Method and system for on-line monitoring and fault early warning of health state of automated stereoscopic warehouse equipment

Abstract

The application discloses an automatic stereoscopic warehouse equipment health state online monitoring and fault early warning method and system, relates to the technical field of intelligent operation and maintenance and fault diagnosis, and realizes layered and accurate early warning of track subsidence and foundation bolt loosening through fusion of shock waveform overlapping degree analysis and pulse time interval change rate accumulation judgment, and significantly improves fault recognition rate and anti-interference capability; absolute value encoders and accelerometers are used for hardware level synchronous latching, so that the shock waveform and the spatial coordinates are accurately associated at the sub-millisecond level, and the accuracy of fault positioning is greatly improved; meanwhile, based on the cumulative criterion of three continuous access cycles, incidental interference such as load fluctuation and speed change is effectively filtered out, false positives and false negatives are avoided, and the effective early warning period is prolonged.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance and fault diagnosis technology, and in particular to a method and system for online monitoring of the health status and fault early warning of automated warehouse equipment. Background Technology

[0002] As a core piece of equipment in modern logistics systems, the health of stacker crane tracks and supporting structures in automated storage and retrieval systems (AS / RS) directly impacts the safety and continuity of warehousing operations. During high-speed reciprocating movement, the periodic impacts between the traveling wheels and the track joints are a major cause of track subsidence, loosening of anchor bolts, and even structural failure. Currently, health monitoring of AS / RS track structures primarily relies on manual inspections and periodic maintenance strategies, employing track straightness measuring instruments, levels, or visual checks to assess track condition while the machine is stopped. Some high-end systems have introduced online monitoring solutions based on vibration accelerometers, installing sensors on the track or column surfaces to identify anomalies through changes in time-domain or frequency-domain characteristics (such as root mean square value, peak value, and spectral energy). Furthermore, impact pulse counting methods based on piezoelectric films are also used to detect the frequency and intensity of the traveling wheels passing over joints, indirectly inferring the degree of track wear.

[0003] However, existing technologies have significant shortcomings in practical applications. First, manual inspection methods are inefficient, highly subjective, and struggle to detect early-stage microscopic defects, often only becoming apparent after the fault has become obvious, leading to unplanned downtime or even safety accidents. Second, while conventional vibration monitoring can achieve continuous data acquisition, it typically extracts only simple statistical features (such as vibration amplitude or RMS), lacking refined analysis of impact waveform morphology. This makes it difficult to distinguish between two distinct fault modes: track subsidence (causing waveform distortion) and loose anchor bolts (causing pulse time interval drift), easily resulting in false alarms or missed alarms. Furthermore, existing piezoelectric film pulse counting methods only record the number of pulses or their frequency, failing to utilize the time interval structure information within the pulse sequence when the same joint is passed multiple times. After anchor bolts loosen, the stiffness and damping of the vibration transmission path change, causing a systematic drift in the relative time interval between adjacent pulses. However, existing technologies lack quantitative characterization and cumulative judgment mechanisms for such fine characteristics, making early warning of loosening trends impossible. 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 abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract and title of the invention. Such simplifications or omissions shall not be used to limit the scope of the present 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: a method for online monitoring of the health status and fault early warning of automated storage and retrieval equipment, characterized by: dividing the travel track into multiple travel zones according to the track joints; recording the factory impact waveform of the joints in each travel zone and the factory pulse reference time sequence of the corresponding anchor bolt piezoelectric film; locating the current travel zone through a position encoder during operation; triggering an accelerometer to collect real-time impact waveforms when passing through a joint, and simultaneously collecting the real-time pulse moments of the travel zone; forming a first feature pair with the real-time waveform and the factory waveform and calculating the waveform overlap; forming a second feature pair with the real-time pulse sequence and the factory reference sequence and calculating the time interval change rate of adjacent pulses; outputting a track sinking early warning when the waveform overlap is lower than a first threshold; outputting an anchor bolt loosening early warning when the time interval change rate exceeds a second threshold for three consecutive access cycles; and uploading the early warning signal to the monitoring terminal according to the zone encoding.

[0007] As a preferred embodiment of the present invention, the step of dividing the travel track into multiple travel zones according to the track joints includes: using each track joint as a dividing point, defining the track segment between two adjacent joints as a travel zone; each travel zone adopts a left-closed and right-open coordinate interval representation method, and establishing a mapping table between the travel coordinate interval of each travel zone and the joint number; the travel zone is activated only when the stacker crane enters the zone coordinate interval in the forward direction, and is not activated when moving in the reverse direction.

[0008] As a preferred embodiment of the present invention, the recording of the factory impact waveform of the seam in each travel zone includes: using an impact hammer to simulate the impact mode when the stacker crane's traveling wheel passes through the seam, and sequentially striking the top of the rail at the rail seam in each travel zone; collecting the standard impact waveform of the seam by an accelerometer installed near the stacker crane's traveling wheel, setting the sampling rate to a high sampling rate sufficient to capture impact details, recording the duration covering the process of impact response decaying to the noise level, and using a pre-trigger buffer method to make the trigger point located at a set proportion position in the recording window.

[0009] As a preferred embodiment of the present invention, the recording of the factory pulse reference time sequence includes: selecting the anchor bolt closest to the horizontal distance of the track joint in the rack column corresponding to each travel zone, and attaching a piezoelectric film to the top of the bolt; the output end of each piezoelectric film is independently connected to a voltage comparator, and the output end of the comparator is connected to different digital input channels of the data acquisition card; during the factory calibration of the stacker crane, no-load operation is performed in a standard access cycle, and the rising edge of the comparator output of each channel is recorded, forming a factory reference time sequence according to the pulse appearance order; for travel zones that have not been traversed, the reference sequence is empty.

