] After inserting a new node into the linked list, check if the length of the linked list is greater than or equal to 2. If so, traverse each pair of adjacent nodes in the linked list. The migration path resolution process is executed first. The location ID and The location ID. If the two are the same, it means that the same goods have been placed independently multiple times in the same location (e.g., goods are temporarily moved out and then put back). In this case, the system does not generate a migration path record item across locations, but marks the two waveforms as continuous operations at the same location, which can be used in subsequent analysis to evaluate the consistency of repeated loading or sensor drift detection of the location.

[0038] If the two are different, then calculate the migration interval. A preset maximum migration time threshold is set (typically 300 seconds, adjustable based on warehouse layout and handling equipment speed). If the migration interval... If the time exceeds the maximum migration time threshold, an error flag indicating a transmission timeout is added to that migration segment; if the migration interval... If the value is less than 0, meaning the start time of the subsequent waveform is earlier than the end time of the previous waveform, then an abnormality flag indicating time overlap is added.

[0039] In the case of time overlap, since the two waveforms cannot coexist in different locations in time, it indicates a serious error in data acquisition or clock synchronization. The following processing will be performed: mark the entire linked list structure corresponding to the EPC as invalid, suspend the generation of subsequent migration paths for the EPC, generate an error alarm (containing the timestamp information of the EPC and the overlapping waveforms), and optionally skip all subsequent processing steps of the EPC (i.e., not participate in S2~S4).

[0040] For migration segments that only have a transmission timeout flag, the system retains the migration path for that segment, but writes the timeout flag into the path record for upper-level decision-making (such as whether to trigger manual review or reduce confidence).

[0041] The migration path record data structure is defined as an ordered array, where each element is a structure {location ID, start time, end time, flags}, where flags is a bitmask. Bit 0 indicates duplicates at the same location (but the actual migration path will not contain consecutive items with the same location; it is reserved here for expansion), bit 1 indicates transmission timeout, and bit 2 indicates time overlap. The system appends adjacent waveform pairs with different locations to the ordered array in chronological order.

[0042] The final generated migration path record is associated with the corresponding EPC and stored. This record does not delete the original waveform data, but only establishes a reference relationship to ensure data traceability.

[0043] S2: Extract the differential waveforms of each relevant cargo location along the migration path and splice them together in chronological order to form a fragment of the cargo migration data stream.

[0044] S2.1: Based on the storage location order corresponding to the same electronic product code in the migration path record, extract the differential waveform complete binary data of each relevant storage location in sequence. The binary data includes a file header and a weight data sequence.

[0045] Specifically, the system reads the ordered storage location sequence corresponding to the specified EPC from the migration path record generated in step S1. For each storage location, the system locates the corresponding differential waveform file in the waveform repository based on the start and end times recorded in the migration path record.

[0046] The specific location method is as follows: the storage path of each waveform file adopts a hierarchical directory structure. The system directly reads the complete binary content of the file based on this path, including the file header (fixed 64 bytes, containing fields such as EPC, start time, end time, sampling frequency, weight unit, and waveform data length) and the weight data sequence that follows (a continuously stored 32-bit single-precision floating-point number). After reading, the binary data body of each waveform is temporarily stored as an independent raw data block in the memory buffer.

[0047] S2.2: Arrange the differential waveforms of each storage location in the order of the trigger start times in the file header.

[0048] Ideally, to ensure that the splicing order is strictly consistent with the actual time sequence of the goods migration, the system does not directly rely on the listing order of the cargo locations in the migration path record. Instead, it rereads the start time field from the header of each waveform file and uses this field as the sorting key to stably sort the read raw data blocks.

[0049] If two waveforms have the same start time (due to clock resolution limitations or system malfunctions), their end times are further compared, with the one ending earlier being ranked first. If they are still the same, the lexicographical order of the storage location IDs determines the order. After sorting, a sequence of waveform data blocks arranged in ascending time order is obtained. ,in, This represents the total number of storage locations involved in the migration path. Each This corresponds to a complete differential waveform (file header + weight data).

[0050] The above operations ensure that even if there is uncertainty in the order of migration path records due to abnormal markings, the actual splicing is still based on the physical timestamp, thereby avoiding splicing errors caused by record generation delays or out-of-order reporting.

