Intelligent warehousing and stacking system for plate warehouse

Abstract

The invention relates to the technical field of industrial visual intelligent stacking, in particular to an intelligent warehousing and stacking system for a plate warehouse. According to the method, firstly, a visual monitoring module is used for obtaining position information data of all structural feature points of each stacked plate, and a data basis is provided for analyzing the stacking abnormal state; further, in an intelligent analysis module, in combination with an anomaly score obtained by an anomaly detection algorithm and the overall deviation degree of the structural feature points, obtaining the confidence anomaly degree of each structural feature point, and obtaining an anomaly coefficient; furthermore, in an early warning and adjusting module, abnormal structure feature points are screened, early warning is given out, and stacking is adjusted. According to the method, the abnormal score and the deviation degree are mutually verified, the confidence abnormal degree is obtained, the accurate abnormal coefficient is obtained accordingly, early warning is given out in time, stacking adjustment is carried out, the safety during intelligent warehousing and stacking of the plate warehouse is guaranteed, and the conditions of plate damage, stacking collapse and the like are prevented.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial visual intelligent stacking, and in particular to an intelligent storage and stacking system for a plate warehouse. Background Art

[0002] In the intelligent storage and stacking scenario of plate warehouses, automated plate stacking equipment can be used to adapt to plates of different sizes and weights, automatically remove the plates from the conveying equipment and stack them in appropriate locations. However, during the top-down stacking process, each stacking of plates will often produce a certain stacking force, which affects the stability of the plate stacking and easily causes deformation, damage, or even collapse of the stack.

[0003] For the historical data of stacking of the same type of plates, the correlation between the stacking height and the stacking stability shows a fluctuation characteristic within a certain range, and the degree of abnormality of the structural feature points of the plates at different positions in the stack is often related to the overall skewness of the stack. When using the traditional anomaly detection algorithm to extract the abnormal points of the structural feature points of the stacked plates, the accuracy of the extraction is affected by the overall skewness of the stack, resulting in inaccurate results of the structural feature points of the abnormal stacked plates, making it difficult to accurately and effectively analyze the abnormal plate positions, and unable to accurately adjust the stacking shape, which is not conducive to improving the safety of intelligent storage and stacking of plate warehouses. Summary of the invention

[0004] In order to solve the technical problem that the abnormal state of stacking is not accurately identified and affects the safety of storage stacking, the purpose of the present invention is to provide an intelligent storage stacking system for plate warehouses. The technical solutions adopted are as follows:

[0005] The visual monitoring module obtains structural feature points at a preset position of each stacked plate; after each stacking operation, the position information data of all the structural feature points of each stacked plate are obtained; the position information data are the coordinate values ​​of the structural feature points in space;

[0006] The intelligent analysis module obtains the offset of each structural feature point at each sampling moment according to the difference of the position information data between adjacent sampling moments of the structural feature point; obtains the offset significance of each structural feature point at each sampling moment according to the difference characteristics of the offset of different structural feature points at the same sampling moment; obtains the offset degree of each structural feature point at each sampling moment according to the difference characteristics of the offset significance at adjacent sampling moments; obtains the anomaly score of each structural feature point at each sampling moment by using a preset anomaly detection algorithm; obtains the confidence anomaly degree of each structural feature point according to the related characteristics of all the anomaly scores and all the offset degrees of each structural feature point; adjusts the anomaly score according to the confidence anomaly degree of each structural feature point to obtain the anomaly coefficient of each structural feature point;

[0007] The early warning and adjustment module selects the abnormal coefficient according to a preset abnormal early warning threshold, obtains abnormal structural feature points, issues an early warning and adjusts the stacked plates of the abnormal structural feature points.

[0008] Furthermore, the method for obtaining the deviation significance degree includes:

[0009] Any one of the structural feature points is selected as the structural feature point to be analyzed; at the same sampling time, the mean of the offsets of all other structural feature points except the structural feature point to be analyzed is used as an offset reference parameter, and the absolute value of the difference between the offset of the structural feature point to be analyzed and the offset reference parameter is used as the offset significance of the structural feature point to be analyzed.

[0010] Furthermore, the method for obtaining the offset degree includes:

[0011] The absolute value of the difference between the deviation significance level of the structural feature point at the current sampling moment and the deviation significance level at the adjacent previous sampling moment is normalized and used as the deviation degree of the structural feature point at the current sampling moment.