[0010] As a preferred embodiment of the present invention, the step of triggering the accelerometer to collect real-time impact waveforms when passing through a seam includes: installing an absolute encoder on the traveling wheel axle of the stacker crane to read the traveling position coordinates in real time; when the coordinates enter the coordinate range of a certain travel zone in the positive direction, activating the data acquisition window of the corresponding travel zone until it is closed when leaving the coordinate range; during the window activation period, comparing the accelerometer signal with a dynamic trigger threshold, the dynamic trigger threshold being obtained by multiplying the dynamic noise level of the accelerometer when the stacker crane travels on a flat track section by a fixed multiple; once the signal amplitude exceeds the dynamic trigger threshold, latching the current coordinates and recording the real-time impact waveform with the same sampling rate and recording length as the factory impact waveform, while using the same pre-trigger ratio as when recording the factory impact waveform; when multiple triggers occur within the same window, each waveform is recorded independently and the coordinates are marked.

[0011] As a preferred embodiment of the present invention, the simultaneous acquisition of real-time pulse moments of the travel partitions includes: the piezoelectric film signal of each travel partition is recorded by hardware interrupt at the moment of each rising edge through its respective voltage comparator and the digital input channel of the data acquisition card, and the partition number is automatically marked according to the channel affiliation; when the stacker crane completes each storage and retrieval operation and returns to the standby position, it summarizes the real-time pulse moments of all travel partitions during the corresponding segment of operation according to the partition to form the original real-time pulse sequence of this storage and retrieval cycle; all pulse moments of each travel partition are extracted from the original sequence and arranged in chronological order to form a real-time pulse sequence, and if a certain travel partition has no pulse, the sequence is empty.

[0012] As a preferred embodiment of the present invention, the step of forming a first feature pair with the real-time waveform and the factory waveform and calculating the waveform overlap includes: normalizing each real-time impact waveform belonging to the same stroke zone and the same joint with the corresponding factory standard impact waveform so that their maximum amplitude values ​​are the same; calculating the cross-correlation coefficient of the two normalized waveforms and using it as the waveform overlap; if multiple real-time impact waveforms are recorded in the same stroke zone and the same joint in a single access loop, the minimum value of all overlaps is taken as the final overlap of the stroke zone in this loop.

[0013] As a preferred embodiment of the present invention, the step of forming the second feature pair and calculating the time interval change rate of adjacent pulses includes: after each access cycle, for each travel partition, if the real-time pulse sequence is not empty and the factory reference sequence is not empty, then the two are aligned according to the order of pulse appearance, and the one with fewer pulses is truncated; the time interval between adjacent pulses in the factory reference sequence and the time interval between adjacent pulses after truncation in the real-time pulse sequence are calculated respectively; if the number of pulses in any sequence is less than two, then the change rate calculation of the travel partition for this time is skipped; for those that can form a time interval, the common interval number is taken, and the time interval change rate of each interval is calculated, that is, the absolute value of the difference between the real-time interval and the reference interval is divided by the reference interval; if the reference interval is zero, then the change rate is defined as zero.

[0014] In a preferred embodiment of the present invention: when the final waveform overlap of a certain travel partition is lower than a first preset threshold in a certain access cycle, it is determined that track subsidence has occurred at the track joint within the travel partition, and a subsidence warning message containing the partition number and the joint number is generated; for each travel partition, each pulse interval number is monitored, and if there is an interval number that causes the partition to be positively entered in three consecutive access cycles and the change rate of the corresponding three time intervals to be greater than a second preset threshold, it is determined that the anchor bolts in the travel partition are loose, and a loosening warning message is generated; if the travel partition is not positively entered in any cycle or the change rate cannot be calculated, the continuous counting is interrupted and accumulation needs to be restarted; the subsidence warning message and the loosening warning message are encoded by partition and sent to the monitoring terminal through the communication bus, and the monitoring terminal displays visual indicators of different warning types by partition, and displays the higher level when both exist.

[0015] The beneficial effects of this invention are as follows: By integrating impact waveform overlap analysis with the cumulative judgment of pulse time interval change rate, this invention achieves layered and precise early warning of track subsidence and anchor bolt loosening, significantly improving fault identification rate and anti-interference capability; by utilizing hardware-level synchronous latching of absolute encoders and accelerometers, it ensures sub-millisecond precise correlation between impact waveforms and spatial coordinates, greatly improving the accuracy of fault location; at the same time, based on the cumulative criterion of three consecutive access cycles, it effectively filters out occasional interference such as load fluctuations and speed changes, avoiding false alarms and missed alarms, and extending the effective early warning cycle; in addition, this invention does not require additional complex equipment, directly reuses the existing sensor resources of stacker cranes, and is easy to deploy at low cost in existing automated warehouse systems, thereby comprehensively improving the intelligence level of equipment operation and maintenance and operational safety. Attached Figure Description

[0016] 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 illustrating an online health status monitoring and fault early warning method for automated storage and retrieval systems (AS / RS) equipment, as shown in this invention.

[0017] Figure 2 This is a structural diagram of an online health status monitoring and fault early warning system for automated three-dimensional warehouse equipment, as shown in this invention. Detailed Implementation

[0018] 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.

[0019] 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.

[0020] 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.

[0021] According to an embodiment of the present invention, in combination Figure 1 The flowchart shown illustrates a method for online monitoring of the health status and fault early warning of automated storage and retrieval systems (AS / RS) equipment, including: S1: Divide the travel track into multiple travel zones according to the track joints, and record the factory impact waveform of the joint in each travel zone and the factory pulse reference time sequence of the corresponding anchor bolt piezoelectric film.