[0051] S2.3: Specify the sequence of separator byte sequences, and specify that if a byte value that is the same as or may conflict with the separator appears in the waveform binary data, an escape byte shall be inserted after it; the separator shall be inserted between the end time of the previous cargo position differential waveform and the start time of the next cargo position differential waveform.

[0052] It should be noted that, to prevent incorrect identification of waveform boundaries in subsequent steps, a unique separator mark is inserted between adjacent waveform data blocks that will not conflict with the internal binary data of the waveform. Specifically, the separator mark sequence is defined as three consecutive bytes 0x7E 0x7E 0x7E (i.e., the ASCII codes for three tilde characters). Since the weight values ​​in the original waveform data are stored as 32-bit floating-point numbers, the binary representation may randomly contain byte combinations that are the same as the separator mark (for example, some byte segments of the floating-point number 1.584e-5 happen to be equal to 0x7E). Therefore, an escaping mechanism (also known as byte stuffing) must be used to eliminate ambiguity.

[0053] A better implementation of the escaping mechanism is as follows: When traversing each byte of the waveform data block, maintain a sliding window comparator. When three consecutive bytes equal to 0x7E 0x7E 0x7E are detected, do not output these three bytes directly. Instead, insert an escape byte 0x7D after outputting the first 0x7E, and then continue outputting the next two 0x7E bytes, thus converting the original three-byte sequence 7E 7E 7E into a four-byte sequence 7E 7D 7E 7E.

[0054] In addition, to avoid confusion between the escape byte itself and the data, when 0x7D appears in the original data, it also needs to be replaced with 0x7D 0x7D (that is, the escape byte itself is repeated once). For non-conflicting bytes, output them as is.

[0055] The above escaping process treats the waveform file header and weight data sequence uniformly, without distinguishing field types, to ensure the integrity of the entire data block.

[0056] After the escape encoding is completed, the escaped waveform data blocks are written to the output buffer in sequence, and the original three-byte separator 0x7E 0x7E 0x7E is appended after each waveform data block (except for the last waveform). (This separator itself is no longer escaped because it is a boundary marker and will be identified and removed during segmentation.)

[0057] It should be noted that the length of the weight data inside the waveform data block will increase dynamically after escaping. Therefore, the total length of the final migration data stream fragment cannot be determined in advance and requires dynamic memory allocation or the use of an expandable streaming buffer.

[0058] S2.4: In this embodiment of the invention, before sequentially combining the arranged differential waveforms and separators into a continuous data string, the 16 bytes of original binary data of the Electronic Product Code are first subjected to the same escape encoding as in S2.3 to obtain the escaped EPC field. Then, the escaped EPC field is placed at the very beginning of the data string, followed immediately by a three-byte separator 0x7E 0x7E 0x7E, and then the escaped waveform data blocks and separators are sequentially concatenated. The entire data string is defined as a fragment of the cargo migration data stream.

[0059] The data structure of this fragment can be formally represented as: [Escaped EPC field] + [0x7E7E7E] + [Escaped EPC field] ] + [0x7E7E7E] + [escaped] ]+ [0x7E7E7E] +... + [escaped text] ].in, to The data has been sorted by timestamp. The system will store these fragments in a temporary memory cache or stream them directly to subsequent steps.

[0060] If subsequent processing requires multiple random accesses to the fragment content, it is recommended to persist the fragment as a temporary file or a memory-mapped area. At the same time, record the metadata of the fragment: the number of original waveforms n, the total byte length, the start time of the first waveform, the end time of the last waveform, and whether there are any transmission timeout or time overlap exception flags (these flags are inherited from the migration path record and written to the fragment tail extension area for reference during health determination).

[0061] It should be noted that this invention employs a three-byte fixed delimiter and a complete escaping (byte padding) mechanism to ensure that even if the same binary combination as the delimiter appears randomly within the waveform data, it will not be misjudged as a waveform boundary. Furthermore, the escaping rules reversibly replace both conflicting bytes and escape bytes in the original data, enabling subsequent steps to accurately restore the original waveform data through reverse escaping. This achieves lossless and unambiguous waveform sequence splicing and segmentation, significantly improving the reliability and robustness of data stream fragmentation.

[0062] S3: Perform layered encoding on the differential waveform of each storage location within the migration data stream segment to obtain the waveform morphology encoding layer and the time constant encoding layer.

[0063] S3.1: Segment the independent differential waveforms of each relevant storage location from the migration data stream fragments according to the position of the delimiter mark. When segmenting, perform escape restoration and read the trigger start time, end time and weight data sequence from the file header of each independent differential waveform.