[0012] Furthermore, the method for obtaining the anomaly score includes:

[0013] At each sampling moment, the mean offset of all the structural feature points at each preset position is used as the first parameter of the structural feature point at each preset position; the absolute value of the difference between the offset of the structural feature point and the corresponding first parameter is used as the second parameter of each structural feature point; and according to the second parameter of the structural feature point, the isolation forest algorithm is used to obtain the abnormal score of the structural feature point.

[0014] Furthermore, the method for obtaining the confidence abnormality degree includes:

[0015] All the deviation degrees of each of the structural feature points constitute a deviation degree sequence of each of the structural feature points, and all the abnormal scores of each of the structural feature points constitute an abnormal score sequence of each of the structural feature points;

[0016] The confidence anomaly degree calculation formula is used to obtain the confidence anomaly degree of each structural feature point; the confidence anomaly degree calculation formula includes:

[0017] Among them, YB i Indicates the confidence abnormality of the i-th structural feature point; exp() represents an exponential function with the natural constant e as the base; {PS i} represents the offset degree sequence of the i-th structural feature point; {C i} represents the abnormal score sequence of the i-th structural feature point; SK() represents the skewness function; (ΔC i ) l，l-1 Represents the difference in anomaly scores between the lth sampling time and the l-1th sampling time of the i-th structural feature point, (Δn) l，l-1 represents the difference in the number of structural feature points between the lth sampling moment and the l-1th sampling moment of the i-th structural feature point; p represents the sampling times of the i-th structural feature point.

[0018] Furthermore, the method for obtaining the abnormal coefficient includes:

[0019] The abnormal coefficient calculation formula is used to obtain the abnormal coefficient of each structural feature point; the abnormal coefficient calculation formula includes:

[0020] Among them, C i ′ represents the abnormal coefficient of the i-th structural feature point; YB i Indicates the confidence abnormality degree of the i-th structural feature point; Represents the mean of the confidence abnormality of all structural feature points; i} represents the abnormal score sequence of the i-th structural feature point; norm[] represents the normalization function; max() represents the maximum value function; avg() represents the average value function.

[0021] Furthermore, the method for obtaining the offset includes:

[0022] The Euclidean distance of the position information data between the current sampling moment of the structural feature point and the previous adjacent sampling moment is used as the offset of the structural feature point corresponding to the current sampling moment.

[0023] Furthermore, the method for acquiring the abnormal structure feature points includes:

[0024] The structural feature point whose abnormal coefficient is greater than the preset abnormal warning threshold is taken as an abnormal structural feature point.

[0025] Furthermore, the preset abnormal warning threshold is 0.92.

[0026] Furthermore, the method for acquiring the structural feature points includes:

[0027] The preset positions are four vertices of the bottom surface of each stacked plate, and the four vertices of the bottom surface of each stacked plate are used as structural feature points.

[0028] The present invention has the following beneficial effects:

[0029] The present invention first obtains the position information data of all structural feature points of each stacked plate to provide a data basis for analyzing the abnormal state of stacking; further obtains the offset of each structural feature point at each sampling moment, obtains the offset significance degree from the difference of different structural points at the same sampling moment, and analyzes the difference in the offset significance degree at adjacent sampling moments, and obtains the offset degree of each structural feature point at each sampling moment from the perspective of the overall stacking skew, so as to provide a basis for the subsequent evaluation of the abnormal score; further, obtains the abnormal score by using a preset abnormality detection algorithm, and then obtains the confidence abnormality degree of each structural feature point by combining the abnormal score obtained by the abnormality detection algorithm and the overall offset degree of the structural feature point, accurately evaluates the credibility of the abnormal score, and prepares for obtaining an accurate abnormality coefficient; further adjusts the abnormal score according to the confidence abnormality degree of each structural feature point, obtains the abnormality coefficient of each structural feature point, and accurately evaluates the abnormality degree of the structural feature point; further screens out abnormal structural feature points, issues an early warning and adjusts the stacking, so as to improve the safety of intelligent storage stacking in the warehouse. The present invention verifies the anomaly score and the degree of deviation with each other, obtains the confidence degree of anomaly and thereby obtains the accurate anomaly coefficient, issues early warning in time and makes stacking adjustments, thereby ensuring the safety of the intelligent storage and stacking of the plate warehouse and preventing damage to the plate and stacking collapse. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0031] Figure 1 A system block diagram of an intelligent storage and stacking system for plate warehouses provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation method, structure, features and effects of a plate warehouse intelligent storage and stacking system proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0033] Unless defined otherwise, 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 belongs.

