Warehouse load grading early warning method and system based on warehouse-in and warehouse-out and fault data

Abstract

The warehouse load grading early warning method and system based on warehouse in-out and fault data comprises: calculating current peer horizontal benchmark, full warehouse seasonal period benchmark and target device period condition benchmark; calculating condition expectation of total work load of target stacker in current statistical period and relative load ratio through all benchmarks; preliminarily judging whether various abnormal states of the current target stacker exist through all benchmarks and the relative load ratio, the abnormal states including peer busy and the machine falling behind, the machine deviating from the historical load and the machine low load and high fluctuation; combining the abnormal states and the set fault data and work load in the short-term period to calculate fault history reinforcement, deviation accumulation and short-time fluctuation index; obtaining comprehensive fault risk score by weighted summation of the fault history reinforcement, the deviation accumulation and the short-time fluctuation index; grading early warning based on the comprehensive fault risk score. The present application can continuously distinguish the overall business fluctuation and the single machine abnormality under the condition of missing data.

Description

Technical Field

[0001] This invention belongs to the field of industrial early warning technology, and more specifically, relates to a warehouse load classification early warning method and system based on inbound / outbound and fault data. Background Technology

[0002] Stacker cranes are crucial logistics equipment in intelligent automated warehouses. They are mainly used to perform tasks such as storing, retrieving, handling, repositioning, and arranging continuous operations of goods within aisles. Their operating status directly affects the overall throughput efficiency, task completion rate, fault handling speed, and maintenance costs of the warehouse. Therefore, it is of great value to conduct operational status assessments, identify anomalies, and issue early warnings for stacker cranes.

[0003] In current existing technologies, it is usually as follows: Figure 1 The judgment shown is based solely on simple workload statistics or, as... Figure 2 The analysis is performed using data from underlying sensors, such as vibration signals, current signals, temperature signals, displacement signals, and other underlying state parameters.

[0004] The above solution has at least the following problems: Judging solely based on changes in the workload of a single stacker crane presents the following problems: it is easily affected by business peaks and troughs, task allocation adjustments, and overall warehouse load fluctuations. When the overall warehouse workload decreases during a certain period, the workload of multiple stacker cranes will decrease synchronously; similarly, when a single stacker crane experiences performance abnormalities or falls behind in load, it may also manifest as a decrease in workload. Without a mechanism for cross-sectional comparison of multiple devices within the same statistical period, it is difficult to distinguish between "overall business fluctuations" and "abnormal lag of a single crane," leading to false alarms or missed alarms.

[0005] Relying on data from underlying sensors for analysis presents several challenges: Firstly, underlying sensor data is more difficult to obtain than historical inbound / outbound data and fault statistics from stacker cranes. It is also harder to retain and uniformly schedule this data within the equipment control and maintenance systems. Secondly, when some existing automated warehouses lack sufficient underlying sensors or their interfaces are not publicly accessible, making it difficult to obtain continuous, stable, and complete underlying status data, early warning schemes cannot be directly implemented. Furthermore, issues such as discontinuous sampling, missing data, inconsistent equipment models, inconsistent data collection standards, or high costs associated with interface modifications may arise. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention proposes a warehouse load classification and early warning method and system based on inbound / outbound and fault data.

[0007] The present invention adopts the following technical solution.

[0008] The first aspect of this invention proposes a warehouse load classification and early warning method based on inbound / outbound and fault data, comprising: A week is divided into multiple daily cycles, and a daily cycle includes multiple statistical cycles. The median of the total workload of all normally operating stacker cranes in the current statistical cycle is used as the current peer horizontal benchmark. The median of the total workload of all normally operating stacker cranes and the target stacker crane in the corresponding statistical cycles of the historical daily cycles is calculated separately, and used as the seasonal time benchmark of the whole warehouse and the time condition benchmark of the target equipment. Calculate the conditional expectation of the total workload of the target stacker crane in the current statistical period using all benchmarks; calculate the relative load ratio based on the actual value and conditional expectation of the total workload of the target stacker crane in the current statistical period. By using all benchmarks and relative load ratios, we can make a preliminary judgment on whether there are any abnormal states of the current target stacker crane. Abnormal states include the companion being busy while the machine is lagging behind, the machine's load deviating from its historical load, and the machine having low load and high fluctuations. The historical reinforcement amount of faults is calculated based on the number of faults and the cumulative duration of faults within the set short-term period. If there is an abnormal state where the machine is lagging behind while the companion is busy or the machine deviates from the historical load, the cumulative deviation amount is calculated based on the relative load ratio within the set long-term period. If there is an abnormal state where the machine has low load and high fluctuations, the short-term fluctuation index is calculated based on the workload of the target stacker crane within the set short-term period. The comprehensive fault risk score is obtained by weighting and summing the historical fault reinforcement amount, cumulative deviation amount, and short-term fluctuation indicators according to the set weights; and graded early warning is carried out based on the comprehensive fault risk score.