[0064] Specifically, the generated migration data stream fragment is received. This fragment is a one-dimensional byte array with the following structure: an escaped EPC field, a three-byte separator 0x7E 0x7E 0x7E, and several escaped and encoded waveform data blocks. Adjacent waveform data blocks are also separated by the three-byte separator 0x7E 0x7E 0x7E.

[0065] The parsing process is as follows: First, scan from the beginning of the migration data stream fragment, finding the first three-byte separator 0x7E 0x7E 0x7E. The byte sequence before the separator is treated as the escaped EPC field. This field is then reversed to obtain 16 bytes of binary data (including any possible 0x00 padding on the right). All 0x00s on the right are removed to obtain the original electronic product code, used for subsequent log recording and result association. Then, parsing continues from after this separator. An empty list, WaveformList, is initialized to store the segmented independent waveform data. The current scan pointer is set to point to the first byte after the separator (i.e., the beginning of the first waveform data block), and the following loop scan continues until the pointer reaches the end of the fragment.

[0066] Within the loop, the system maintains a byte buffer, BlockBuffer, to accumulate escaped bytes of the current waveform data block. Bytes are read one by one, while simultaneously checking for a delimiter: since the delimiter consists of three consecutive 0x7E bytes, and the waveform data block has already undergone escape encoding (i.e., 0x7E 0x7E 0x7E in the original data has been converted to 0x7E0x7D 0x7E 0x7E, and 0x7D in the original data has been converted to 0x7D 0x7D), the delimiter is unique in the escaped data stream and will not be confused with the internal waveform data.

[0067] The specific scanning logic is as follows: Maintain a sliding window of length 3. When the window content equals [0x7E, 0x7E, 0x7E], it indicates the end of the current waveform data block. At this time, the accumulated bytes in the BlockBuffer are used as the escaped complete content of the waveform data block. After performing reverse escape and restoration, the original waveform binary data is obtained and stored in WaveformList. Then, the BlockBuffer is cleared, and the pointer skips the three-byte separator mark to continue parsing the next waveform data block.

[0068] If the end of the segment is reached without encountering a delimiter during the scanning process, the remaining bytes in the BlockBuffer are escaped and restored as the last waveform data block and added to the list.

[0069] Furthermore, the specific algorithms for reverse escape restoration include: Iterate through the input byte sequence and initialize the output buffer OutBuf. When the byte 0x7D is encountered, it indicates that it is followed by an escape byte, and the next byte next needs to be read. If next is equal to 0x7D, it is restored to a single 0x7D; if next is equal to 0x7E, it is restored to 0x7E; other combinations are considered data corruption, and an exception is thrown or the waveform is marked as invalid.

[0070] For bytes other than 0x7D, copy them directly to OutBuf. Note that since there are no separate delimiters in the original data, the restored byte sequence will not contain three consecutive 0x7E bytes, thus ensuring the uniqueness of the waveform boundaries. After reverse escaping, the original binary data of each independent differential waveform is obtained, where the first 64 bytes are the file header, and the subsequent bytes are the weight data sequence (a 32-bit floating-point array).

[0071] The system parses the trigger start time, end time, and sampling frequency from the file header. Including metadata such as weight units, and converting the weight data sequence into a float array. ,in .in, This represents the total length of the file.

[0072] S3.2: Perform morphological classification coding for each independent differential waveform.

[0073] Specifically, the first step is to calculate the stable value of the waveform: take the last preset duration before the waveform ends. Weight data points within a period of 0.5 seconds (same as in S1.1, typically). Based on the sampling frequency. Calculate the number of sampling points corresponding to this duration. From the last element of the weight array Calculate the arithmetic mean from the beginning to the end of each element. If the waveform length is insufficient... If there are 10 points, the average value of all waveform data is calculated, and a warning that the waveform is too short is recorded in the log.

[0074] Furthermore, it is necessary to detect whether there is an overshoot during the period from the start time of triggering to the end time. The specific process includes: Iterate through the entire weight array and record the maximum value. Preset overshoot ratio threshold (Typical value 0.05, i.e., 5%). If If the waveform pattern is determined to be overshoot, the waveform pattern code is assigned a predefined fixed-length string, such as OVS (3 characters); otherwise, the exponential decision branch is entered.