[0034] The specific scheme of the intelligent storage and stacking system for plate warehouse provided by the present invention is described in detail below with reference to the accompanying drawings.

[0035] See also Figure 1 , which shows a system block diagram of an intelligent storage and stacking system for a plate warehouse provided by an embodiment of the present invention. The system includes a visual monitoring module 101, an intelligent analysis module 102, and an early warning and adjustment module 103.

[0036] The visual monitoring module 101 obtains structural feature points at a preset position of each stacked plate; after each stacking operation, obtains position information data of all structural feature points of each stacked plate; the position information data is the coordinate value of the structural feature point in space.

[0037] By monitoring the spatial position of the structural feature points of the plate, the stability of the stack can be evaluated, the height of the stack can be controlled, and the misalignment can be monitored, which is of great significance for maintaining the stability of the plate stack and protecting the plate. Therefore, in an embodiment of the present invention, the visual monitoring module is used to obtain the position information data of each structural feature point after each stacking operation.

[0038] Considering that the common form of a plate is a cuboid, the vertices of a cuboid plate can accurately reflect the morphological characteristics of the plate. Therefore, in a preferred embodiment of the present invention, the preset positions are the four vertices of the bottom surface of each stacked plate, and the four vertices of the bottom surface of each stacked plate are used as structural feature points. Each time a new plate is added to the top of the stack is regarded as a stacking operation, and the existing plates in the stack are called stacked plates. After each stacking operation, the position information data of all the structural feature points of each stacked plate is obtained to provide data support for subsequent analysis of abnormal states of the stack, issuing warnings and making adjustments.

[0039] It should be noted that multiple industrial cameras are used to acquire images of structural feature points, and then the image data is analyzed and three-dimensional reconstruction is performed using stereo vision technology or multi-view geometry analysis and other methods, so as to obtain the coordinate value of each structural feature point in space as position information data. In one embodiment of the present invention, the bottom surface of the stacking board closest to the ground is taken as the xoy plane, and any side surface of the stacking board closest to the ground is taken as the xoz plane to obtain the position information data of the structural feature points of the stacking board. This is a technical means well known to those skilled in the art and will not be elaborated here.

[0040] The intelligent analysis module 102 obtains the offset of each structural feature point at each sampling moment according to the difference in position information data between adjacent sampling moments of the structural feature point; obtains the offset significance of each structural feature point at each sampling moment according to the difference characteristics of the offsets of different structural feature points at the same sampling moment; obtains the offset degree of each structural feature point at each sampling moment according to the difference characteristics of the offset significance of adjacent sampling moments; obtains the anomaly score of each structural feature point at each sampling moment using a preset anomaly detection algorithm; obtains the confidence anomaly degree of each structural feature point according to the relevant characteristics of all anomaly scores and all offset degrees of each structural feature point; adjusts the anomaly score according to the confidence anomaly degree of each structural feature point to obtain the anomaly coefficient of each structural feature point.

[0041] In the embodiment of the present invention, each time a new plate is added to the stack, it will have a certain impact on the existing stacked plates, and the spatial position of the structural feature points will fluctuate to a certain extent, affecting the stability of the stack. In order to measure the position fluctuation characteristics of the structural feature points, the abnormal coefficient of each structural feature point is obtained.

[0042] When traditional anomaly detection algorithms analyze the abnormal characteristics of structural feature points and obtain the abnormal score of each structural feature point at each sampling moment, they are easily affected by the overall skewness of the stack, resulting in the abnormal score of the structural feature point being unable to accurately characterize the fluctuation characteristics of the stacked plate position information data. Therefore, the offset degree is obtained by combining the offset characteristics of the overall structural feature points, and then the related characteristics of the abnormal score and the offset degree are analyzed to obtain the confidence abnormality degree, provide a credibility weight for the abnormal score, and prepare for accurately adjusting the abnormal score; finally, adjust the abnormal score according to the confidence abnormality degree, obtain the abnormal coefficient, and accurately evaluate the abnormal state of the structural feature point, so as to prepare for the subsequent system to perform related operations according to the abnormal coefficient of the structural feature point to improve the safety of stacking.