[0009] Preferably, the conditional expectation of the total workload of the target stacker crane in the current statistical period is calculated using all benchmarks, specifically as follows: The current peer horizontal benchmark and the full position seasonal benchmark are weighted and summed according to the set weights to obtain the full position reference benchmark; If, when calculating the target equipment time period condition baseline, the total number of total operations of the target stacker crane corresponding to the corresponding historical daily cycle and statistical cycle exceeds the set first quantity threshold, then the target equipment time period condition baseline will be used as the condition expectation. Otherwise, the historical busy statistical period is selected from the corresponding daily period and the historical busy statistical period. The median of the total workload of all normally operating stacker cranes in the historical busy statistical period is greater than the set busy threshold. If the number of historical busy statistical periods exceeds the set second threshold, the average of the ratio of the total workload of the target set in all busy statistical periods and when running normally to the full position benchmark of the corresponding busy statistical period is obtained as the historical reference ratio; then the product of the current full position benchmark and the historical reference ratio is used as the conditional expectation; otherwise, the current full position benchmark is used as the conditional expectation.

[0010] Preferably, the relative load ratio is calculated based on the actual value and expected value of the total workload of the target stacker crane in the current statistical period, specifically as follows: The relative load ratio is the actual value of the total workload of the target stacker crane in the current statistical period divided by the sum of the expected value and the set fixed value.

[0011] Preferably, the presence of various abnormal conditions of the current target stacker crane is initially determined by using all benchmarks and relative load ratios; If the overall reference benchmark is greater than the set busy threshold and the relative load ratio is less than the set low percentile threshold, it is determined that there is an abnormal state where the companion is busy but the local machine has fallen behind. If the target device's time period baseline is greater than the set busy threshold and the relative load ratio is less than the set low percentile threshold, then it is determined that there is an abnormal state where the local load deviates from the historical load. If the total workload of the target stacker crane in the current statistical period is less than the set low load threshold, and the variance of the total workload of the target stacker crane in the short-term period is greater than the set dispersion threshold, then it is determined that there is an abnormal state of low load and high fluctuation of the machine.

[0012] Preferably, the fault history reinforcement amount is calculated based on the number of faults and the cumulative duration of faults within a set short-term period, specifically as follows: The sum of the number of failures within the set short-term period and 1 is input into the ln function to obtain the first product term; Divide the cumulative fault duration within the set short-term period by the difference between the number of statistical periods included in the short-term period and 1, and add 1 to the result of the division to obtain the second product term; Multiply the difference between the set first adjustment parameter and the time interval of the most recent fault by the set second adjustment parameter, and input the multiplication result into the sigmoid function to obtain the third product term; Multiplying the first, second, and third product terms together yields the fault history reinforcement quantity.

[0013] Preferably, the cumulative deviation is calculated based on the relative load ratio over a predetermined long-term period, specifically as follows: For a given long-term period, if the relative load ratio of the corresponding statistical period is greater than or equal to the set lower quantile threshold, then the deviation is 0. Otherwise, the deviation is the square of the difference between the low quantile threshold and the relative load ratio, multiplied by the time decay term, and then multiplied by the abnormal state correction term. The time decay term is the result of exponential calculation with the natural base e as the base, multiplying the time decay coefficient by the difference between the index of the current statistical period and the index of the corresponding statistical period, and then multiplying by -1 to obtain the exponent. If there is an abnormal state where the companion is busy and the local machine is out of sync in the corresponding statistical period, the first abnormal state factor of the corresponding statistical period is 1, otherwise it is 0. If there is an abnormal state where the local machine deviates from the historical load in the corresponding statistical period, the second abnormal state factor of the corresponding statistical period is 1, otherwise it is 0. The abnormal state correction term is 1 plus the product of the set first enhancement coefficient and the first abnormal state factor, plus the product of the set second enhancement coefficient and the second abnormal state factor. The cumulative deviation is obtained by summing the deviations over a set long-term period. If neither the companion is busy and the local machine falls behind, nor the local machine deviates from its historical load, then the cumulative deviation is 0.

[0014] Preferably, a short-term fluctuation index is calculated based on the workload of the target stacker crane within a set short-term period, specifically as follows: Obtain the variance and skewness of the total workload of the target stacker crane in a set short-term period; Divide the variance by the set normalization threshold and add 1. Input the sum into the ln function. Multiply the result of the ln function by 1 and the sum of the absolute values ​​of the skewness to obtain the short-term volatility index. If there is no abnormal state of low load and high fluctuation on the machine, the short-term fluctuation index is 0.

[0015] Preferably, a statistical period contains multiple sampling points. If there are missing data when obtaining the workload of each sampling point in the statistical period, the missing data is filled in by interpolation. If missing data needs to be filled within the set long-term period, then calculate the uncertainty correction amount; For each statistical period with missing data, the total number of missing data in the corresponding statistical period is calculated and multiplied by the corresponding time decay term to obtain the uncertainty of the corresponding statistical period. Divide the sum of uncertainties for all statistical periods with missing data by the sum of time decay terms for the statistical periods with missing data; Multiply the uncertainty result by the set first adjustment coefficient and then by -1 as the exponent, with the natural base e as the base, and perform exponential operation. Subtract the product of the set second adjustment coefficient and the exponential operation result from 1. If the result of the subtraction exceeds the set maximum adjustment coefficient, the maximum adjustment coefficient is used as the uncertainty correction coefficient; otherwise, the result of the subtraction is used as the uncertainty correction coefficient. The upper limit of the comprehensive failure risk score is obtained by multiplying the comprehensive failure risk score by 1 and summing it with the uncertainty correction coefficient. The lower limit of the comprehensive failure risk score is obtained by multiplying the comprehensive failure risk score by 1 and summing it with the uncertainty correction coefficient.