[0075] Exponential determination requires exponential curve fitting to the rising segment of the waveform. First, the start and end indices of the rising segment are determined: the start index is 0 (corresponding to the trigger start time); the end index is when the weight value first enters the interval. The index, where The stable fluctuation threshold is (same as S1.1, typical value 0.05 kg).

[0076] The consecutive data points from start to end form an ascending segment array. The corresponding time series ,in (Relative time, in seconds).

[0077] The exponential curve model is fitted using the nonlinear least squares method: in, This is the empty load baseline value for this storage location (obtained from the system configuration). is the time constant to be fitted.

[0078] The fitting process can be solved iteratively using the Levenberg-Marquardt algorithm, or approximated using a logarithmic linearization method: Let: but Obtained through linear regression The estimated value, and then the calculation .

[0079] After the fitting is completed, the goodness of fit is calculated. in, This is the predicted value from the exponential curve fitting model. This is the arithmetic mean of all actual weight values ​​during the ascent segment. No. The actual weight value (measured value) of each sampling point.

[0080] like If the goodness-of-fit value is greater than the preset threshold (typically 0.95), it is classified as exponential and coded as EXP; otherwise, it is classified as stepwise and coded as STP. All morphological codes are stored as 3-character ASCII strings.

[0081] Furthermore, for each independent differential waveform, after completing the morphological encoding, the time constant encoding is further calculated. Specifically, the target weight threshold is first defined as... in, The preset scaling factor is used (typically 0.9, representing 90% of the step response completion point). Starting from the trigger start time (index 0), the weight array is traversed in chronological order. Seeking the first satisfaction index If it exists, then (Unit: seconds); if the weight never reaches [a certain value] during the entire waveform period. ,but The duration of the entire window is calculated by subtracting the start time from the end time of the waveform. This is to prevent abnormal waveforms from causing... If too large, the maximum encoding time is preset, with a typical value of 10 seconds. If it exceeds the maximum coded time, then force It equals the maximum coded time.

[0082] Furthermore, the preset unit time length (Typical value 0.1 seconds, corresponding to a quantization resolution of 10 Hz). The original quantization value is: This value may be a non-integer. This embodiment of the invention uses rounding rules: Furthermore, the lower limit of the preset quantization interval is 0, and the upper limit is determined by the number of bits in the encoding. This embodiment of the invention uses a fixed 3-bit decimal string encoding, therefore the upper limit of the quantization interval is 999. If... If it is less than the lower limit of the quantization interval, then the lower limit of the quantization interval is used. If the value is greater than the upper limit of the quantization interval, then the upper limit of the quantization interval is used. The quantized integer value... Convert to a decimal string, ensuring the string length is 3: If the value is less than 100, padding with 0 to 3 characters on the left is used. For example, Q=5 is converted to 005, Q=42 to 042, and Q=123 to 123. This fixed-length string of numbers is the time constant code (3 characters). If the quantization value exceeds the upper limit of the quantization interval, it is encoded as 999, and a truncation warning is logged.

[0083] It should be noted that the time constant code and the waveform morphology code have the same length (both are 3 characters), which facilitates subsequent splicing and alignment.

[0084] Finally, for each independent differential waveform, the waveform morphology code and the time constant code are concatenated directly in sequence to form a 6-character string, where the waveform morphology code occupies the high 3 bits (first layer) and the time constant code occupies the low 3 bits (second layer).

[0085] The total length of this hierarchical encoding is a fixed number of characters (6 characters in this embodiment of the invention), and the encoding character set is limited to uppercase letters (AZ) and numbers (0-9), which facilitates storage and comparison.

[0086] The hierarchical codes for each storage location are generated sequentially according to the migration path order (i.e., the order of the waveforms in WaveformList), and these codes are stored in list form, while being associated with the corresponding storage location identifier and EPC. If an anomaly occurs during the parsing or encoding process (such as data corruption or fitting failure), its hierarchical code is marked as ERRERR (6 characters), and it is considered a candidate for health status deterioration in subsequent steps.

[0087] As can be seen, this invention employs a paired mechanism of escape byte padding and reverse escape restoration, combined with a unique three-byte separator, ensuring accurate segmentation of the migrated data stream. Even when the waveform data contains random byte sequences identical to the separator, the original waveform can be unambiguously recovered. This mechanism avoids the complexity and vulnerability of traditional fixed-length segmentation or reliance on external index tables, providing a complete and reliable data foundation for subsequent encoding. Furthermore, the overshoot, exponential, and stepped loading morphologies are derived from typical mechanical response characteristics during cargo loading, corresponding to different physical mechanisms such as sensor elastic deformation overshoot, viscoelastic damping-dominated smooth rise, and discrete stepped loading, respectively. By setting overshoot ratio thresholds and exponential goodness-of-fit thresholds, objective and repeatable automatic classification is achieved, avoiding subjective judgment.

