Intelligent archiving method and system for unmanned archive room

Abstract

The present application relates to the technical field of data management, in particular to a kind of intelligent archiving method and system of unmanned archives room.The method sets a tentative pick-and-place action process for target archive location, and real-time monitors its feedback motor current response sequence and motor speed response sequence, obtains instantaneous input power sequence and instantaneous dissipation power sequence under local time period, solves the allocation mode of system energy from input band to dissipation band under local time period, and obtains the matching result under each time window.By calculating the difference characteristics of continuous matching results and comparing with threshold, the contact event state of target archive location can be judged.The present application realizes the effective distinction of two resistance states in the process of picking and placing archives by carrying out frequency domain decoupling analysis on the response information under the tentative pick-and-place action process and difference analysis on the energy allocation path under the process.

Description

Technical Field

[0001] This invention relates to the field of data management technology, specifically to an intelligent archiving method and system for an unmanned archive warehouse. Background Technology

[0002] In the high-density storage environment of unmanned archive warehouses, robotic systems are responsible for the automated storage and retrieval of archive boxes. Due to the long-term effects of their own weight and stacking pressure, and the cyclical changes in temperature and humidity, the box materials undergo deformations such as creep, expansion, or warping. Simultaneously, the surface coatings or labels of adjacent boxes may adhere due to pressure and environmental factors. These factors collectively complicate the actual contact state of the archive boxes within the shelves. When a robot attempts to retrieve a box, its movement encounters two completely different types of resistance. The first type of resistance is the contact resistance caused by the static friction between the boxes and surface adhesion; applying appropriate force can separate the box and allow it to be successfully retrieved. The second type of resistance is the rigid jamming formed by the geometric deformation of the box (such as expansion or warping) and the shelf or adjacent boxes. In this case, the contact system forms a stable mechanical constraint; further force will not produce effective displacement but may instead cause the box to be crushed or damage the robot's end effector.

[0003] Current technologies for analyzing the status of file boxes in unmanned archives typically rely on computer vision. However, computer vision can only observe the visible external outline of the file boxes and cannot detect invisible surface adhesions or jamming tendencies caused by internal stress. If force feedback sensors are used for auxiliary judgment, inaccurate identification can occur because a single threshold cannot distinguish between the two resistance states mentioned above. A high threshold may apply destructive external force to the "geometric jamming" state, posing a safety risk; while a low threshold may fail to identify the "static friction breakthrough" state, which requires a large separation force, leading to frequent task termination and reduced system efficiency. Summary of the Invention

[0004] To address the technical problem that existing technologies cannot effectively distinguish between geometric jamming states and static friction breakthrough states between file boxes, the present invention aims to provide an intelligent archiving method and system for unmanned archive storage, the specific technical solution of which is as follows:

[0005] This invention proposes an intelligent archiving method for unmanned archive storage facilities, the method comprising:

[0006] The file retrieval and placement robot is controlled to perform a trial retrieval and placement action at the target file location, and the motor current response sequence and motor speed response sequence are collected during the trial retrieval and placement action.

[0007] Based on the local sequences of the motor current response sequence and the motor speed response sequence within a preset time window, the instantaneous input power sequence and instantaneous dissipated power sequence for each time window are obtained; spectral analysis is performed on the instantaneous input power sequence and instantaneous dissipated power sequence to obtain the instantaneous input power spectrum vector and instantaneous dissipated power spectrum vector.

[0008] The instantaneous input power spectrum vector and the instantaneous dissipated power spectrum vector are matched and analyzed to obtain the matching results under each time window; the difference features of the matching results under all adjacent time windows are statistically analyzed, and the difference features are compared with a preset threshold to determine the contact event state of the target file position; the contact event state includes geometric jamming, static friction rate breakthrough, and uncertain events.

[0009] Furthermore, the method for obtaining the instantaneous input power sequence includes:

[0010] At each moment, the current is multiplied by the motor torque constant to obtain the torque, and the torque is multiplied by the motor speed to obtain the instantaneous input power;

[0011] The instantaneous input power at each moment within the time window constitutes the instantaneous input power sequence.