[0016] Preferably, if there is no missing data to fill in within the set long-term period, a high-risk warning is issued when the overall failure risk is greater than or equal to the set high-risk threshold, a normal risk warning is issued when the overall failure risk is less than the set high-risk threshold but greater than or equal to the set medium-risk threshold, and no warning is issued when the overall failure risk is less than the set low-risk threshold. If missing data is required within the set long-term period, a high-risk warning will be issued when the upper limit of the overall failure risk is greater than or equal to the set high-risk threshold; a normal risk warning will be issued when the upper limit of the overall failure risk is less than the set high-risk threshold and the overall failure risk is greater than or equal to the set medium-risk threshold; and no warning will be issued when the overall failure risk is less than the set low-risk threshold.

[0017] The second aspect of this invention proposes a warehouse load classification and early warning system based on inbound / outbound and fault data using the method described in the first aspect of this invention, including a baseline calculation module, a relative load ratio calculation module, an abnormal state preliminary judgment module, an abnormal index calculation module, and a risk classification and early warning module, specifically: The benchmark calculation module is used to divide a week into multiple daily cycles, and to include multiple statistical cycles in one daily cycle. The median of the total workload of all normally operating stacker cranes in the current statistical cycle is used as the current peer horizontal benchmark. The median of the total workload of all normally operating stacker cranes and the target stacker crane in the corresponding statistical cycles of the historical daily cycles is calculated separately, and used as the seasonal time period benchmark for the whole warehouse and the time period condition benchmark for the target equipment. Relative load ratio calculation module: used to calculate the conditional expectation of the total workload of the target stacker crane in the current statistical period using all benchmarks; and to calculate the relative load ratio based on the actual value and conditional expectation of the total workload of the target stacker crane in the current statistical period. The preliminary judgment module for abnormal conditions is used to make a preliminary judgment on whether the current target stacker crane has various abnormal conditions based on all benchmarks and relative load ratios. Abnormal conditions include the companion being busy while the machine is lagging behind, the machine's load deviating from its historical load, and the machine having low load and high fluctuations. Anomaly indicator calculation module: used to calculate the historical reinforcement amount of faults based on the number of faults and the cumulative duration of faults within a set short period; if there is an abnormal state where the companion is busy and the machine is lagging behind or the machine deviates from the historical load, the cumulative deviation amount is calculated based on the relative load ratio within a set long period; if there is an abnormal state where the machine has low load and high fluctuations, the short-term fluctuation index is calculated based on the workload of the target stacker crane within a set short period. Risk classification and early warning module: It is used to obtain a comprehensive fault risk score by weighting and summing the historical fault reinforcement amount, cumulative deviation amount and short-term fluctuation indicators according to the set weights; and to conduct classified early warning based on the comprehensive fault risk score.

[0018] A third aspect of the present invention provides an apparatus comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor performing steps of the warehouse load classification and early warning method based on inbound / outbound and fault data as described in the first aspect of the present invention.

[0019] A fourth aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, uses the steps of the warehouse load classification and early warning method based on inbound / outbound and fault data described in the first aspect of the present invention.

[0020] The beneficial effects of this invention are as follows: Compared with the prior art, this invention calculates the expected value of conditions by constructing the current peer horizontal benchmark, the seasonal time benchmark of the whole warehouse, and the time condition benchmark of the target equipment. The multi-layer dynamic benchmark establishes a reasonable expected value for each equipment in each statistical period that conforms to historical and current scenarios. By judging abnormal states through multi-layer dynamic benchmarks and expected values ​​of conditions, and combining abnormal states to calculate comprehensive abnormal indicators, compared with traditional methods based on a single threshold or post-event alarms, this invention can effectively distinguish whether the equipment is reasonably idle or abnormally out of service, significantly reducing false alarms. By quantifying the uncertainty of missing data, the warning results are more accurate. In the intelligent automated warehouse environment where multiple stacker cranes operate in parallel, a complete technical chain is formed that can be used to identify overall business fluctuations and single machine anomalies, can continue to operate under missing conditions, and can directly drive control actions. Attached Figure Description

[0021] Figure 1 This is a diagram illustrating how existing similar technologies rely solely on simple workload statistics for judgment. Figure 2 This is a schematic diagram of a typical device early warning scheme based on underlying sensor data in the existing technology; Figure 3 This is a schematic diagram illustrating the application scenarios and data processing objects of the present invention in an automated storage and retrieval system. Figure 4 This is a schematic diagram of data processing and control output; Figure 5 This is a detailed flowchart of the method of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0023] like Figure 5 As shown, Embodiment 1 of the present invention proposes a warehouse load classification and early warning method based on inbound / outbound and fault data, including: A week is divided into multiple daily cycles, and a daily cycle is divided into multiple statistical cycles. The median of the total workload of all normally operating stacker cranes in the current statistical period is used as the current peer benchmark, and the formula is:

[0024] in, As the current peer's lateral reference; For the stacker crane to operate normally j In the current statistical period t Total workload; For the normal operation of stacker crane assembly; Calculate the median of the total workload for all normally operating stacker cranes and the target stacker crane for the corresponding statistical period of the historical daily cycle, respectively. Use this median as the seasonal benchmark for the entire warehouse and the time-condition benchmark for the target equipment. The formula is:

[0025] in, , These represent the seasonal benchmark for the entire warehouse and the time-based benchmark for the target equipment, respectively. , These are stacker cranes in normal operation. j The target stacker crane in the statistical period Total workload; For the current statistical period t Located in the One-day cycle; For the current statistical period t In the Index for daily cycles; Indicates the statistical period The number of the day cycle located at, Indicates the statistical period In the Index for daily cycles.