[0043] Considering that in order to obtain the offset degree of the structural feature points, it is first necessary to obtain the offset of the structural feature points. The changes in the position information data at adjacent sampling moments can reflect the real-time offset fluctuations of the structural feature points. Therefore, the offset of each structural feature point at each sampling moment is obtained according to the difference characteristics of the position information data between adjacent sampling moments of the structural feature points.

[0044] Preferably, in one embodiment of the present invention, considering that the fluctuation of the position information data of the structural feature point corresponds to the stability of the stack, even if a certain offset occurs and then the original position is restored, it also represents that the stack is in an unstable state. Therefore, the Euclidean distance of the position information data between the current sampling moment of the structural feature point and its adjacent previous sampling moment is used as the offset of the structural feature point corresponding to the current sampling moment.

[0045] Since the plates are subjected to pressure from the upper plates when stacked, and the plates have certain elastic characteristics and will produce certain deformations when subjected to pressure, the offset cannot accurately represent the offset degree of the structural feature points. However, the overall offset at the same sampling time should be roughly the same. When the plate is abnormally offset, it shows a more significant feature relative to the overall offset. Therefore, according to the difference characteristics of the offsets of different structural feature points at the same sampling time, the offset significance of each structural feature point at each sampling time is obtained. The greater the offset significance, the greater the abnormality of the structural feature point at the corresponding sampling time. However, the continuous stacking of plates will cause the continuous fluctuation of the position of the structural feature points, and the offset significance will also fluctuate. The smaller the fluctuation, the more stable the stacking state of the corresponding structural feature point is when the plates are stacked. The smaller the difference between the offset significance of the structural feature point at the current sampling time and the offset significance at the previous sampling time, and the more similar it is to the difference value with other structural points, the smaller the offset degree of the structural point at the current sampling time is.

[0046] Preferably, in one embodiment of the present invention, taking into account that the mean can represent the overall characteristics of the data and is less affected by outliers, the mean is selected as a statistical characteristic value to measure the significance of the offset of the structural feature points. Based on this, any one structural feature point is selected as the structural feature point to be analyzed; at the same sampling time, the mean of the offsets of all other structural feature points except the structural feature point to be analyzed is used as the offset reference parameter, and the absolute value of the difference between the offset of the structural feature point to be analyzed and the offset reference parameter is used as the significance of the offset of the structural feature point to be analyzed.

[0047] It should be noted that, in other embodiments of the present invention, the implementer may also select other statistical characteristic values ​​such as the median to represent the central tendency of the data as the offset reference parameter.

[0048] Preferably, in one embodiment of the present invention, considering that the difference in the offset significance between adjacent sampling moments can reflect the rate of change of the offset significance, the greater the rate of change, the greater the impact on the corresponding structural feature point during the stacking process, and the lower the stability, so the absolute value of the difference between the offset significance of the structural feature point at the current sampling moment and the offset significance of the adjacent previous sampling moment is normalized as the offset degree of the structural feature point at the current sampling moment, which is expressed by the formula:

[0049] PS il =Max_Min{|f i (l)-f i (l-1)|}

[0050] Among them, PS il represents the degree of deviation of the i-th structural feature point at the l-th sampling moment; f i (l) represents the deviation significance of the i-th structural feature point at the l-th sampling moment; fi(l-1) represents the deviation significance of the i-th structural feature point at the l-1-th sampling moment; Max_Min{} represents the maximum and minimum value normalization function.

[0051] In the calculation formula of the degree of deviation, the greater the difference in the degree of deviation between adjacent sampling moments, the more obvious the fluctuation of the position information data of the structural feature points, the worse the stability, and the greater the degree of deviation.

[0052] It should be noted that in other embodiments of the present invention, other basic mathematical operations or function mappings may be used to implement relevant mappings, and other normalization functions may be used for processing. These are technical means well known to those skilled in the art and will not be elaborated here.