[0012] Furthermore, the method for obtaining the instantaneous power dissipation sequence includes:

[0013] At each moment, the instantaneous input power and motor speed are substituted into the instantaneous power balance equation of the rotating system to obtain the instantaneous dissipated power;

[0014] The instantaneous power dissipation at each moment within the time window constitutes the instantaneous power dissipation sequence.

[0015] Further, obtaining the instantaneous input power spectrum vector and the instantaneous dissipation power spectrum vector includes:

[0016] Either the instantaneous input power sequence or the instantaneous dissipated power sequence is taken as the sequence to be analyzed. The sequence to be analyzed is processed using the Fast Fourier Transform, and the resulting power spectrum is downsampled in multiple preset frequency bands to obtain the corresponding power spectrum vector.

[0017] Furthermore, the matching analysis of the instantaneous input power spectrum vector and the instantaneous dissipation power spectrum vector includes:

[0018] A matching matrix is ​​constructed based on the instantaneous input power spectrum vector and the instantaneous dissipation power spectrum vector. The rows of the matching matrix represent the positions of elements in the instantaneous input power spectrum vector, the columns represent the positions of elements in the instantaneous dissipation power spectrum vector, and the element values ​​in the matrix are the absolute values ​​of the differences between two elements at corresponding positions.

[0019] The matching matrix is ​​optimized using the Hungarian algorithm to obtain an instantaneous energy distribution matrix of the same size as the matching matrix. The instantaneous energy distribution matrix is ​​a binary matrix. The instantaneous energy distribution matrix is ​​used as the matching result.

[0020] Furthermore, the difference feature is the Flobenius norm between two adjacent matching matrices.

[0021] Furthermore, the determination of the contact event status of the target file location includes:

[0022] The differential features are summed to obtain the overall difference. The overall difference is then normalized to obtain the normalized path change.

[0023] If the change in the normalized path is less than a preset first threshold, it is determined to be a geometrically stuck state; if the change in the normalized path is greater than a preset second threshold, it is determined to be a static friction force breakthrough state; other cases are determined to be an uncertain state, where the first threshold is less than the second threshold.

[0024] Furthermore, the method also includes a threshold update process, which includes:

[0025] The location of all archives in the unmanned archive storage is divided into multiple local areas. For each local area, the number of event samples diagnosed as geometric jamming and static friction breakthrough is counted. If the number of event samples is less than a preset threshold, the threshold is not updated. If the number of event samples is not less than the preset threshold, the normalized path change of the geometric jamming event samples is counted to obtain the updated first threshold, and the normalized path change of the static friction breakthrough event samples is counted to obtain the updated second threshold.

[0026] Furthermore, the method also includes a file box status analysis process for the target file location, including:

[0027] For the target file location, all event samples judged as static friction breakthroughs are statistically analyzed within a preset analysis time period, and the sequence of normalized path changes corresponding to these events is used as the analysis sequence. If the length of the analysis sequence is less than a preset length threshold, the file box status analysis process is not performed. If the length of the analysis sequence is greater than or equal to the preset length threshold, a straight line is fitted to the analysis sequence. If the slope of the fitted straight line is greater than a preset slope threshold, the file box status is determined to need maintenance; otherwise, it is determined not to need maintenance.

[0028] This invention proposes an intelligent archiving system for an unmanned archive repository, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the steps of the intelligent archiving method for an unmanned archive repository.

[0029] The present invention has the following beneficial effects:

[0030] This invention sets up a trial pick-and-place action process for the target file location and monitors the feedback motor current response sequence and motor speed response sequence in real time. These two response information can characterize the force feedback information of the current trial pick-and-place action process. Since a single threshold analysis cannot accurately identify the contact event state of the target file location, this embodiment of the invention performs local analysis, conducting a detailed analysis of the entire process to obtain the instantaneous input power sequence and instantaneous dissipation power sequence for the local time period. The distribution of system energy from the input frequency band to the dissipation frequency band under the local time period is solved to obtain the matching result for each local time period, i.e., each time window. This matching result can decouple the energy input to dissipation from the frequency domain through matching. The matching results under continuous time windows can describe the law of energy distribution path structure evolution over time. For the geometric jamming state, its energy distribution path is time-invariant; while for the static friction breakthrough state, because this state can continuously apply force to remove the file, it has a significant change in energy distribution path compared to the geometric jamming state. Therefore, by calculating the difference characteristics and comparing them with the threshold, the contact event state of the target file location can be determined. This invention achieves effective differentiation of two resistance states during the file retrieval and placement process by performing frequency domain decoupling analysis on the response information during the trial retrieval and placement process and performing difference analysis on the energy distribution path during the process. Attached Figure Description

[0031] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art 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.