[0022] S1.1: In this embodiment, along the entire length of the stacker crane's travel track, each track joint is used as a dividing point, and the track segment between two adjacent joints is defined as a travel zone. Specifically, there are multiple track joints sequentially along the track travel direction (referred to as the forward direction), and their absolute coordinates are respectively... ,in, Let be the total number of seams. Then the th The travel coordinate interval of each travel zone is defined as a left-closed, right-open interval. ,in, The coordinates of the starting joint of this partition are... This is the coordinate of the next seam. This interval represents the coordinates from which the stacker crane starts. (Including) exercise to (Excluding) the track segments traversed.

[0023] Establish a mapping table between the travel coordinate intervals of each partition and the seam number. Since each partition contains only one track seam (located at the left endpoint of the interval), the unique seam number within that partition is always 1. The purpose of retaining the seam number index is to later expand to scenarios where each partition contains multiple seams, such as when the track seam spacing is uneven, multiple spacings can be divided into one partition.

[0024] The mapping table is stored in the controller and is used to quickly find the corresponding partition number and seam number based on the real-time location coordinates at runtime.

[0025] In this embodiment, a partition is activated only when the stacker crane moves in the forward direction (the direction of increasing coordinates) and enters a certain partition's coordinate range; when the stacker crane moves in the reverse direction and leaves the range, the partition is not activated. This operation avoids the same partition being repeatedly triggered during the round trip, reducing invalid data collection.

[0026] As can be seen, this invention ensures the unique ownership of the partition boundary by defining the interval as left-closed and right-open and by directional constraints, thus avoiding ambiguity when the position coordinates happen to fall on the boundary. At the same time, the mechanism of only positive activation is consistent with the physical process of the stacker crane passing through the seam in one direction during normal operation, which simplifies the triggering logic.

[0027] S1.2: In the rack uprights corresponding to each travel zone, select the anchor bolts with the closest horizontal distance to the track joint within that zone. This selection principle is based on structural dynamics analysis: when the stacker crane's traveling wheels pass over the track joint, the resulting impact vibration is transmitted along the track and the ground to the anchor bolts of the rack uprights. The bolts closest to the track experience the most significant vibration and have the highest signal-to-noise ratio. A piezoelectric film is attached to the top of this bolt; the piezoelectric film converts the strain changes caused by the vibration into a voltage signal.

[0028] Each piezoelectric film's output is independently connected to a voltage comparator. This comparator has a fixed voltage threshold (e.g., 1.5V) and a preamplifier circuit to amplify the weak piezoelectric signal to a triggerable level. The comparator outputs are connected to different digital input channels of the data acquisition card, enabling multi-channel parallel acquisition, with each channel corresponding to a travel partition.

[0029] During the factory calibration of the stacker crane, a standard access cycle is executed: the stacker crane starts from the standby position, travels forward along the track to the first storage location after the farthest seam, and then returns to the standby position in reverse. This standard cycle covers all possible travel zones. During this process, the data acquisition card records the rising edge of the comparator output on each digital input channel, i.e., the moment the piezoelectric film generates a pulse, and these pulses are arranged in order to form a factory reference time sequence. , where i is the partition number. For partitions that are not visited in the standard loop (e.g., track segments located behind the standby position), their reference sequence is defined as an empty sequence.

[0030] S1.3: This invention uses an impact hammer to simulate the impact mode when the traveling wheels of a stacker crane pass through a joint. The impact hammer is equipped with a hard rubber hammerhead, and its impact force spectrum is calibrated to match the force spectrum generated when the traveling wheels roll over the joint. The top of the rail at the joint within each travel zone is struck sequentially, with the striking point located at the center line of the rail top and the direction perpendicular to the rail surface.

[0031] The standard impact waveform of the joint is acquired by an accelerometer installed near the stacker crane's traveling wheels. The accelerometer is either piezoelectric or MEMS type, with a range and frequency response suitable for impact signal acquisition. The sampling rate is set to a high enough level to capture impact details, for example, no less than 10 kHz. The recording duration covers the entire process from the onset of the impact to the decay of the impact response below the noise level, typically tens of milliseconds.

[0032] Furthermore, the accelerometer signal is continuously written to the circular buffer. When an impact pulse is detected (triggered by a level comparator), a certain percentage (e.g., 10%) of the data before the trigger point and the remaining percentage after the trigger point are stored together as a complete waveform. This pre-trigger percentage is consistent with the percentage in subsequent real-time acquisition to ensure accurate waveform alignment. This waveform is denoted as... ,in, For partition number, This is the seam number (currently always 1).

[0033] It should be noted that by actively exciting the impact hammer, a standard waveform with a high signal-to-noise ratio can be obtained, and the pre-triggering mechanism retains the background signal before the impact arrives, which is convenient for identifying the starting point of the waveform during subsequent normalization processing.

[0034] S1.4: Under conditions where the stacker crane is stationary and there is no external impact, such as during the equipment installation and commissioning phase, with all power sources turned off and personnel movement avoided, the accelerometer continuously collects signals for 1 second, and the standard deviation of this signal segment is calculated as the static noise level. This value reflects the accelerometer's own electronic noise and ambient background vibration. It is stored in the controller and used to verify whether the accelerometer channel is working properly during the device's power-on self-test, or as a benchmark for dynamic noise level calibration.

[0035] Furthermore, when the stacker crane travels unloaded at its lowest stable speed on a flat track section without any joints (e.g., on a continuous rail), accelerometer signals are collected for 1 second, and the standard deviation of the signal for that section is calculated as the dynamic noise level. The dynamic noise primarily originates from the rolling vibration of the wheels and electromagnetic interference from the motor. The dynamic trigger threshold is set to a fixed multiple (e.g., 3 times) of the dynamic noise level, denoted as... .in, This is a preset multiple, usually a fixed value between 3 and 5, which is determined based on the signal-to-noise ratio test during equipment debugging.