[0026] Calculate the conditional expectation of the total workload of the target stacker crane in the current statistical period using all benchmarks; calculate the relative load ratio based on the actual value and conditional expectation of the total workload of the target stacker crane in the current statistical period. By using all benchmarks and relative load ratios, we can make a preliminary judgment on whether there are any abnormal states of the current target stacker crane. Abnormal states include the companion being busy while the machine is lagging behind, the machine's load deviating from its historical load, and the machine having low load and high fluctuations. The historical reinforcement amount of faults is calculated based on the number of faults and the cumulative duration of faults within the set short-term period. If there is an abnormal state where the machine is lagging behind while the companion is busy or the machine deviates from the historical load, the cumulative deviation amount is calculated based on the relative load ratio within the set long-term period. If there is an abnormal state where the machine has low load and high fluctuations, the short-term fluctuation index is calculated based on the workload of the target stacker crane within the set short-term period. It should be noted that both the long-term and short-term periods set in this embodiment include several statistical periods. Specifically, the long-term period in this embodiment is set to 24 hours, that is, the long-term period includes the current statistical period and the previous 23 statistical periods. The short-term period is set to 6 hours, that is, the long-term period includes the current statistical period and the previous 5 statistical periods. The comprehensive fault risk score is obtained by weighting and summing the historical fault reinforcement amount, cumulative deviation amount, and short-term fluctuation indicators according to the set weights; and graded early warning is carried out based on the comprehensive fault risk score.

[0027] It should be noted that, as Figure 3 As shown, the workload is obtained from the historical inbound and outbound database, and the fault data such as the number of faults and the cumulative duration of faults are obtained from the fault database. The historical inbound / outbound database and the fault database are two types of macro-industrial data. The historical inbound / outbound data includes the actual operating status of each stacker crane within a continuous statistical period, including at least the stacker crane identification, operating time, operating direction, and operating quantity. The fault statistics are used to characterize the occurrence of fault events of each stacker crane, including at least the equipment identification or aisle identification, fault occurrence time, fault duration, and fault information.

[0028] In one implementation, when the original inbound / outbound data is an event-level job log, its minute-level or second-level time information is retained so that auxiliary statistical features such as task interval within an hour, task distribution entropy, and number of task records can be further generated later; when the original fault data is an event-level log, the fault start time and fault duration are retained to support subsequent fault mapping according to a unified statistical period.

[0029] The historical inbound / outbound data and fault statistics are processed using a unified time axis, mapping data at different time granularities to a unified statistical period. Specifically, in this embodiment, the daily period is one day, and one daily period is divided into 24 hours, with each hour serving as a statistical period. Furthermore, the association between fault events and target stacker cranes is established based on stacker crane identifiers, aisle identifiers, or preset correspondences.

[0030] For historical inbound and outbound data, the inbound operation volume, outbound operation volume, and total operation volume are first summarized according to the target stacker crane and the statistical period. For fault statistics, fault events are mapped to one or more corresponding hourly statistical periods based on the fault occurrence time and duration. Subsequently, a continuous statistical grid is constructed in the form of "stacker crane - hour" so that each stacker crane forms a continuous hourly statistical sequence throughout the entire analysis period.

[0031] If a stacker crane lacks historical inbound and outbound data within a certain statistical period, the subsequent analysis will not be terminated directly. Instead, the statistical period will be retained in the continuous statistical grid, and a corresponding missing marker will be set after the data is completed, so that the continuity of values ​​and the reliability of data will be considered simultaneously during the subsequent risk assessment.

[0032] When obtaining the total workload, interpolation is performed to complete any missing data. Preferably, the conditional expectation of the total workload of the target stacker crane in the current statistical period is calculated using all benchmarks, specifically as follows: The current peer horizontal benchmark and the full position seasonal benchmark are weighted and summed according to the set weights to obtain the full position reference benchmark;

[0033] in, Used as a benchmark for the entire portfolio; , These are the weights of the current peer horizontal benchmark and the weights of the full position seasonal benchmark, respectively. If, when calculating the target equipment time period baseline, the total number of operations collected for the target stacker crane corresponding to the historical daily cycle and the corresponding statistical period exceeds the set first quantity threshold, then it is determined that... Available, setting the target device time period condition benchmark. As a conditional expectation; Otherwise, the historical busy statistical period is selected from the corresponding daily period and the historical busy statistical period. The median of the total workload of all normally operating stacker cranes in the historical busy statistical period is greater than the set busy threshold. If the number of historically busy statistical periods exceeds the set second threshold, then the historical reference ratio will be used. Available, obtain the average of the ratio of the total workload of the target set across all busy statistical periods and during normal operation to the full position benchmark of the corresponding busy statistical period as the historical reference ratio; then, the product of the current full position benchmark and the historical reference ratio is used as the conditional expectation; otherwise, the current full position benchmark is used as the conditional expectation; conditional expectation The formula is:

[0034] Preferably, the relative load ratio is calculated based on the actual value and expected value of the total workload of the target stacker crane in the current statistical period, specifically as follows: The relative load ratio is calculated by dividing the actual total workload of the target stacker crane in the current statistical period by the sum of the expected and fixed values. The formula is as follows:

[0035] in, This is a fixed value that is set. This represents the actual total workload of the target stacker crane during the current statistical period. For the current statistical period t The relative load ratio.