[0088] S4: Compare the hierarchical coding consistency of each storage location within the migration data stream segment. When a coding level mismatch occurs, determine that the health status of the corresponding storage location has deteriorated.

[0089] It should be noted that migration paths with time overlap anomalies have been discarded or skipped in stage S1, therefore all paths entering S4 satisfy the physical rationality in terms of time. For migration segments with transmission timeout markers, these segments are still retained, but before comparison, it can be selectively configured whether to ignore timeout locations (for example, if the timeout period is too long, the sensor may drift, and the corresponding location's code can be marked as suspicious but not directly included in the comparison). By default, this embodiment of the invention includes timeout locations in the comparison normally, but additional timeout information is noted when an alarm is triggered.

[0090] In addition, before the comparison begins, the system performs a validity pre-check on the hierarchical coding sequences: if any code equals ERRERR (i.e., the erroneous code generated in S3 due to waveform abnormality or fitting failure), the location is directly marked as invalid data and will no longer participate in subsequent comparisons of morphological mismatch or constant mismatch, and will be recorded as a candidate for health status deterioration in the final output. After the pre-check is completed, the morphological code and time constant code of each valid location are extracted, and the morphological code sequence and time constant code sequence are used as comparison inputs.

[0091] S4.1: Traverse each storage location sequentially along the migration path: (1) If the current storage location is neither the first nor the last, then compare the morphology code with the morphology codes of the previous and next storage locations. If the current morphology code is different from the morphology code of the previous storage location and also different from the morphology code of the next storage location, that is, the morphology code of the current storage location is different from both its left and right neighbors, then the storage location is marked as a morphology mismatch storage location. For example, in the sequence [EXP, OVS, EXP], the middle OVS is different from both sides, so it is marked; while in the sequence [EXP, OVS, OVS], the second OVS is the same as the right side, which does not meet the condition of being different from both sides, so it is not marked. This is consistent with the reality that adjacent storage locations may simultaneously experience abnormalities when the health status continuously deteriorates.

[0092] (2) If the current storage location is the first storage location, compare the morphology code with the morphology code of the second storage location. If they are different, mark it as a morphology mismatch, which means that the load-bearing characteristics of the storage location have deviated from the normal baseline when the goods are initially placed, and an alarm should be issued.

[0093] (3) If the current storage location is the last storage location, compare the morphology code with the morphology code of the second to last storage location. If they are different, mark it as a morphology mismatch. This is used to detect abnormal states of the storage location where the goods finally stop.

[0094] The above judgment results are stored in a Boolean array, initially all set to false, and set to true when the condition is met. Simultaneously, the neighboring code values ​​of each mismatched storage location are recorded for subsequent traceability analysis. It is worth noting that if a storage location has already been excluded due to invalid data in the pre-screening, the morphology judgment for that location is skipped and remains false (it will not participate in the constant mismatch judgment later).

[0095] S4.2: For each storage location not marked as morphological mismatch, compare the time constant code with the time constant codes of adjacent storage locations according to the same boundary rules described above. If the time constant code is different from that of all adjacent storage locations, then mark the storage location as a constant mismatch storage location.

[0096] It should be noted that mismatched cargo locations typically reflect a fundamental change in the physical response type during cargo loading (e.g., a sudden shift from a smooth exponential type to an overshoot type). The health of these locations has clearly deteriorated, and further comparison of their time constants is unnecessary. Therefore, locations already marked as mismatched are first filtered out, and only the remaining locations (i.e., those whose shape codes are identical to or at least one end of their adjacent locations) undergo time constant code consistency checks. This hierarchical, cascaded decision-making strategy avoids interference with time constant statistics due to shape anomalies and also reduces the false alarm rate.

[0097] For each location index that is not marked as morphological mismatch (and the location data is valid), the same boundary rules as those used in S4.1 morphological determination are used to compare the time constant code with the time constant code of the adjacent location.