[0036] During long-term operation of the equipment, dynamic noise levels may drift due to track wear, wheel aging, or environmental changes. The system automatically or manually repeats the dynamic noise level measurement process at preset calibration cycles (e.g., weekly or monthly) or after a certain cumulative operating time to update the levels. and This is to ensure that the trigger threshold always adapts to the current operating conditions.

[0037] It should be noted that the static and dynamic noise levels are calibrated separately in the embodiments, and a dynamic threshold is used to adapt to the background vibration during track operation, so as to avoid false triggering or missed triggering due to environmental vibration fluctuations.

[0038] S2: During operation, the current travel partition is located by the position encoder. When passing through the seam, the accelerometer is triggered to collect the real-time impact waveform, and the real-time pulse moment of the travel partition is collected at the same time.

[0039] S2.1: An absolute encoder is installed on the axle of the stacker crane's traveling wheel. The encoder rotates coaxially with the traveling wheel and outputs a Gray code or SSI signal proportional to the traveling distance. After decoding by the controller, the traveling position coordinates are obtained in real time. The absolute encoder has a power-off memory function, eliminating the need to return to zero every time power is restored, and has a sampling frequency of no less than 100Hz to meet the accuracy requirements of position tracking during high-speed movement.

[0040] Meanwhile, the piezoelectric film signal of each travel zone is connected to the control system through its own independent voltage comparator and the digital input channel of the data acquisition card. The voltage comparator threshold is fixed to the same value set at the factory calibration (e.g., 1.5V), and the amplification factor of the preamplifier circuit is consistent with that in S1.2, ensuring that the pulse triggering conditions acquired in real time are exactly the same as those at the factory calibration.

[0041] The data acquisition card is equipped with a hardware interrupt function. When a rising edge occurs on any digital input channel (i.e., the comparator output changes from low to high), the hardware interrupt immediately responds to the event, records the current time (based on the controller's high-precision timer, with an accuracy better than 1 millisecond), and automatically marks the partition number i to which the pulse belongs according to the channel number. This process does not require CPU software polling, avoiding timing errors caused by operating system scheduling delays.

[0042] Each time the stacker crane completes a storage / retrieval operation and returns to the standby position, the controller summarizes the pulse moments recorded by all partitions during that operation period (i.e., the current storage / retrieval cycle) to form the real-time pulse raw sequence for the current storage / retrieval cycle. The pulse times in this original sequence are arranged in chronological order, but have not yet been length-aligned or truncated with the factory reference sequence.

[0043] S2.2: This embodiment adopts a triggering strategy that combines position window activation with acceleration threshold comparison, specifically including the following sub-steps: S2.2.1: The controller continuously reads the position coordinates of the absolute encoder. And the travel coordinate ranges of each partition established in S1.1 Comparison. When The interval for positive entry into the i-th partition (i.e.) From less than Change to greater than or equal to When the stacker crane is moving in a positive direction (and the stacker crane is moving in a positive direction), the controller activates the data acquisition window for that partition. This window remains active until... Leaving this interval (i.e.) When the window is active, it closes. During window activation, the system monitors the accelerometer signal of that partition in real time.

[0044] S2.2.2: Once the instantaneous amplitude of the accelerometer signal exceeds... The analog comparator output goes high, and this high-level signal acts as a trigger pulse to drive the data acquisition card to perform the following two operations: First, the data acquisition card immediately latches the current position coordinates of the absolute encoder. .

[0045] The latching operation is performed directly by hardware with a delay of less than 1 microsecond, ensuring strict synchronization between the trigger point and the position coordinates.

[0046] Second, the data acquisition card reads a complete waveform from the front-end analog or digital buffer of the accelerometer signal at the exact same sampling rate (e.g., no less than 10kHz) and recording length as in S1.3. The waveform is recorded using the same pre-trigger buffering method as the factory calibration: the signal is continuously written to the circular buffer, and when the trigger signal arrives, a certain proportion (e.g., 10%) of the data before the trigger point is concatenated with the remaining proportion of data after the trigger point to form a complete waveform. The pre-trigger proportion is exactly the same as in S1.3 (e.g., both are 10%) to ensure the comparability of the real-time waveform and the factory waveform in phase alignment. This waveform is denoted as... .

[0047] S2.2.3: In actual operation, multiple impacts may occur when the traveling wheel passes over the seam (e.g., the rebound after the wheel flange collides with the edge of the seam), or the residual waves of adjacent vibrations may exceed the threshold again, resulting in multiple triggering events within the same data acquisition window. The handling rule for this in this embodiment is: each trigger is recorded as a separate shock wave. ,in This indicates the first [number] in this window. Each trigger is recorded, and the location coordinates at the time of triggering are marked. .

[0048] All waveforms are stored in chronological order and associated with the same partition and the same seam. During subsequent feature extraction, these multiple waveforms are compared with the factory waveform for overlap calculation, and the most unfavorable result (lowest overlap) is taken as the final feature of the seam in this loop to capture the most severe anomalies.

[0049] As can be seen, the present invention uses a dual determination of position window + dynamic threshold to ensure that data acquisition is only triggered when the stacker crane actually passes through the seam, thus avoiding false triggering by other vibration sources on the track.

[0050] S2.3: After each trigger recording is completed, the controller determines the position coordinates of the latch. According to the left-closed, right-open boundary assignment rules defined in S1.1, find the partition i to which the coordinate belongs and the corresponding joint number j. Then, record the real-time impact waveform. The corresponding factory standard waveform stored in S1.3 Associate the data and create waveform pairs for subsequent calculations.

[0051] Meanwhile, after each access cycle ends (i.e., after the stacker crane returns to the standby position), the controller retrieves the raw real-time pulse sequence summarized in S2.1. Retrieve all pulse moments recorded in each partition within the current access loop, arrange them in chronological order, and form the real-time pulse sequence for that partition. If no pulses are recorded for a certain partition during the entire cycle (possibly because the stacker crane did not pass through that partition, or the piezoelectric film did not generate a voltage of sufficient amplitude when it did), then Defined as an empty sequence.