[0036] Preferably, the presence of various abnormal conditions of the current target stacker crane is initially determined by using all benchmarks and relative load ratios; If the overall reference benchmark is greater than the set busy threshold and the relative load ratio is less than the set low percentile threshold, it is determined that there is an abnormal state where the companion is busy but the local machine has fallen behind.

[0037] in, This is the first abnormal state factor. A value of 1 indicates an abnormal state where the companion is busy and the local machine has fallen behind, while a value of 0 indicates that it does not exist. The set busy threshold; The set lower quantile threshold; If the target device's time period baseline is greater than the set busy threshold and the relative load ratio is less than the set low percentile threshold, then it is determined that there is an abnormal state where the local load deviates from the historical load.

[0038] in, This is the second abnormal state factor. A value of 1 indicates the existence of an abnormal state where the local load deviates from the historical load, while a value of 0 indicates that it does not exist. If the total workload of the target stacker crane in the current statistical period is less than the set low load threshold, and the variance of the total workload of the target stacker crane in the short-term period is greater than the set dispersion threshold, then it is determined that there is an abnormal state of low load and high fluctuation of the machine.

[0039] in, This is the third abnormal state factor. A value of 1 indicates the existence of an abnormal state of low load and high fluctuation on the local machine, while a value of 0 indicates that it does not exist. The set low load threshold; The variance of the total workload for the target stacker crane in the short-term cycle; The set dispersion threshold.

[0040] Preferably, the fault history reinforcement amount is calculated based on the number of faults and the cumulative duration of faults within a set short-term period, specifically as follows: The sum of the number of failures within the set short-term period and 1 is input into the ln function to obtain the first product term; Divide the cumulative fault duration within the set short-term period by the difference between the number of statistical periods included in the short-term period and 1, and add 1 to the result of the division to obtain the second product term; Multiply the difference between the set first adjustment parameter and the time interval of the most recent fault by the set second adjustment parameter, and input the multiplication result into the sigmoid function to obtain the third product term; Multiplying the first, second, and third product terms together yields the fault history reinforcement quantity.

[0041]

[0042] in, This is an enhancement of the fault history. The number of failures within a set short-term period; The cumulative duration of faults within a set short-term period; The total number of statistical periods included in the defined short-term period. That is, before the current statistical period Average cumulative fault duration over a statistical period; , These are the first and second adjustment parameters, respectively; The time interval between the most recent failure; This item represents the timeliness of the fault. The shorter the time since the most recent fault occurred in the current statistical period, the more likely the repair was incomplete or there is an inherent defect, and the higher the probability of another fault in the short term. In other words, the more faults there are, the longer the cumulative fault duration, and the closer the fault is to the present, the greater the historical reinforcement of the fault. Preferably, the cumulative deviation is calculated based on the relative load ratio over a predetermined long-term period, specifically as follows: For a given long-term period, if the relative load ratio of the corresponding statistical period is greater than or equal to the set lower quantile threshold, then the deviation is 0. Otherwise, the deviation is the square of the difference between the low quantile threshold and the relative load ratio, multiplied by the time decay term, and then multiplied by the abnormal state correction term. The time decay term is the result of exponential calculation with the natural base e as the base, multiplying the time decay coefficient by the difference between the index of the current statistical period and the index of the corresponding statistical period, and then multiplying by -1 to obtain the exponent. If there is an abnormal state where the companion is busy and the local machine is out of sync in the corresponding statistical period, the first abnormal state factor of the corresponding statistical period is 1, otherwise it is 0. If there is an abnormal state where the local machine deviates from the historical load in the corresponding statistical period, the second abnormal state factor of the corresponding statistical period is 1, otherwise it is 0. The abnormal state correction term is 1 plus the product of the set first enhancement coefficient and the first abnormal state factor, plus the product of the set second enhancement coefficient and the second abnormal state factor. The cumulative deviation is obtained by summing the deviations over a set long-term period. The formula is: in, This is the deviation from the cumulative amount; This is the difference between the index of the current statistical period and the index of the corresponding statistical period. This is the time decay coefficient; The total number of statistical periods included in the set long-term period; For statistical period The relative load ratio; , These are the first enhancement coefficient and the second enhancement coefficient, respectively. , Statistical periods The first abnormal state factor and the second abnormal state factor; If neither the companion is busy and the local machine falls behind, nor the local machine deviates from its historical load, then the cumulative deviation is 0. Preferably, a short-term fluctuation index is calculated based on the workload of the target stacker crane within a set short-term period, specifically as follows: Obtain the variance and skewness of the total workload of the target stacker crane in a set short-term period; Divide the variance by the set normalization threshold and add 1. Input the sum into the ln function. Multiply the result of the ln function by 1 and the sum of the absolute values ​​of the skewness to obtain the short-term volatility index.

[0043] in, This is a short-term fluctuation indicator; The skewness of the total workload of the target stacker crane within a set short period; The set normalization threshold; If there is no abnormal state of low load and high fluctuation on the machine, the short-term fluctuation index is 0; The short-term fluctuation index indicates that the larger the variance of the short-term workload, the more obvious the skewness, and the greater the coupling of short-term fluctuations when it is in a low-load, high-fluctuation mode.