[0098] The determination result is stored in another Boolean array. It's important to note that even if a storage location's shape code is the same as one side, its time constant code may still differ from both sides. In this case, the storage location will be marked as having a constant mismatch. For example, in the sequence [EXP, EXP, EXP], all shapes are identical, but the time constant sequence is [010, 050, 010]. Therefore, the middle storage location 050 differs from both sides and is marked as having a constant mismatch, indicating an abnormal response speed for that location. Conversely, if a shape mismatch has occurred, the time constant is no longer checked, because shape mismatch itself is sufficient to determine health degradation, and shape mismatch may lead to unreliable time constant calculation benchmarks (such as stable values).

[0099] S4.3: Record the electronic product code corresponding to the storage location marked as morphological mismatch or constant mismatch, as well as the location of the storage location in the migration path, as a storage location with deteriorated health status, and output it to the log or alarm system.

[0100] Finally, the system uses the union of morphologically mismatched and constant mismatched storage locations as the final set of deteriorated storage locations in terms of health status. Specifically, for each storage location index, if both the Boolean arrays in S4.1 and S4.2 are true, then the storage location is included in the deterioration report. The report data structure is defined as: {EPC, list of deteriorated storage locations}, where each element in the list of deteriorated storage locations contains: storage location identifier, sequence number in the migration path, mismatch type (morphological mismatch, constant mismatch, or both), and corresponding neighborhood coding information (e.g., morphological coding and time constant coding of the previous and next storage locations, used for manual review).

[0101] If the number of deteriorated storage locations reaches a preset threshold (e.g., ≥1), an alarm log will be generated, with a record level of warning or critical, and the alarm information will be pushed to a message queue or operations dashboard. Simultaneously, the deteriorated storage location information will be persisted to a database table, with fields including: timestamp, EPC, storage location ID, path sequence number, mismatch type, and code details.

[0102] If the same storage location is marked as deteriorated in multiple different migration paths, the system can accumulate confidence scores. When the score exceeds the threshold, a higher-level maintenance work order will be triggered.

[0103] In addition, the system provides optional post-processing mechanisms. For example, for migration segments marked solely due to transmission timeouts, if the timed-out storage location itself is not mismatched in shape or constants, it will not be considered a health degradation, but an abnormal migration interval will be noted in the alarm remarks. If the timed-out storage location also has a mismatch, it will be marked as degraded normally, and timeout information will be added to the alarm to distinguish the cause of the degradation.

[0104] It should be noted that this invention uses direct comparison of adjacent cargo locations instead of global averages or thresholds, enabling it to sensitively capture isolated waveform morphology or response speed abrupt changes in a continuous migration path. Specialized boundary rules are designed for the first and last cargo locations to ensure that path endpoints can also be properly evaluated. Simultaneously, by using different judgment conditions than those on both sides, false alarms caused by simultaneous changes in the same trend in two consecutive adjacent cargo locations are avoided (for example, when multiple cargo locations drift due to consistent ambient temperature, their relative differences may still be zero, thus not triggering false alarms). Furthermore, morphological encoding and time constant encoding are separated into two independent comparison levels, with morphological judgment taking priority; cargo locations with morphological mismatches are no longer included in time constant comparisons. This design reduces unnecessary computational overhead and prevents distortions in stable values ​​or rise time calculations caused by morphological anomalies from contaminating the time constant encoding. Morphological mismatch reflects a fundamental change in the cargo loading response type (e.g., from viscoelastic to overshoot), while constant mismatch indicates a shift in response speed; together, they constitute a multi-dimensional characterization of the cargo location's health status.

[0105] This invention also includes an intelligent shelf location health status assessment system, comprising: The data acquisition and recording module collects the differential waveform of each storage location during the transition from empty to full load, and records the migration path of goods between different storage locations. The segmentation generation module extracts the differential waveforms of each relevant cargo location along the migration path and splices them together in chronological order to form the migration data stream segments of the cargo. The layered coding module performs layered coding on the differential waveform of each storage location within the migration data stream slice, resulting in a waveform morphology coding layer and a time constant coding layer. The degradation judgment module compares the hierarchical coding consistency of each storage location within the migration data stream segment. When a coding level mismatch occurs, it determines that the health status of the corresponding storage location has deteriorated.

[0106] The system also includes one or more processors and memory.

[0107] The memory is used to store operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the flow of the intelligent shelf location health status assessment method of the foregoing embodiments, in particular... Figure 1 The flowchart of the method is shown.