[0052] As can be seen, this invention links two independent data streams, waveform acquisition and pulse acquisition, through a unified coordinate system (partition index and seam index), ensuring that the first feature pair and the second feature pair are based on the same spatiotemporal reference when constructing subsequent features.

[0053] S3: Combine the real-time waveform with the factory waveform to form the first feature pair and calculate the waveform overlap; combine the real-time pulse sequence with the factory reference sequence to form the second feature pair and calculate the time interval change rate of adjacent pulses.

[0054] S3.1: Acquiring real-time impact waveforms ,in For partition number, For the seam number, For the first in this partition After recording the waveform sequence number of the next trigger, the controller compares it with the corresponding factory standard waveform. The first feature pair is formed. To eliminate the influence of differences in accelerometer installation sensitivity and signal path gain drift, the two waveforms are first normalized.

[0055] The specific method of normalization is as follows: find the... and The maximum amplitude value (the absolute maximum value) at all sampling points is used to divide the amplitude of each sampling point by this maximum value, so that the maximum amplitude of each waveform after normalization is 1. This process ensures that the waveform amplitude reflects only the relative shape and is not affected by changes in absolute amplitude, thereby decoupling the difference in impact energy from the difference in waveform shape.

[0056] After normalization, the cross-correlation coefficient between the two waveforms is calculated as the waveform overlap. The cross-correlation coefficient is calculated by treating the two waveforms as discrete sequences of equal length. and ( ),in, Let be the number of sampling points. The numerator is the sum of the products of the corresponding points. The denominator is the geometric mean of the sum of squares of the energies of the two waveforms: The value of this ratio ranges from [0,1], where 1 indicates that the two waveforms are completely identical (linearly correlated and consistent after amplitude normalization), and 0 indicates that they are completely orthogonal (no similarity). Since the factory waveform and the real-time waveform are both generated by the same mechanical impact under healthy conditions, their normalized waveforms should be highly similar, and their cross-correlation coefficient should be close to 1. When the track sinks, the impact waveform will be distorted (e.g., pulse width increases, sidelobes appear, or multi-peak splitting occurs), resulting in a significant decrease in the cross-correlation coefficient.

[0057] For multiple real-time impact waveforms recorded in a single access cycle within the same partition and the same seam (e.g., multiple triggers due to residual vibration or rebound), calculate the cross-correlation coefficient between each waveform and the factory waveform to obtain... The minimum value among these coefficients is taken as the final waveform overlap of the joint in this cycle, denoted as . The rationale for selecting the minimum value is that even if only a single impact causes an anomaly, such as the distorted waveform resulting from a wheel flange hitting the edge of a seam, an early warning should be triggered. Taking the average or maximum value might mask individual abnormal waveforms. This design demonstrates early and sensitive detection of faults.

[0058] S3.2: After each access cycle, the controller acquires the real-time pulse sequence of each partition. and the reference pulse sequence recorded at the time of manufacture. Before calculating the rate of change of the time interval, both sequences need to be preprocessed.

[0059] First, determine Is it an empty sequence? If empty, it means the stacker crane did not pass through this zone in this cycle, or the piezoelectric film did not generate any pulse when it passed through. In this case, all time interval comparisons for this zone are skipped, and it will not participate in subsequent loosening warning judgments. Secondly, if Since it is not empty, further judgment is needed. Is the sequence empty? If the factory reference sequence is empty, it means that the partition was not traversed during factory calibration. It is typically located outside the start or end of the track. Therefore, the time interval comparison for this partition will be skipped in this and all subsequent access cycles. This is because the lack of reference data makes it impossible to determine if the track is loose.

[0060] When both sequences are not empty, they are aligned according to the absolute order of pulse occurrence; that is, the first pulse is aligned with the first pulse, the second pulse with the second pulse, and so on. Alignment is based on the pulse order, not specific time values, because the timing of the stacker crane passing through the same section may vary slightly in different cycles due to acceleration / deceleration strategies, but the pulse generation order should remain consistent under healthy conditions. If the number of pulses in the two sequences is inconsistent, the sequence with fewer pulses is used as the baseline, and the sequence with more pulses is truncated to the same length, retaining only the first pulse. pulses ( (Equal to the number of pulses of the smaller one). The purpose of truncation is to make the two sequences have the same length, so that the time interval between adjacent pulses can be calculated pairwise. After truncation, only pulse segments that exist in both sequences are compared, and redundant pulses are discarded. The reason for redundant pulses may be that the number of times the partition is passed through during real-time operation is more than the number of times it is in the factory cycle (e.g., the stacker crane stops or moves back and forth in this partition multiple times). However, considering that the standard cycle has covered all the number of times the joints are passed through during normal operation, redundant pulses are usually atypical operations, and truncation of them does not affect the detection of core abnormal patterns.

[0061] S3.3: For two aligned and truncated pulse sequences, calculate the time interval between adjacent pulses. Assume there are n pulses after the reference sequence is truncated. If there are pulses, then the number of intervals between adjacent pulses is... , record the The intervals are: Similarly, after real-time sequence truncation, there is pulses ( equal (because it has been truncated to the smaller value), calculate the real-time interval. Since the number of pulses in the two truncated sequences is equal, their interval numbers are... The range is the same.

[0062] If the number of pulses in any sequence is less than 2, no time interval can be formed. In this case, the controller skips the calculation of the interval change rate for this access cycle and does not participate in the subsequent loosening warning judgment. This situation usually occurs when the factory reference sequence contains only one pulse or the real-time sequence contains only one pulse, for example, when the stacker crane passes through the partition only once and the piezoelectric film is triggered only once.