[0044] Preferably, a statistical period contains multiple sampling points. If there are missing data when obtaining the workload of each sampling point in the statistical period, the missing data is filled in by interpolation. If missing data needs to be filled within the set long-term period, then calculate the uncertainty correction amount; For each statistical period with missing data, the total number of missing data in the corresponding statistical period is calculated and multiplied by the corresponding time decay term to obtain the uncertainty of the corresponding statistical period. Divide the sum of uncertainties for all statistical periods with missing data by the sum of time decay terms for the statistical periods with missing data; Multiply the uncertainty result by the set first adjustment coefficient and then by -1 as the exponent, with the natural base e as the base, and perform exponential operation. Subtract the product of the set second adjustment coefficient and the exponential operation result from 1. If the result of the subtraction exceeds the set maximum adjustment coefficient, the maximum adjustment coefficient is used as the uncertainty correction coefficient; otherwise, the result of the subtraction is used as the uncertainty correction coefficient. The formula is:

[0045] in, For uncertainties, correction factors are used. , These are the first adjustment coefficient and the second adjustment coefficient, respectively. For the corresponding statistical period The total number of missing data; The upper limit of the comprehensive failure risk score is obtained by multiplying the comprehensive failure risk score by 1 and summing it with the uncertainty correction coefficient. The lower limit of the comprehensive failure risk score is obtained by multiplying the comprehensive failure risk score by 1 and summing it with the uncertainty correction coefficient.

[0046] Preferably, if there is no missing data to fill in within the set long-term period, a high-risk warning is issued when the overall failure risk is greater than or equal to the set high-risk threshold, a normal risk warning is issued when the overall failure risk is less than the set high-risk threshold but greater than or equal to the set medium-risk threshold, and no warning is issued when the overall failure risk is less than the set low-risk threshold. If missing data is required within the set long-term period, a high-risk warning will be issued when the upper limit of the overall failure risk is greater than or equal to the set high-risk threshold; a normal risk warning will be issued when the upper limit of the overall failure risk is less than the set high-risk threshold and the overall failure risk is greater than or equal to the set medium-risk threshold; and no warning will be issued when the overall failure risk is less than the set low-risk threshold.

[0047] Control actions under different risks are as follows: (1) Low risk: Status logging, prompting display, and allowance for full-load operation; (2) Medium risk: trigger manual review, restrict the allocation of heavy-load tasks; and modify the suggested capacity limit: set a high quantile (e.g., the third quarter quantile), obtain the statistical period corresponding to the full warehouse reference benchmark being greater than the set busy threshold as the busy period, obtain the high quantile value of the total workload of the target stacker crane during the busy period; and sum the high quantile values ​​of the total workload of the current statistical period and the total workload of the target stacker crane during the busy period according to the set weight to obtain the suggested capacity limit.

[0048] (3) High risk: Triggers a high priority alarm. It is recommended to suspend high-risk tasks, arrange maintenance, or implement protective controls.

[0049] Embodiment 2 of the present invention proposes a warehouse load classification and early warning system based on inbound / outbound and fault data using the method described in Embodiment 1 of the present invention, comprising, specifically: The benchmark calculation module is used to divide a week into multiple daily cycles, and to include multiple statistical cycles in one daily cycle. The median of the total workload of all normally operating stacker cranes in the current statistical cycle is used as the current peer horizontal benchmark. The median of the total workload of all normally operating stacker cranes and the target stacker crane in the corresponding statistical cycles of the historical daily cycles is calculated separately, and used as the seasonal time period benchmark for the whole warehouse and the time period condition benchmark for the target equipment. Relative load ratio calculation module: used to calculate the conditional expectation of the total workload of the target stacker crane in the current statistical period using all benchmarks; and to calculate the relative load ratio based on the actual value and conditional expectation of the total workload of the target stacker crane in the current statistical period. The preliminary judgment module for abnormal conditions is used to make a preliminary judgment on whether the current target stacker crane has various abnormal conditions based on all benchmarks and relative load ratios. Abnormal conditions include the companion being busy while the machine is lagging behind, the machine's load deviating from its historical load, and the machine having low load and high fluctuations. Anomaly indicator calculation module: used to calculate the historical reinforcement amount of faults based on the number of faults and the cumulative duration of faults within a set short period; if there is an abnormal state where the companion is busy and the machine is lagging behind or the machine deviates from the historical load, the cumulative deviation amount is calculated based on the relative load ratio within a set long period; if there is an abnormal state where the machine has low load and high fluctuations, the short-term fluctuation index is calculated based on the workload of the target stacker crane within a set short period. Risk classification and early warning module: It is used to obtain a comprehensive fault risk score by weighting and summing the historical fault reinforcement amount, cumulative deviation amount and short-term fluctuation indicators according to the set weights; and to conduct classified early warning based on the comprehensive fault risk score.

[0050] Embodiment 3 of the present invention proposes a device including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the steps of the warehouse load classification and early warning method based on inbound / outbound and fault data described in Embodiment 1 of the present invention. The processor execution flow is as follows: Figure 4 As shown.