[0108] Other aspects disclosed in the embodiments of the present invention also propose a computer-readable medium for storing software including instructions executable by one or more computers, which, upon execution, cause the one or more computers to perform operations including the flow of the intelligent shelf location health status assessment method of the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0109] It should be recognized that embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium.

[0110] The method can be implemented using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program in the computer program, wherein the storage medium is configured such that the computer operates in a specific and predefined manner.

[0111] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if required, the program can be implemented in assembly or machine language.

[0112] In any case, the language can be either compiled or interpreted.

[0113] Furthermore, for this purpose, the program can run on programmed application-specific integrated circuits.

[0114] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0115] Furthermore, the method can be implemented in any suitable computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices.

[0116] Various aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether portable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein.

[0117] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0118] When such media includes instructions or programs that combine with a microprocessor or other data processor to implement the steps described above, the invention described herein includes these and other different types of non-transitory computer-readable storage media.

[0119] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for assessing the health status of intelligent shelving locations, characterized by: include: Collect differential waveforms for each storage location during the transition from empty to full load, and record the migration path of goods between different storage locations. Differential waveforms of each relevant cargo location are extracted along the migration path and spliced ​​together in chronological order to form a fragment of the cargo migration data stream; The differential waveform of each storage location within the migration data stream fragment is hierarchically encoded to obtain a waveform morphology encoding layer and a time constant encoding layer; By comparing the hierarchical coding consistency of each storage location within the migration data stream segment, when a coding level mismatch occurs, the health status of the corresponding storage location is determined to have deteriorated.

2. The method for assessing the health status of intelligent shelving locations as described in claim 1, characterized in that: The differential waveforms collected for each cargo location during the transition from empty to full load include: When the weight sensor detects that the weight is continuously rising from the empty load baseline value, and the rate of weight change is greater than the rate of increase threshold and continues to be no less than the shortest time of increase, the data acquisition window of the corresponding cargo location is triggered. Record the continuous weight change curve from the start of the trigger to the end of the time when the weight first meets the stability judgment condition. Each sampling time and the corresponding weight value constitute a time-weight sequence, which is denoted as a differential waveform. If, during the period from the start to the end of the trigger, the weight has exceeded the no-load reference value plus twice the abnormal drop margin, and then the weight value is less than the current maximum weight value minus the abnormal drop margin, the data acquisition window will be terminated in advance and the differential waveform will be marked as abnormal and will not participate in subsequent processing. The stability determination condition is: within the stable duration window, the absolute value of the weight change is less than the stable fluctuation threshold. During the data acquisition window, the electronic tag reader reads the electronic tags attached to the goods placed on the storage location at preset reading intervals; if the unique identifier of the goods, i.e., the electronic product code, is successfully obtained at any reading time within the window, the electronic product code is appended to the header of the differential waveform file at the end of the window.

3. The method for assessing the health status of intelligent shelving locations as described in claim 2, characterized in that: The recorded migration path of goods between different storage locations includes: Maintain a list sorted by waveform trigger start time for each electronic product code; Whenever a non-abnormal differential waveform is generated, the header information is added to the list of the corresponding electronic product codes and automatically sorted according to the trigger start time; When the list length is not less than 2, adjacent differential waveform pairs are checked in chronological order to generate migration path records: If two adjacent differential waveforms have the same storage location, they are considered as multiple independent operations at the same storage location. No migration path record is generated, and only the waveform is retained for other analyses. If the storage locations are different, the migration interval is calculated. The migration interval is equal to the start time of the subsequent differential waveform trigger minus the end time of the previous differential waveform. If the migration interval is greater than the preset maximum migration time threshold, the corresponding migration segment is marked as a transmission timeout. If the migration interval is less than zero, it is marked as a time overlap anomaly; Generate migration path records sequentially; When a time overlap anomaly is detected, the corresponding migration path is discarded or the subsequent processing of the electronic product code is skipped.

4. The intelligent shelf location health status assessment method as described in claim 3, characterized in that: The differential waveforms extracted from each relevant cargo location include: Based on the location order corresponding to the same electronic product code in the migration path record, the differential waveform complete binary data of each relevant location is extracted sequentially. The binary data includes a file header and a weight data sequence. Arrange the differential waveforms of each storage location in chronological order of the trigger start times in the file header; The separator includes: specifying a separator byte sequence, and specifying that if a byte value that is the same as or may conflict with the separator appears in the waveform binary data, an escape byte is inserted after it; the separator is inserted between the end time of the previous cargo location differential waveform and the start time of the next cargo location differential waveform.