[0032] Figure 1 This is a flowchart illustrating an intelligent archiving method for an unmanned archive warehouse, as provided in one embodiment of the present invention. Detailed Implementation

[0033] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent archiving method and system for an unmanned archive warehouse proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0035] The following description, in conjunction with the accompanying drawings, details the specific scheme of the intelligent archiving method and system for an unmanned archive warehouse provided by the present invention.

[0036] This invention addresses the scenario of unmanned retrieval and placement of archives in an unmanned archive repository. Unmanned archive repositories are primarily used in fields requiring massive storage, efficient traceability, and long-term secure preservation of physical archives, such as compliant contracts of financial institutions, long-term medical records of medical institutions, core corporate documents, and valuable documents of cultural institutions. Through highly automated and intelligent management, it replaces traditional manual searching and handling, significantly improving storage density, retrieval efficiency, and archive security while meeting stringent regulatory requirements. In the unmanned archive repository, the robot's retrieval and placement process is a fully automated and precise logistics operation. Upon receiving an instruction, the management system dispatches a track-based shuttle robot or AGV to the designated shelf location, retrieves the archive storage box using a robotic arm or forklift mechanism, and transports it to the workstation via a lift and path network. This invention embodiment is based on a process where, before the robot retrieves and places archives, a trial retrieval and placement action is initiated to assess the contact status of the target archive location.

[0037] Please see Figure 1 The diagram illustrates a flowchart of an intelligent archiving method for an unmanned archive repository according to an embodiment of the present invention. The method includes:

[0038] Step S1: Control the file retrieval robot to perform a trial retrieval action at the target file location, and collect the motor current response sequence and motor speed response sequence during the trial retrieval action.

[0039] In this embodiment of the invention, the target file location is any file location that needs to be retrieved or placed. That is, the method proposed in this embodiment of the invention is applicable to every file location in an unmanned archive. This embodiment of the invention only uses one file location as the target file location for illustration.

[0040] In order to assess the contact event status by analyzing the feedback information when the robot picks up and places files, this embodiment of the invention sets a standardized short-term propulsion action as a trial pick-up and place action. By actively and repeatedly stimulating this process and collecting the corresponding response data, the original dynamic response data reflecting the contact status of the target file box at the target file location is obtained.

[0041] It should be noted that the probing and placement action should be a completely standardized process, meaning the same process should be performed for different file locations. Therefore, to ensure that the difference in each probing action stems only from the contact state of the file box itself, rather than changes in test conditions, the system must first define and load a set of standardized probing parameters before executing the advancement action. In this embodiment of the invention, the probing parameters specifically include: probing duration, set to 0.15 seconds, which is sufficient to fully capture the transient process from contact to stabilization or breakthrough; data sampling frequency, set to 2 kHz, to capture high-frequency dynamic details in motor current and speed during the probing and placement action through a high sampling rate; and command angular velocity, set to a constant value, which in this embodiment is set to 0.01 radians / second. This low-speed command aims to start the system in a quasi-static state, thereby more clearly exposing the nature of the contact resistance. In addition to the standardized trial parameters used for probing the pick-up and place actions, various prior parameters used in subsequent processes in this embodiment of the invention also need to be pre-loaded. For example, the motor torque constant is a scalar coefficient that converts motor current into output torque; the system equivalent moment of inertia represents the magnitude of the inertia of the motor rotor and transmission mechanism. These two prior parameters can be directly used in subsequent calculations. In other specific implementations of this invention, the standardized trial parameters can be fine-tuned according to the specific implementation scenario to ensure that the parameters no longer change after adjustment.

[0042] In this embodiment of the invention, the moment the robot's end effector confirms contact with the target file box, the controller sends a command to the servo motor responsible for linear propulsion, instructing it to execute with the standardized trial parameters described above, and simultaneously recording the motor current response sequence and the click speed response sequence. It should be noted that the two sequences are of equal length and aligned in time. After collecting these two types of response information, all information can be encapsulated into a structured, self-contained data unit for subsequent processing. That is, after the trial action ends, the processor creates a single trial record, which includes the identifier of this trial event, the timestamp of the event, the standardized trial parameter set, the motor current response sequence, and the motor speed response sequence. Once constructed, this single trial record can be used as input for feature analysis in subsequent steps.