[0063] For cases where a time interval can be formed, the controller takes the common interval number to calculate the current access cycle (denoted as the th interval). The first iteration of the loop) Rate of change over time intervals : This rate of change represents the degree of deviation of the real-time interval from the reference interval. If the reference interval... The value is zero (theoretically impossible, because adjacent pulses cannot occur simultaneously in the same piezoelectric film under healthy conditions, but to prevent division by zero errors caused by abnormal data), then it is defined as... It is 0.

[0064] It should be noted that when anchor bolts loosen, their vibration modes change, such as a decrease in natural frequency and a change in damping ratio, causing the time interval between output pulses of the piezoelectric film to drift relative to its healthy state. For example, a previously tight bolt experiences rapid vibration decay after impact, with short pulse durations and relatively uniform intervals between adjacent pulses; after loosening, the vibration duration lengthens, and the intervals between pulse sequences may increase or exhibit irregular fluctuations. By monitoring the rate of change in intervals, the degree of bolt loosening can be quantitatively assessed.

[0065] S4: When the waveform overlap is lower than the first threshold, output a track sinking warning; when the time interval change rate exceeds the second threshold for three consecutive access cycles, output a loose anchor bolt warning; the warning signal is uploaded to the monitoring terminal according to the zone code.

[0066] S4.1: After each access cycle, the controller obtains the final waveform overlap of the seam of each traversed travel partition. The overlap is compared with a pre-stored first preset threshold. This first preset threshold is an empirical value determined statistically through multiple normal impact tests during the equipment commissioning phase, typically ranging from 0.6 to 0.8, with the specific value depending on the processing precision of the track joint, the accelerometer installation method, and the on-site noise level. This threshold represents the upper limit of the permissible waveform distortion.

[0067] when When the value falls below a first preset threshold, the controller determines that track subsidence has occurred at the j-th track joint within the i-th travel zone. Specifically, the rails on both sides of the track joint experience relative vertical displacement. When the traveling wheel passes over it, the impact changes from a single-peak, symmetrical waveform to a multi-peak, asymmetrical, or broadened waveform, resulting in a significant decrease in the cross-correlation coefficient between the normalized waveform and the factory standard waveform. The greater the subsidence, the more severe the waveform distortion and the lower the cross-correlation coefficient.

[0068] Upon triggering, the controller immediately generates a subsidence warning message. This message contains at least the following information: warning type identifier (distinguishing between subsidence and loosening warnings), zone number, seam number (currently always 1, reserved for future expansion), trigger timestamp, and the current overlap value (optional). The message format uses structured encoding, such as fixed-length binary frames or compact ASCII strings, to meet the transmission requirements of low-speed industrial buses such as RS485.

[0069] S4.2: Judgment of anchor bolt looseness is based on the rate of change of time interval. Furthermore, the condition must be met for three consecutive access cycles to prevent false alarms caused by occasional disturbances. The specific operation is as follows: S4.2.1: For the i-th travel partition, the controller monitors the sequence number of each pulse interval. In one access loop (denoted as the c-th loop), if the partition is visited (determined by the position coordinates of the positive entry into the partition's coordinate range), and the coordinates can be effectively calculated... (i.e., the number of pulses is not less than 2), then each k corresponds to It is compared with a second preset threshold. The second preset threshold is a warning value determined by statistical analysis of the interval change rate under multiple normal cycles during factory calibration, and typically ranges from 0.15 to 0.25. This threshold represents the upper limit of the allowable pulse interval drift.

[0070] S4.2.2: Since the process from the initial loosening of anchor bolts to the occurrence of obvious failure is usually a gradual one, and the time interval variation rate may occasionally exceed the limit due to factors such as changes in stacker crane load and fluctuations in travel speed within a single cycle, this embodiment uses the cumulative criterion of three consecutive access cycles to improve the reliability of the early warning.

[0071] Specifically, the controller maintains a counter for each partition i and each pulse interval number k. For partition i, the following logic is executed after each access cycle: If the partition is not traversed in the current loop (determined by the position coordinates), or if it is traversed but cannot be calculated. (For example, if the number of pulses is less than 2), then the counters for all k corresponding to that partition are cleared, and the accumulation restarts in the next loop; if the partition has been traversed in this loop and can be calculated... Then check each k one by one: when When the value is greater than the second preset threshold, the counter corresponding to k is incremented by 1; when... If the value is not greater than the threshold, the counter corresponding to k is reset to zero. When the counter value corresponding to a certain k reaches 3, it indicates that the partition has been visited in three consecutive access cycles, and the three cycles... If all values ​​exceed the second preset threshold, a loosening warning is triggered.

[0072] S4.2.3: Once the conditions for the above three consecutive cycles are met, the controller determines that the anchor bolts in the i-th travel zone are loose and generates a loosening warning message. This message must contain at least the following information: warning type identifier (loosening warning), zone number i, trigger timestamp, and pulse interval sequence number k that caused the warning (optional, to facilitate analysis of specific vibration beat changes).

[0073] It should be noted that after the anchor bolts loosen, the preload between the bolts and the column decreases, and the stiffness and damping of the vibration transmission path change, causing a systematic drift in the pulse sequence interval output by the piezoelectric film, and this drift repeats in multiple access cycles.

[0074] S4.2.4: If, during the accumulation process, the counter is reset to zero because the partition has not been visited or R cannot be calculated, then three consecutive cycles that meet the conditions must be accumulated again before an alert can be triggered. For example, if a partition meets the conditions in the first and second cycles but is not visited in the third cycle, the counter is reset to zero. Even if the conditions are met consecutively in the fourth, fifth, and sixth cycles, accumulation can only begin again from the fourth cycle, and the alert will be triggered at the end of the sixth cycle (assuming all four cycles are met). This design avoids false triggers due to sporadic missing counts and ensures that true persistent faults are not missed when stacker crane operations are discontinuous.