5. The intelligent shelf location health status assessment method as described in claim 4, characterized in that: The process of splicing together the migration data stream fragments of the goods includes: The 16 bytes of raw binary data of the electronic product code are escaped to obtain the escaped EPC field; Place the escaped EPC field at the very beginning of the data string, followed by a three-byte separator, and then concatenate the escaped waveform data block and the separator in sequence; The combined data string is defined as a fragment of the cargo migration data stream.

6. The method for assessing the health status of intelligent shelving locations as described in claim 5, characterized in that: The layered encoding of the differential waveform of each storage location within the migration data stream segment includes: The independent differential waveforms of each relevant storage location are segmented from the migration data stream fragments according to the position of the delimiter mark. During segmentation, escape restoration is performed, and the trigger start time, end time, and weight data sequence in the file header of each independent differential waveform are read. For each independent differential waveform, a stable value is calculated, which is the arithmetic mean of the weight values ​​within the last preset duration of the window; If the maximum weight exceeds the stable value multiplied by a preset overshoot ratio threshold, it is determined to be an overshoot type and encoded with a preset morphological code; otherwise, the rising segment data from the start of the trigger to the first time the weight enters the range of the stable value plus or minus a preset stable fluctuation threshold is curve fitted. The fitting adopts an exponential curve shape. If the goodness of fit is greater than the preset goodness of fit threshold, it is determined to be an exponential type and encoded with a preset morphological code; otherwise, it is determined to be a step type and encoded with a preset morphological code; the waveform morphological code is a fixed-length character sequence.

7. The method for assessing the health status of intelligent shelving locations as described in claim 6, characterized in that: The time interval from the trigger start time to the first stable weight value minus the no-load reference value is multiplied by a preset ratio and added to the no-load reference value; if the time interval exceeds the preset maximum coded time, the maximum value is taken. Divide the rise time by the unit time length and round to the nearest integer as the quantified value; The quantized value is converted into a fixed-length number string. If the length is insufficient, the high-order bits are padded with zeros to obtain the time constant code. If the quantized value exceeds the upper limit of the code, the upper limit value is used for encoding. The waveform morphology code and time constant code of each independent differential waveform are concatenated in sequence to form the layered code of the cargo location. The format is waveform morphology code first, time constant code second, and the total length is a fixed number of characters.

8. The method for assessing the health status of intelligent shelving locations as described in claim 7, characterized in that: The determination of the health status of the storage location includes: Traverse each storage location sequentially along the migration path: If the current storage location is neither the first nor the last, then compare the shape code with the shape codes of the previous and next storage locations. If the current shape code is different from the shape code of the previous storage location and also different from the shape code of the next storage location, then mark the storage location as a shape mismatch storage location. If the current storage location is the first storage location, compare the shape code with the shape code of the second storage location. If they are different, mark it as a shape mismatch. If the current storage location is the last storage location, compare the morphology code with the morphology code of the second-to-last storage location. If they are different, mark it as a morphology mismatch.

9. The method for assessing the health status of intelligent shelving locations as described in claim 8, characterized in that: The determination of the health status of the storage location also includes: For each storage location not marked as morphological mismatch, the time constant code is compared with the time constant codes of adjacent storage locations. If it is different from the time constant codes of all adjacent storage locations, the storage location is marked as a constant mismatch storage location. The electronic product codes corresponding to the storage locations marked as morphological mismatch or constant mismatch, along with the location of those storage locations in the migration path, are recorded as storage locations with deteriorated health status.

10. An intelligent shelving location health status assessment system, based on the intelligent shelving location health status assessment method according to any one of claims 1 to 9, characterized in that: Also includes: The data acquisition and recording module collects the differential waveform of each storage location during the transition from empty to full load, and records the migration path of goods between different storage locations. The segmentation generation module extracts the differential waveforms of each relevant cargo location along the migration path and splices them together in chronological order to form the migration data stream segments of the cargo. The layered coding module performs layered coding on the differential waveform of each storage location within the migration data stream slice, resulting in a waveform morphology coding layer and a time constant coding layer. The degradation judgment module compares the hierarchical coding consistency of each storage location within the migration data stream segment. When a coding level mismatch occurs, it determines that the health status of the corresponding storage location has deteriorated.

Images