[0043] It should be noted that the two types of response data can be collected directly using the robot's built-in sensor modules, which are well-known techniques in the field and will not be elaborated upon here.

[0044] Step S2: Based on the local sequences of the motor current response sequence and the motor speed response sequence within a preset time window, obtain the instantaneous input power sequence and instantaneous dissipated power sequence for each time window; perform spectral analysis on the instantaneous input power sequence and instantaneous dissipated power sequence to obtain the instantaneous input power spectrum vector and instantaneous dissipated power spectrum vector.

[0045] Existing technologies that distinguish between two resistance states using a single force feedback threshold are limited to simplifying the dynamic contact process into a static scalar, leading to inaccurate judgments. This invention considers that the difference between the two resistances lies not in the magnitude of the resistance, but in the different evolutionary characteristics of the robot's internal energy distribution during the force application process. Therefore, step S2, through local analysis and frequency domain feature extraction, facilitates subsequent steps in effectively analyzing the evolutionary characteristics of energy distribution during the trial pick-and-place action.

[0046] In the geometrically locked state, the file box and the shelf form a stable mechanical structure, and its energy transfer mode remains unchanged for a short period of time. However, in the static friction breakthrough state, the microscopic adhesion points on the surface of the file box undergo a rapid dynamic evolution from "stickiness" to "fracture," causing the energy transfer mode to be reconstructed on a millisecond scale. By observing the changes in the energy distribution path over time with high temporal resolution, the essential difference between these two states can be directly captured. Therefore, this embodiment of the invention performs local analysis of the entire process, dividing the complete trial-and-place action time period into multiple sub-time periods by setting time windows, thereby achieving dynamic segmentation of details.

[0047] In order to analyze the energy transfer process, in each time window, the embodiments of the present invention obtain the input power sequence and the instantaneous dissipated power sequence according to the local sequence corresponding to the response sequence. The input power sequence and the instantaneous dissipated power sequence have the same length, which is the time length of the time window. Each element in the sequence represents the input power and dissipated power generated by the motor at a certain moment.

[0048] In a geometrically locked state, since the mechanical system forms a stable mechanical structure, the energy transfer path remains fixed for a short period of time. The input energy is mainly converted into structural strain energy and dissipated in a relatively stable manner (such as low-frequency heat conduction). In the frequency domain, the energy should exhibit the characteristic that the input and dissipation are at the same frequency band height.

[0049] For the static friction breakthrough state, the microscopic adhesion points undergo a dynamic evolution of "viscosity-fracture," accompanied by complex processes such as transient impact, high-frequency vibration, and acoustic emission. The energy transfer path is rapidly reconfigured, causing part of the input low-frequency mechanical energy to be converted into high-frequency vibrational energy and dissipated, i.e., significant cross-frequency energy transfer occurs.

[0050] Therefore, to reflect the difference between the two states, this embodiment of the invention further analyzes the data in the frequency domain to obtain the instantaneous input power spectrum vector and the instantaneous dissipation power spectrum vector. Because time-series analysis can only reflect the total energy change and cannot show the energy transfer mechanism at different frequency components, the instantaneous input power spectrum vector is used to reflect the distribution of energy applied by the motor at different frequencies, and the instantaneous dissipation power spectrum vector is used to reflect the distribution of energy dissipated in the robot drive system in the frequency domain as heat, vibration, and sound waves when subjected to force. The characteristics of these two distribution features can be applied to the subsequent diagnostic process in step S3, reflecting the energy generation mechanism generated in each time window.

[0051] Preferably, in this embodiment of the invention, the method for obtaining the instantaneous input power sequence includes:

[0052] At each moment, the current is multiplied by the motor torque constant to obtain the torque, and the torque is multiplied by the motor speed to obtain the instantaneous input power. It should be noted that the motor torque constant is a priori parameter pre-loaded in this embodiment of the invention. It is a priori constant well known to those skilled in the art. The power input during motor operation can be directly obtained using the current information through this constant. The specific technique is well known to those skilled in the art and will not be elaborated here.