[0075] As can be seen, the present invention uses a cumulative criterion of three consecutive cycles to effectively filter out single, occasional interferences, such as temporary load changes and instantaneous impacts caused by foreign objects on the track, thereby improving the reliability and accuracy of the early warning. Furthermore, by counting independently for each pulse interval number k, the different effects of loosening on different vibration beats (e.g., the first rebound and the second rebound) can be detected, providing richer fault characteristics.

[0076] S4.3: The generated subsidence warning messages and loosening warning messages are uniformly encoded within the controller according to the travel zone number. The encoding method adopts a compact binary format.

[0077] The encoded message is sent to the monitoring terminal via the RS485 bus. The RS485 bus uses half-duplex differential transmission, has strong anti-interference capabilities, and is suitable for long-distance communication in industrial environments (up to 1200 meters). The controller acts as the master, and the monitoring terminal acts as the slave, using a master-slave query-response or active reporting communication protocol. This embodiment uses the active reporting method: once an alarm message is generated, the controller immediately sends it to the bus, and the monitoring terminal continuously listens and parses it.

[0078] After receiving the message, the monitoring terminal (usually an industrial computer or touchscreen) parses and displays it according to the partition number. The display method uses a graphical interface: the floor plan of the automated warehouse is used as the background, and each travel partition corresponds to a color block or icon. When a subsidence warning is received, the corresponding partition is highlighted in yellow; when a loosening warning is received, it is highlighted in red; if two warnings exist for the same partition at the same time (e.g., subsidence and loosening occur concurrently), the one with higher priority is displayed in red (loosening warnings are usually more serious than subsidence warnings because they may cause overall structural instability). In addition, the terminal also records the warning history, generates reports, and can be linked to audible and visual alarms.

[0079] like Figure 2 As shown, the present invention also provides an online health status monitoring and fault early warning system for automated storage and retrieval systems (AS / RS), comprising: The zone calibration module divides the travel track into multiple travel zones according to the track joints, and records the factory impact waveform of the joint in each travel zone and the factory pulse reference time sequence of the corresponding anchor bolt piezoelectric film. The joint acquisition module locates the current travel zone through the position encoder during operation. When passing through the seam, it triggers the accelerometer to acquire the real-time impact waveform and simultaneously acquires the real-time pulse moment of the travel zone. The feature calculation module combines the real-time waveform and the factory waveform into a first feature pair and calculates the waveform overlap; it also combines the real-time pulse sequence and the factory reference sequence into a second feature pair and calculates the rate of change of the time interval between adjacent pulses. The layered early warning module outputs a track sinking warning when the waveform overlap is below the first threshold; and outputs a loose anchor bolt warning when the time interval change rate exceeds the second threshold for three consecutive access cycles; the warning signals are uploaded to the monitoring terminal according to the zone code.

[0080] The system also includes one or more processors and memory.

[0081] 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 an online health status monitoring and fault early warning method for automated storage and retrieval systems according to the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0082] 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 online health status monitoring and fault early warning method for automated warehouse equipment of the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0083] 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.

[0084] 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.

[0085] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if necessary, the program can be implemented in assembly or machine language.

[0086] In any case, the language can be either compiled or interpreted.

[0087] Furthermore, for this purpose, the program can run on a programmed application-specific integrated circuit.

[0088] 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.

[0089] 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.

[0090] 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.

[0091] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0092] 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.

[0093] 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 online monitoring of the health status and fault early warning of automated storage and retrieval system equipment, characterized by: include: The travel track is divided into multiple travel zones according to the track joints, and the factory impact waveform of the joint in each travel zone and the factory pulse reference time sequence of the corresponding anchor bolt piezoelectric film are recorded. During operation, the current travel zone is located by the position encoder. When passing through the seam, the accelerometer is triggered to collect the real-time impact waveform, and the real-time pulse moment of the travel zone is collected at the same time. The real-time waveform and the factory waveform are combined to form the first feature pair and the waveform overlap is calculated; the real-time pulse sequence and the factory reference sequence are combined to form the second feature pair and the time interval change rate of adjacent pulses is calculated. When the waveform overlap is below the first threshold, a track sinking warning is output; when the time interval change rate exceeds the second threshold for three consecutive access cycles, a loose anchor bolt warning is output; the warning signal is uploaded to the monitoring terminal according to the zone code.

2. The method for online health status monitoring and fault early warning of automated storage and retrieval system equipment as described in claim 1, characterized in that: The method of dividing the travel track into multiple travel zones according to the track joints includes: Using each track joint as a dividing point, the track segment between two adjacent joints is defined as a travel zone; Each travel zone uses a left-closed, right-open coordinate interval representation method, and establishes a mapping table between the travel coordinate interval of each travel zone and the joint number; The travel partition is activated only when the stacker crane enters the partition coordinate range in the forward direction, and is not activated when moving in the reverse direction.

3. The method for online monitoring of the health status and fault early warning of automated storage and retrieval system equipment as described in claim 2, characterized in that: The factory impact waveform recorded for each stroke zone of the joint includes: An impact hammer was used to simulate the impact of a stacker crane's traveling wheels passing through a joint, and the top of the rail at the rail joint in each travel zone was struck sequentially. The standard impact waveform of the joint is collected by an accelerometer installed near the stacker crane's traveling wheels. The sampling rate is set to a high sampling rate sufficient to capture impact details. The recording duration covers the process of the impact response decaying to the noise level. A pre-trigger buffer method is used to ensure that the trigger point is located at a set proportion position in the recording window.