[0053] After obtaining the degree of deviation of the structural feature points at each sampling moment, the degree of abnormality of the structural feature points is evaluated from the perspective of the overall skew characteristics, which provides a basis for the subsequent credibility judgment of the anomaly scores obtained by the anomaly detection algorithm. Therefore, the anomaly detection algorithm is used to obtain the anomaly score of each structural feature point at each sampling moment.

[0054] Preferably, in one embodiment of the present invention, considering that the isolation forest algorithm can quickly and effectively analyze abnormal data and performs well on large-scale data, the isolation forest algorithm is used to obtain the abnormal score of the structural feature point at each sampling moment. Based on this, at each sampling moment, the mean offset of all structural feature points at each preset position is used as the first parameter of the structural feature point at each preset position; the absolute value of the difference between the offset of the structural feature point and the corresponding first parameter is used as the second parameter of each structural feature point; according to the second parameter of the structural feature point, the isolation forest algorithm is used to obtain the abnormal score of the structural feature point. Taking a sampling moment as an example, each stacked plate has structural feature points at four positions A, B, C, and D, then the mean offset corresponding to the structural feature points of all stacked plates at position A is used as the first parameter of the structural feature point at position A, and the absolute value of the difference between the offset of each structural feature point and the corresponding first parameter is used as the second parameter of each structural feature point; according to the second parameter of each structural feature point, the isolation forest algorithm is used to obtain the abnormal score of each structural feature point. Taking a sampling moment as an example, each stacked plate has structural feature points at four positions A, B, C, and D, then the mean offset corresponding to the structural feature points of all stacked plates at position A is used as the first parameter of the structural feature point at position A, and the absolute value of the difference between the offset of each structural feature point and the corresponding first parameter is used as the second parameter of each structural feature point; according to the second parameter of each structural feature point, the isolation forest algorithm is used to obtain the abnormal score of each structural feature point.

[0055] It should be noted that obtaining anomaly scores by using the anomaly score calculation formula in the isolation forest algorithm is a technical means well known to those skilled in the art and will not be described in detail here. In other embodiments of the present invention, implementers may also use other anomaly detection algorithms such as the LOF algorithm to obtain the anomaly scores of structural feature points at each sampling moment.

[0056] In the embodiment of the present invention, considering that the more similar the fluctuations of the anomaly score and the degree of deviation with the sampling time are, the stronger the correlation is, which means that the anomaly analysis of the structural feature point is more accurate and the credibility is higher, the anomaly score and the degree of deviation are mutually verified, and the confidence anomaly degree of each structural feature point is obtained according to the relevant features of all anomaly scores and all deviation degrees of each structural feature point.

[0057] Preferably, in one embodiment of the present invention, considering that skewness can measure the distribution characteristics of data, the more similar the skewness is, the more similar the distribution characteristics of the data sets are; at the same time, considering that as the plates in the stack are continuously stacked, the number of structural feature points is constantly changing, which has a certain impact on the analysis of the isolated forest algorithm, so corrections are made according to the difference in the number of structural feature points and the difference in anomaly scores. Based on this, all the offset degrees of each structural feature point constitute the offset degree sequence of each structural feature point, and all the anomaly scores of each structural feature point constitute the anomaly score sequence of each structural feature point; the confidence anomaly degree calculation formula is used to obtain the confidence anomaly degree of each structural feature point; the confidence anomaly degree calculation formula includes:

[0058]

[0059] Among them, YB iIndicates the confidence abnormality of the i-th structural feature point; exp() represents an exponential function with the natural constant e as the base; {PS i} represents the offset degree sequence of the i-th structural feature point; {C i} represents the abnormal score sequence of the i-th structural feature point; SK() represents the skewness function; (ΔC i ) l，l-1 Represents the difference in anomaly scores between the lth sampling time and the l-1th sampling time of the i-th structural feature point, (Δn) l，l-1 represents the difference in the number of structural feature points between the lth sampling moment and the l-1th sampling moment of the i-th structural feature point; p represents the sampling times of the i-th structural feature point.

[0060] In the confidence anomaly degree calculation formula, the smaller the skewness difference between the anomaly score sequence and the offset degree sequence, the greater the correlation between the two sequences, and the greater the confidence anomaly degree; the greater the cumulative impact of the change in the number of structural feature points on the anomaly score value, the greater the impact of the change in the number of structural feature points on the anomaly score value, the greater the impact of the change in the number of structural feature points on the anomaly score, the smaller the skewness difference, and the greater the confidence anomaly degree.