[0053] The instantaneous input power at each moment within the time window constitutes the instantaneous input power sequence.

[0054] Preferably, in this embodiment of the invention, the method for obtaining the instantaneous power dissipation sequence includes:

[0055] At each instant, the instantaneous input power and motor speed are substituted into the instantaneous power balance equation of the rotating system to obtain the instantaneous power dissipation. It should be noted that the instantaneous power balance equation of the rotating system is a well-known technique, also known as the electromechanical system power equation, and in servo control systems, it is referred to as a power loss observer or heat loss calculation model. The equation has the following form:

[0056] ;in Let be the instantaneous power dissipation at time t. Let be the input power at time t; The equivalent rotational inertia of the system, These are prior parameters that are preloaded in the embodiments of the present invention, and will not be described in detail here. Let t be the motor speed (angular velocity).

[0057] In this equation, Expressed as the instantaneous kinetic energy of the system, obtained through differentiation. This is the rate of change of instantaneous kinetic energy. Subtracting the power used to change the kinetic energy from the input power leaves all the power dissipated in the system in the form of heat, etc. This dissipated power includes all forms of loss such as copper loss (heat generated by winding resistance), iron loss (eddy current and hysteresis losses in the iron core), and mechanical aerodynamic losses. The specific content of this equation is a well-known technique to those skilled in the art and will not be elaborated upon here.

[0058] The instantaneous power dissipation at each moment within the time window constitutes the instantaneous power dissipation sequence.

[0059] Preferably, in this embodiment of the invention, obtaining the instantaneous input power spectrum vector and the instantaneous dissipation power spectrum vector includes:

[0060] Either the instantaneous input power sequence or the instantaneous dissipated power sequence is taken as the sequence to be analyzed. The sequence is processed using a Fast Fourier Transform (FFT), and the resulting power spectrum is downsampled across multiple preset frequency bands to obtain the corresponding power spectrum vector. It should be noted that in this invention, the preset frequency bands are set to 128, allowing for the uniform selection of 128 frequency points within a frequency range. The resulting instantaneous input power spectrum vector and instantaneous dissipated power spectrum vector are identical, with each element representing the energy at the corresponding frequency on the power spectrum. Frequency domain analysis using FFT can clearly characterize details such as which frequencies the dissipated power is concentrated at and the magnitude of dissipated energy within the controller bandwidth.

[0061] Step S3: Perform matching analysis on the instantaneous input power spectrum vector and the instantaneous dissipation power spectrum vector to obtain the matching results under each time window; count the difference features of the matching results under all adjacent time windows, compare the difference features with the preset threshold, and determine the contact event state of the target file position; the contact event state includes geometric jamming, static friction rate breakthrough, and uncertain events.

[0062] In step S2, the frequency domain response information generated during the trial pick-up and release operation is obtained. Further, the instantaneous input power spectrum vector and the instantaneous dissipated power spectrum vector need to be integrated into a feature data that can characterize the energy transfer characteristics at that instant. Therefore, this embodiment of the invention uses vector matching to determine the matching result that can characterize the energy transfer through element-by-element matching.

[0063] Furthermore, the differences in matching results between consecutive time windows can be analyzed to characterize the dynamic changes in energy transfer. A large difference indicates that the process of attempting to pick up and place objects is not in a steady state, consistent with the characteristics of a static friction breakthrough state; conversely, a small difference indicates a geometric jamming state. Therefore, the differences can be compared with preset thresholds to determine the contact event state of the target file location.