4. The method for online monitoring of the health status and fault early warning of automated storage and retrieval system equipment as described in claim 2, characterized in that: The recorded factory pulse reference time series includes: In each shelf column corresponding to each travel zone, select the anchor bolt that is closest to the horizontal distance of the track joint within the travel zone, and attach a piece of piezoelectric film to the top of the bolt. The output of each piezoelectric film is independently connected to a voltage comparator, and the output of the comparator is connected to different digital input channels of the data acquisition card. During the factory calibration of the stacker crane, no-load operation is performed using the standard access cycle. The rising edge of the comparator output of each channel is recorded, and the pulse occurrence order is used to form the factory reference time sequence. For unexplored travel partitions, the baseline sequence is empty.

5. The method for online monitoring of the health status and fault early warning of automated storage and retrieval system equipment as described in claim 1, characterized in that: The step of triggering the accelerometer to collect real-time impact waveforms when passing through a seam includes: An absolute encoder is installed on the traveling wheel axle of the stacker crane to read the traveling position coordinates in real time; When the coordinate enters the coordinate range of a certain travel partition in the positive direction, the data acquisition window of the corresponding travel partition is activated and closed when it leaves the coordinate range; During the window activation period, the accelerometer signal is compared with a dynamic trigger threshold, which is obtained by multiplying the dynamic noise level of the accelerometer when the stacker crane is traveling on a flat track section by a fixed factor. Once the signal amplitude exceeds the dynamic trigger threshold, the current coordinates are latched and the real-time impact waveform is recorded with the same sampling rate and recording length as the factory impact waveform, while the same pre-trigger ratio is used as when recording the factory impact waveform. When triggered multiple times within the same window, each waveform is recorded independently and its coordinates are marked.

6. The method for online health status monitoring and fault early warning of automated storage and retrieval system equipment as described in claim 5, characterized in that: The simultaneous acquisition of real-time pulse moments of the travel partition includes: The piezoelectric film signal of each travel zone is recorded by hardware interrupts through its respective voltage comparator and digital input channel of the data acquisition card, and the zone number is automatically marked according to the channel. When the stacker crane completes each storage and retrieval operation and returns to the standby position, it summarizes the real-time pulse moments of all travel partitions during the corresponding segment of operation to form the original real-time pulse sequence of this storage and retrieval cycle. Extract all pulse moments from each travel partition from the original sequence and arrange them into a real-time pulse sequence in chronological order. If a travel partition has no pulses, the sequence is empty.

7. The method for online health status monitoring and fault early warning of automated storage and retrieval system equipment as described in claim 1, characterized in that: The step of forming a first feature pair from the real-time waveform and the factory-issued waveform and calculating the waveform overlap includes: Each real-time impact waveform belonging to the same stroke zone and the same joint is normalized to the corresponding factory standard impact waveform so that the maximum amplitude of the two is the same. Calculate the cross-correlation coefficient of the two normalized waveforms and use it as the waveform overlap. If multiple real-time impact waveforms are recorded for the same travel partition and the same seam within a single access loop, the minimum value of all overlaps is taken as the final overlap of the travel partition in this loop.

8. The method for online health status monitoring and fault early warning of automated storage and retrieval system equipment as described in claim 1, characterized in that: The process of composing the second feature pair and calculating the rate of change of the time interval between adjacent pulses includes: After each access cycle ends, for each stroke partition, if the real-time pulse sequence is not empty and the factory reference sequence is not empty, the two are aligned according to the order of pulse appearance, and the one with fewer pulses is truncated. Calculate the time interval between adjacent pulses in the factory reference sequence and the time interval between adjacent truncated pulses in the real-time pulse sequence, respectively. If the number of pulses in any sequence is less than two, then skip the calculation of the rate of change for this stroke partition; For intervals that can form time intervals, take the common interval number and calculate the time interval change rate of each interval, which is the absolute value of the difference between the real-time interval and the reference interval divided by the reference interval. If the baseline interval is zero, then the rate of change is defined as zero.

9. The method for online health status monitoring and fault early warning of automated storage and retrieval system equipment as described in claim 1, characterized in that: When the final waveform overlap of a certain travel partition is lower than the first preset threshold under a certain access cycle, it is determined that track sinking has occurred at the track joint within the travel partition, and a sinking early warning message containing the partition number and the joint number is generated. For each travel partition, monitor each pulse interval number. If there is an interval number that causes the partition to be entered in the positive direction in three consecutive access cycles and the change rate of the corresponding three time intervals is greater than the second preset threshold, then it is determined that the anchor bolts in the travel partition are loose and a loosening warning message is generated. If any of the loops in the cycle is not entered in the forward direction or the rate of change cannot be calculated, the continuous counting is interrupted and the accumulation must be restarted. Subsidence warning messages and loosening warning messages are encoded by zone and sent to the monitoring terminal via the communication bus. The monitoring terminal displays visual identifiers for different warning types by zone, and displays the higher level when both exist.

10. An online health status monitoring and fault early warning system for automated storage and retrieval systems (AS / RS), based on the online health status monitoring and fault early warning method for AS / RS as described in any one of claims 1 to 9, characterized in that: Also includes: The zone calibration module divides the travel track into multiple travel zones according to the track joints, and records the factory impact waveform of the joint in each travel zone and the factory pulse reference time sequence of the corresponding anchor bolt piezoelectric film. The joint acquisition module locates the current travel zone through the position encoder during operation. When passing through the seam, it triggers the accelerometer to acquire the real-time impact waveform and simultaneously acquires the real-time pulse moment of the travel zone. The feature calculation module combines the real-time waveform and the factory waveform into a first feature pair and calculates the waveform overlap; it also combines the real-time pulse sequence and the factory reference sequence into a second feature pair and calculates the rate of change of the time interval between adjacent pulses. The layered early warning module outputs a track sinking warning when the waveform overlap is below the first threshold; and outputs a loose anchor bolt warning when the time interval change rate exceeds the second threshold for three consecutive access cycles; the warning signals are uploaded to the monitoring terminal according to the zone code.

Images