[0051] Embodiment 4 of the present invention proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, uses the steps of the warehouse load classification and early warning method based on inbound / outbound and fault data described in Embodiment 1 of the present invention.

[0052] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0053] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0054] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0055] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0056] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A warehouse load grading early warning method based on warehouse in and out and fault data, characterized in that, include: A week is divided into multiple daily cycles, and a daily cycle is divided into multiple statistical cycles. The median of the total workload of all normally operating stacker cranes in the current statistical period is used as the current peer horizontal benchmark. The median of the total workload of all normally operating stacker cranes and the target stacker crane in the corresponding statistical period of the historical corresponding daily period is calculated separately, and used as the seasonal time period benchmark of the whole warehouse and the time period condition benchmark of the target equipment. Based on all benchmarks, calculate the conditional expectation of the total workload of the target stacker crane in the current statistical period. Specifically, the current peer horizontal benchmark and the seasonal benchmark of the entire warehouse are weighted and summed according to the set weights to obtain the reference benchmark of the entire warehouse. If, when calculating the target equipment time period condition baseline, the total number of total operations of the target stacker crane corresponding to the corresponding historical daily cycle and statistical cycle exceeds the set first quantity threshold, then the target equipment time period condition baseline will be used as the condition expectation. Otherwise, select historical busy statistical periods from the corresponding historical daily period statistical periods. The median of the total workload of all normally operating stacker cranes in the historical busy statistical period is greater than the set busy threshold. If the number of historical busy statistical periods exceeds the set second quantity threshold, obtain the average of the ratio of the total workload of the target set in all busy statistical periods when it is normally operating to the full warehouse benchmark of the corresponding busy statistical period as the historical reference ratio. Then, the product of the current full warehouse reference benchmark and the historical reference ratio is used as the conditional expectation. Otherwise, the current full position reference benchmark will be used as the conditional expectation; Calculate the relative load ratio based on the actual value and expected condition of the total workload of the target stacker crane in the current statistical period; The system uses all benchmarks and relative load ratios to initially determine whether the current target stacker crane is in any abnormal state. Abnormal states include when the companion is busy but the machine is lagging behind, when the machine's load deviates from its historical load, and when the machine has a low load with high fluctuations. Specifically, if the overall warehouse reference benchmark is greater than the set busy threshold and the relative load ratio is less than the set low percentile threshold, then it is determined that there is an abnormal state where the companion is busy but the machine is lagging behind. If the target device's time period baseline is greater than the set busy threshold and the relative load ratio is less than the set low percentile threshold, then it is determined that there is an abnormal state where the local load deviates from the historical load. If the total workload of the target stacker crane in the current statistical period is less than the set low load threshold, and the variance of the total workload of the target stacker crane in the short-term period is greater than the set dispersion threshold, then it is determined that there is an abnormal state of low load and high fluctuation of the machine. The historical fault reinforcement amount is calculated based on the number of faults and the cumulative fault duration within a set short-term period. If there is an abnormal state where the machine is lagging behind while its companions are busy, or where the machine's load deviates from its historical load, the cumulative deviation amount is calculated based on the relative load ratio within a set long-term period. Specifically, for the set long-term period, if the relative load ratio of the corresponding statistical period is greater than or equal to the set lower quantile threshold, the deviation is 0; otherwise, the deviation is the square of the difference between the lower quantile threshold and the relative load ratio multiplied by a time decay term and then multiplied by an abnormal state correction term. The time decay term is calculated by multiplying the time decay coefficient by the current statistical period using the natural base e as the base. The difference between the index and the index of the corresponding statistical period is multiplied by -1 to obtain the exponent, and the result of the exponentiation is calculated. If there is an abnormal state in the corresponding statistical period where the companion is busy and the local machine is lagging behind, the first abnormal state factor of the corresponding statistical period is 1, otherwise it is 0. If there is an abnormal state in the corresponding statistical period where the local machine deviates from the historical load, the second abnormal state factor of the corresponding statistical period is 1, otherwise it is 0. The abnormal state correction term is 1 plus the product of the set first enhancement coefficient and the first abnormal state factor, plus the product of the set second enhancement coefficient and the second abnormal state factor. The deviations of the set long-term periods are superimposed to obtain the cumulative deviation. If neither the companion is busy and the local machine falls behind, nor the local machine deviates from its historical load, then the cumulative deviation is 0. If there is an abnormal state of low load and high fluctuation in the machine, the short-term fluctuation index is calculated based on the workload of the target stacker crane in the set short-term period. The comprehensive fault risk score is obtained by weighting and summing the historical fault reinforcement amount, cumulative deviation amount and short-term fluctuation indicators according to the set weights. A graded early warning system is implemented based on a comprehensive fault risk score.

2. The warehouse load classification and early warning method based on inbound / outbound and fault data according to claim 1, characterized in that: Based on the actual value and expected value of the total workload of the target stacker crane in the current statistical period, the relative load ratio is calculated as follows: The relative load ratio is the actual value of the total workload of the target stacker crane in the current statistical period divided by the sum of the expected value and the set fixed value.