[0064] Preferably, in this embodiment of the invention, considering that simple vector distance or correlation coefficients in the matching process can only measure the difference in capabilities within the same frequency band or the entire frequency band, and cannot capture cross-band energy transfer (i.e., the energy of low-frequency input is transferred and dissipated in high-frequency vibrations), this embodiment of the invention adopts a matrix construction method. A matching matrix is ​​constructed based on the instantaneous input power spectrum vector and the instantaneous dissipated power spectrum vector. The rows of the matching matrix represent the positions of elements in the instantaneous input power spectrum vector, and the columns represent the positions of elements in the instantaneous dissipated power spectrum vector. The element values ​​in the matrix are the absolute values ​​of the differences between two elements at corresponding positions. The matching matrix is ​​then optimized using the Hungarian algorithm to obtain an instantaneous energy allocation matrix of the same size as the matching matrix. This instantaneous energy allocation matrix is ​​a binary matrix; the instantaneous energy allocation matrix is ​​used as the matching result. It should be noted that the Hungarian algorithm is a combinatorial optimization algorithm well-known to those skilled in the art, used to solve minimum weight matching in allocation problems. In this embodiment of the invention, the goal of this algorithm is to find a one-to-one allocation scheme with the lowest total mismatch degree from all possible energy transfer paths. That is, in the final binary matrix, elements with a value of 1 represent the one-to-one correspondence between energy from each input frequency band to each dissipation frequency band within the current time window, while elements with a value of 0 represent that there is no correspondence under the current frequency band combination. Finally, a set of binary matrices can be obtained from consecutive time windows. The specific algorithm is a technique well-known to those skilled in the art and will not be described in detail here.

[0065] Preferably, in this embodiment of the invention, the difference feature is the Flobenius norm between two adjacent matching matrices. The Flobenius norm is used to measure the overall difference between two matrices, and it is calculated by taking the square root of the square of the differences between all elements of the matrices. This is a technique well known to those skilled in the art and will not be elaborated here.

[0066] Furthermore, since each adjacent time window can yield a difference feature, resulting in several difference features, in order to quantify the contact event state identification process, this embodiment of the invention accumulates the difference features to obtain the overall difference, and then normalizes the overall difference to obtain the normalized path change.

[0067] In this embodiment of the invention, the normalization operation can be selected by range standardization. A database is constructed through a large number of data samples, and the maximum and minimum values ​​of the overall differences in the database are statistically analyzed to achieve normalization. The specific normalization method is a well-known technical means in the art and will not be described in detail here.

[0068] If the change in the normalized path is less than a preset first threshold, it indicates that the energy transfer is relatively stable, and the condition is judged as a geometric dead state. If the change in the normalized path is greater than a preset second threshold, it indicates that the energy transfer is chaotic, and the condition is judged as a static friction breakthrough state. Other conditions are judged as uncertain states. The first threshold is less than the second threshold.

[0069] It should be noted that, for the first and second thresholds, this embodiment of the invention includes an initial threshold calibration process to address the problem that the system cannot set effective discrimination thresholds during initial deployment due to a lack of historical data. Upon system startup, if empty historical diagnostic data is detected, or after receiving a "recalibrate" command from the administrator, the following operations are performed:

[0070] (1) Calibrate the geometric dead baseline: The system commands the robot to move to a known target with rigid constraints, such as the metal column of the shelf, and perform a standard trial pick-up and put-down action on it. Since the internal mechanical structure of the target is constant under force, its corresponding "PTV" value represents the pure system measurement background noise. Therefore, the normalized path change generated in this process can be used as the geometric dead baseline.

[0071] (2) Calibration of static friction force breakthrough baseline: The system commands the robot to perform a standard trial pick-and-place action on a new file box in a warehouse that is known to be in good condition and with a clean surface. The goal of this operation is a typical, non-destructive static friction force breakthrough process. The normalized path change generated in this process can then be used as the static friction force breakthrough baseline.

[0072] (3) Setting thresholds: A safety margin can be added to the geometrically locked baseline as the first threshold. To ensure that the second threshold is always greater than the first threshold, the second threshold can be selected as follows: The max() function is the maximum value selection function. For static friction to break through the baseline, The first threshold, The preset minimum positive interval, d, is used to ensure that the second threshold is always greater than the first threshold, thereby forming a discrimination interval. In this embodiment of the invention, the settings of the safety margin and the minimum positive interval can be set according to the accuracy requirements of the diagnostic process under specific implementation. This embodiment of the invention does not elaborate on or limit these settings.

[0073] Preferably, in this embodiment of the invention, the method further includes a threshold update process, the threshold update process comprising:

[0074] Firstly, considering that the statistical results obtained when a resistance event occurs may differ due to factors such as the material, storage age, and temperature and humidity environment of the file boxes in different areas of the storage room, in order to achieve accurate threshold updates, this embodiment of the invention divides all file locations in the unmanned archive storage room into multiple local areas and performs individual threshold updates for each local area. The specific division method can be adaptively divided based on the size of the storage space, which will not be elaborated or limited in this embodiment of the invention.