[0061] It should be noted that, since the stacking is continuously stacking plates, the number of stacked plates is increasing, the number of structural feature points is also increasing, and the difference in the number of structural feature points at adjacent sampling moments cannot be zero; in one embodiment of the present invention, since the offset, the degree of offset and the anomaly score of the structural feature point are all obtained by the difference between a certain sampling moment and its previous sampling moment, before the first sampling moment, the corresponding plate has not been stacked on the stack, and therefore the offset, degree of offset and anomaly score of each structural feature point at the first sampling moment are all zero; in other embodiments of the present invention, when calculating the confidence anomaly degree, other basic mathematical operations or function mappings may also be used to implement relevant mappings, and other normalization functions may be used for processing, which are all technical means well known to those skilled in the art and will not be elaborated herein.

[0062] After obtaining the confidence anomaly degree of the structural feature points, the anomaly score can be adjusted according to the confidence anomaly degree of each structural feature point, the accurate anomaly coefficient of each structural feature point can be obtained, the structural feature points can be accurately evaluated, and accurate data basis can be provided for subsequent early warning and adjustment operations.

[0063] Preferably, in one embodiment of the present invention, considering that the mean can represent the overall characteristics of the data and is less affected by outliers, the average confidence anomaly degree is used as the judgment criterion, and the average anomaly score is used as the basis. When the confidence anomaly degree of the structural feature point is less than the average confidence anomaly degree of all structural feature points, it means that the credibility of the anomaly score of the structural feature point is low, and the average anomaly score is normalized as the anomaly coefficient of the corresponding structural feature point; when the confidence anomaly degree of the structural feature point is greater than or equal to the average confidence anomaly degree of all structural feature points, it means that the credibility of the anomaly score of the structural feature point is high, so the maximum value of the anomaly score is normalized as the anomaly coefficient of the corresponding structural feature point, based on this, an anomaly coefficient calculation formula is constructed, and the anomaly coefficient calculation formula is used to obtain the anomaly coefficient of each structural feature point; the anomaly coefficient calculation formula includes:

[0064]

[0065] Among them, C i ′ represents the abnormal coefficient of the i-th structural feature point; YB i Indicates the confidence abnormality degree of the i-th structural feature point; Represents the mean of the confidence abnormality of all structural feature points; i} represents the abnormal score sequence of the i-th structural feature point; norm[] represents the normalization function; max() represents the maximum value function; avg() represents the average value function.

[0066] In the anomaly coefficient calculation formula, when the confidence anomaly degree of a structural feature point is less than the average confidence anomaly degree of all structural feature points, it means that the credibility of the anomaly score of the structural feature point is low. The average anomaly score is normalized and used as the anomaly coefficient of the corresponding structural feature point to reduce the impact of abnormal data points in the anomaly score; when the confidence anomaly degree of a structural feature point is greater than or equal to the average confidence anomaly degree of all structural feature points, it means that the credibility of the anomaly score of the structural feature point is high. From a safety perspective, the peak value of the anomaly score corresponds to the moment when the stacking stability is the worst, so the maximum value of the anomaly score is normalized and used as the anomaly coefficient of the corresponding structural feature point to avoid accidents.

[0067] The early warning and adjustment module 103 selects abnormal coefficients according to a preset abnormal early warning threshold, obtains abnormal structural feature points, issues an early warning, and adjusts the stacked plates of the abnormal structural feature points.

[0068] After being processed by the intelligent analysis module 102, the accurate abnormal coefficient of each structural feature point is obtained, providing a reliable data basis for the system to perform prediction and adjustment operations. In an embodiment of the present invention, the larger the abnormal coefficient, the greater the abnormal possibility of the structural feature point. Therefore, the abnormal coefficient is screened according to the preset abnormal warning threshold to obtain abnormal structural feature points. In one embodiment of the present invention, the preset abnormal warning threshold is set to 0.92, and the structural feature points with abnormal coefficients greater than the preset abnormal warning threshold are used as abnormal structural feature points. When abnormal structural feature points appear, the corresponding plate and its abnormal structural feature points are displayed on the display screen through visual operations, and an early warning notification is issued to remind relevant personnel of the abnormality, and the intelligent and precise adjustment of the plate stacking position is performed using automatic or semi-automatic stacking equipment to improve the safety of intelligent stacking of plate warehouses.