3. The warehouse load classification and early warning method based on inbound / outbound and fault data according to claim 1, characterized in that: The historical reinforcement amount for faults is calculated based on the number of faults and the cumulative duration of faults within a set short-term period, specifically as follows: The sum of the number of failures within the set short-term period and 1 is input into the ln function to obtain the first product term; Divide the cumulative fault duration within the set short-term period by the difference between the number of statistical periods included in the short-term period and 1, and add 1 to the result of the division to obtain the second product term; Multiply the difference between the set first adjustment parameter and the time interval of the most recent fault by the set second adjustment parameter, and input the multiplication result into the sigmoid function to obtain the third product term; Multiplying the first, second, and third product terms together yields the fault history reinforcement quantity.

4. The warehouse load classification and early warning method based on inbound / outbound and fault data according to claim 1, characterized in that: The short-term fluctuation index is calculated based on the target stacker crane's workload within a set short-term period, specifically as follows: Obtain the variance and skewness of the total workload of the target stacker crane in a set short-term period; Divide the variance by the set normalization threshold and add 1. Input the sum into the ln function. Multiply the result of the ln function by 1 and the sum of the absolute values ​​of the skewness to obtain the short-term volatility index. If there is no abnormal state of low load and high fluctuation on the machine, the short-term fluctuation index is 0.

5. The warehouse load classification and early warning method based on inbound / outbound and fault data according to claim 1, characterized in that: A statistical period contains multiple sampling points. If there are missing data when obtaining the workload of each sampling point in the statistical period, the missing data is filled in by interpolation. If missing data needs to be filled within the set long-term period, then calculate the uncertainty correction amount; For each statistical period with missing data, the total number of missing data in the corresponding statistical period is calculated and multiplied by the corresponding time decay term to obtain the uncertainty of the corresponding statistical period. Divide the sum of uncertainties for all statistical periods with missing data by the sum of time decay terms for the statistical periods with missing data; Multiply the uncertainty result by the set first adjustment coefficient and then by -1 as the exponent, with the natural base e as the base, and perform exponential operation. Subtract the product of the set second adjustment coefficient and the exponential operation result from 1. If the result of the subtraction exceeds the set maximum adjustment coefficient, the maximum adjustment coefficient is used as the uncertainty correction coefficient; otherwise, the result of the subtraction is used as the uncertainty correction coefficient. The upper limit of the comprehensive failure risk score is obtained by multiplying the comprehensive failure risk score by 1 and summing it with the uncertainty correction coefficient. The lower limit of the comprehensive failure risk score is obtained by multiplying the comprehensive failure risk score by 1 and summing it with the uncertainty correction coefficient.

6. The warehouse load classification and early warning method based on inbound / outbound and fault data according to claim 5, characterized in that: If no missing data is filled within the set long-term period, a high-risk warning will be issued when the overall failure risk is greater than or equal to the set high-risk threshold; a normal risk warning will be issued when the overall failure risk is less than the set high-risk threshold but greater than or equal to the set medium-risk threshold; and no warning will be issued when the overall failure risk is less than the set low-risk threshold. If missing data is required within the set long-term period, a high-risk warning will be issued when the upper limit of the overall failure risk is greater than or equal to the set high-risk threshold; a normal risk warning will be issued when the upper limit of the overall failure risk is less than the set high-risk threshold and the overall failure risk is greater than or equal to the set medium-risk threshold; and no warning will be issued when the overall failure risk is less than the set low-risk threshold.

7. A warehouse load classification and early warning system based on inbound / outbound and fault data using the method of any one of claims 1-6, comprising a baseline calculation module, a relative load ratio calculation module, an abnormal state preliminary judgment module, an abnormal index calculation module, and a risk classification and early warning module, characterized in that: The benchmark calculation module is used to divide a week into multiple daily cycles, and to include multiple statistical cycles in one daily cycle. The median of the total workload of all normally operating stacker cranes in the current statistical cycle is used as the current peer horizontal benchmark. The median of the total workload of all normally operating stacker cranes and the target stacker crane in the corresponding statistical cycles of the historical daily cycles is calculated separately, and used as the seasonal time period benchmark for the whole warehouse and the time period condition benchmark for the target equipment. Relative load ratio calculation module: used to calculate the conditional expectation of the total workload of the target stacker crane in the current statistical period using all benchmarks; Calculate the relative load ratio based on the actual value and expected condition of the total workload of the target stacker crane in the current statistical period; The preliminary judgment module for abnormal conditions is used to make a preliminary judgment on whether the current target stacker crane has various abnormal conditions based on all benchmarks and relative load ratios. Abnormal conditions include the companion being busy while the machine is lagging behind, the machine's load deviating from its historical load, and the machine having low load and high fluctuations. Anomaly indicator calculation module: used to calculate the historical reinforcement amount of faults based on the number of faults and the cumulative duration of faults within a set short period; if there is an abnormal state where the companion is busy and the machine is lagging behind or the machine deviates from the historical load, the cumulative deviation amount is calculated based on the relative load ratio within a set long period; if there is an abnormal state where the machine has low load and high fluctuations, the short-term fluctuation index is calculated based on the workload of the target stacker crane within a set short period. Risk grading and early warning module: used to obtain a comprehensive fault risk score by weighting and summing the historical fault reinforcement amount, cumulative deviation amount and short-term fluctuation indicators according to the set weights; A graded early warning system is implemented based on a comprehensive fault risk score.

8. An apparatus comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor performs the steps of the warehouse load classification and early warning method based on inbound / outbound and fault data as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program that, when executed by a processor, uses the steps of the warehouse load classification and early warning method based on inbound / outbound and fault data as described in any one of claims 1 to 6.

Images