[0075] For each local region, the number of event samples diagnosed as geometric jamming and static friction exceedance is counted. If the number of event samples is less than a preset threshold, the current statistics are considered unreliable, and the threshold is not updated. If the number of event samples is not less than the preset threshold, the normalized path change of the geometric jamming event samples is counted to obtain the updated first threshold, and the normalized path change of the static friction exceedance event samples is counted to obtain the updated second threshold. In subsequent real-time diagnosis, the updated two thresholds are used as the diagnostic criteria for that local region and are continuously updated.

[0076] In this embodiment of the invention, the updated first threshold can be selected as the 95th percentile of the normalized path change in the statistical geometric jamming state event samples. The percentile is chosen here instead of the mean because it is less sensitive to a few extreme outliers in the data distribution and can more stably define the upper bound of the normalized path change value distribution under the "geometric jamming" state. After obtaining the updated first threshold, the 5th percentile of the normalized path change in the statistical static friction breakthrough state is then used... The updated second threshold is obtained, where This represents the 5th percentile of the normalized path change in static friction force at the breakthrough state. This is the first threshold after the update. The quantity threshold can be set to 20. The purpose of this threshold is to control the frequency and accuracy of the updates, and it can be specifically set in the implementation.

[0077] Furthermore, since the deterioration of the file box, such as surface stickiness and swelling due to moisture, is a slow and gradual process, its characteristics will be reflected in the long-term trend of the normalized path change value of multiple access events. Therefore, this embodiment of the invention also includes a file box status analysis process for the target file location, including:

[0078] For the target file location, within a preset analysis time period, all event samples judged as static friction breakthroughs are statistically analyzed, and the sequence of normalized path changes corresponding to these events is used as the analysis sequence. If the length of the analysis sequence is less than a preset length threshold, it indicates that the sample data is insufficient and the analysis is not meaningful; therefore, the file box state analysis process is not performed. If the length of the analysis sequence is greater than or equal to the preset length threshold, a linear fit is performed on the analysis sequence. If the slope of the fitted line is greater than a preset slope threshold, it indicates that the internal state reconstruction required for each successful retrieval of the file box is becoming increasingly drastic. This directly corresponds to the deterioration process of the file box's surface adhesion degree increasing over time, indicating that the file box state requires maintenance; otherwise, it is determined that maintenance is not required. Staff can perform targeted checks and maintenance on documents based on whether maintenance is needed. Similarly, if a geometric jam is directly reported during the above process, the corresponding document location can be directly judged as requiring maintenance, and staff can be reminded to handle it.

[0079] In this embodiment of the invention, the file box status analysis process at the target file location can be set to be performed once a week, and the linear fitting can be performed using a linear regression algorithm. The specific algorithm is a technical means well known to those skilled in the art and will not be described in detail here.

[0080] In summary, this invention sets up a trial pick-and-place action process for the target file location and monitors the feedback motor current response sequence and motor speed response sequence in real time to obtain the instantaneous input power sequence and instantaneous dissipated power sequence for a local time period. The distribution of system energy from the input frequency band to the dissipation frequency band during the local time period is solved to obtain the matching result for each time window. By calculating the difference characteristics of continuous matching results and comparing them with a threshold, the contact event state of the target file location can be determined. This invention achieves effective differentiation of two resistance states during file pick-and-place by performing frequency domain decoupling analysis on the response information during the trial pick-and-place action process and performing difference analysis on the energy distribution path during the process.

[0081] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0082] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. An intelligent archiving method for an unmanned archive repository, characterized in that, The method includes: The file retrieval and placement robot is controlled to perform a trial retrieval and placement action at the target file location, and the motor current response sequence and motor speed response sequence are collected during the trial retrieval and placement action. Based on the local sequences of the motor current response sequence and the motor speed response sequence within a preset time window, the instantaneous input power sequence and instantaneous dissipated power sequence for each time window are obtained; spectral analysis is performed on the instantaneous input power sequence and instantaneous dissipated power sequence to obtain the instantaneous input power spectrum vector and instantaneous dissipated power spectrum vector. The instantaneous input power spectrum vector and the instantaneous dissipated power spectrum vector are matched and analyzed to obtain the matching results under each time window; the difference features of the matching results under all adjacent time windows are statistically analyzed, and the difference features are compared with a preset threshold to determine the contact event state of the target file position; the contact event state includes geometric jamming, static friction force breakthrough, and uncertain events; The process of obtaining the instantaneous input power spectrum vector and the instantaneous dissipation power spectrum vector includes: Take either the instantaneous input power sequence or the instantaneous dissipated power sequence as the sequence to be analyzed, process the sequence to be analyzed using the Fast Fourier Transform, and downsample the obtained power spectrum in multiple preset frequency bands to obtain the corresponding power spectrum vector. The matching analysis of the instantaneous input power spectrum vector and the instantaneous dissipated power spectrum vector includes: A matching matrix is ​​constructed based on the instantaneous input power spectrum vector and the instantaneous dissipation power spectrum vector. The rows of the matching matrix represent the positions of elements in the instantaneous input power spectrum vector, the columns represent the positions of elements in the instantaneous dissipation power spectrum vector, and the element values ​​in the matrix are the absolute values ​​of the differences between two elements at corresponding positions. The matching matrix is ​​optimized using the Hungarian algorithm to obtain an instantaneous energy distribution matrix of the same size as the matching matrix. The instantaneous energy distribution matrix is ​​a binary matrix. The instantaneous energy distribution matrix is ​​used as the matching result.

2. The intelligent archiving method for an unmanned archive repository according to claim 1, characterized in that, The method for obtaining the instantaneous input power sequence includes: At each moment, the current is multiplied by the motor torque constant to obtain the torque, and the torque is multiplied by the motor speed to obtain the instantaneous input power; The instantaneous input power at each moment within the time window constitutes the instantaneous input power sequence.

3. The intelligent archiving method for an unmanned archive repository according to claim 1, characterized in that, The method for obtaining the instantaneous power dissipation sequence includes: At each moment, the instantaneous input power and motor speed are substituted into the instantaneous power balance equation of the rotating system to obtain the instantaneous dissipated power; The instantaneous power dissipation at each moment within the time window constitutes the instantaneous power dissipation sequence.

4. The intelligent archiving method for an unmanned archive repository according to claim 1, characterized in that, The difference feature is the Flobenius norm between two adjacent matching matrices.

5. The intelligent archiving method for an unmanned archive repository according to claim 4, characterized in that, The determination of the contact event status of the target file location includes: The differential features are summed to obtain the overall difference. The overall difference is then normalized to obtain the normalized path change. If the change in the normalized path is less than a preset first threshold, it is determined to be a geometrically stuck state; if the change in the normalized path is greater than a preset second threshold, it is determined to be a static friction force breakthrough state; other cases are determined to be an uncertain state, where the first threshold is less than the second threshold.

6. The intelligent archiving method for an unmanned archive repository according to claim 5, characterized in that, The method further includes a threshold update process, which includes: The location of all archives in the unmanned archive storage is divided into multiple local areas. For each local area, the number of event samples diagnosed as geometric jamming and static friction breakthrough is counted. If the number of event samples is less than a preset threshold, the threshold is not updated. If the number of event samples is not less than the preset threshold, the normalized path change of the geometric jamming event samples is counted to obtain the updated first threshold, and the normalized path change of the static friction breakthrough event samples is counted to obtain the updated second threshold.

7. The intelligent archiving method for an unmanned archive repository according to claim 6, characterized in that, The method also includes a file box status analysis process for the target file location, including: For the target file location, all event samples judged as static friction breakthroughs are statistically analyzed within a preset analysis time period, and the sequence of normalized path changes corresponding to these events is used as the analysis sequence. If the length of the analysis sequence is less than a preset length threshold, the file box status analysis process is not performed. If the length of the analysis sequence is greater than or equal to the preset length threshold, a straight line is fitted to the analysis sequence. If the slope of the fitted straight line is greater than a preset slope threshold, the file box status is determined to need maintenance; otherwise, it is determined not to need maintenance.

8. An intelligent archiving system for an unmanned archive repository includes a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent archiving method for an unmanned archive warehouse as described in any one of claims 1 to 7.

Images