[0069] In other embodiments of the present invention, implementers may set other appropriate preset abnormal warning thresholds according to actual application scenarios.

[0070] In summary, the present invention aims at the technical problem that the abnormal state of stacking is not accurately identified, which affects the safety of storage stacking, and proposes an intelligent storage stacking system for plate warehouses. The present invention first uses the visual monitoring module 101 to obtain the position information data of all structural feature points of each stacked plate, providing a data basis for analyzing the abnormal state of stacking; further in the intelligent analysis module 102, the abnormal score obtained by the abnormal detection algorithm and the overall offset degree of the structural feature point are combined to obtain the confidence abnormality degree of each structural feature point, and obtain the abnormality coefficient; further in the early warning and adjustment module 103, the abnormal structural feature points are screened, an early warning is issued, and the stacking is adjusted. The present invention verifies the abnormal score and the offset degree with each other, obtains the confidence abnormality degree and thereby obtains the accurate abnormality coefficient, issues an early warning in time and performs stacking adjustments, ensures the safety of the intelligent storage stacking of the plate warehouse, and prevents damage to the plate and collapse of the stack.

[0071] It should be noted that the sequence of the above embodiments of the present invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0072] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An intelligent storage and stacking system for plate warehouses, characterized in that: The system comprises: The visual monitoring module obtains structural feature points at a preset position of each stacked plate; after each stacking operation, the position information data of all the structural feature points of each stacked plate are obtained; the position information data are the coordinate values ​​of the structural feature points in space; The intelligent analysis module obtains the offset of each structural feature point at each sampling moment according to the difference of the position information data between adjacent sampling moments of the structural feature point; obtains the offset significance of each structural feature point at each sampling moment according to the difference characteristics of the offset of different structural feature points at the same sampling moment; obtains the offset degree of each structural feature point at each sampling moment according to the difference characteristics of the offset significance at adjacent sampling moments; obtains the anomaly score of each structural feature point at each sampling moment by using a preset anomaly detection algorithm; obtains the confidence anomaly degree of each structural feature point according to the related characteristics of all the anomaly scores and all the offset degrees of each structural feature point; adjusts the anomaly score according to the confidence anomaly degree of each structural feature point to obtain the anomaly coefficient of each structural feature point; The early warning and adjustment module selects the abnormal coefficient according to a preset abnormal early warning threshold, obtains abnormal structural feature points, issues an early warning and adjusts the stacked plates of the abnormal structural feature points.

2. According to the intelligent storage and stacking system for plate warehouses described in claim 1, it is characterized in that: The method for obtaining the deviation significance degree includes: Any one of the structural feature points is selected as the structural feature point to be analyzed; at the same sampling time, the mean of the offsets of all other structural feature points except the structural feature point to be analyzed is used as an offset reference parameter, and the absolute value of the difference between the offset of the structural feature point to be analyzed and the offset reference parameter is used as the offset significance of the structural feature point to be analyzed.

3. According to the intelligent storage and stacking system for plate warehouses as described in claim 1, it is characterized in that: The method for obtaining the offset degree includes: The absolute value of the difference between the deviation significance level of the structural feature point at the current sampling moment and the deviation significance level at the adjacent previous sampling moment is normalized and used as the deviation degree of the structural feature point at the current sampling moment.

4. According to the intelligent storage and stacking system for plate warehouses as described in claim 1, it is characterized in that: The method for obtaining the anomaly score includes: At each sampling moment, the mean offset of all the structural feature points at each preset position is used as the first parameter of the structural feature point at each preset position; the absolute value of the difference between the offset of the structural feature point and the corresponding first parameter is used as the second parameter of each structural feature point; and according to the second parameter of the structural feature point, the isolation forest algorithm is used to obtain the abnormal score of the structural feature point.

5. According to the intelligent storage and stacking system for plate warehouses as described in claim 1, it is characterized in that: The method for obtaining the confidence abnormality degree includes: All the deviation degrees of each of the structural feature points constitute a deviation degree sequence of each of the structural feature points, and all the abnormal scores of each of the structural feature points constitute an abnormal score sequence of each of the structural feature points; The confidence anomaly degree calculation formula is used to obtain the confidence anomaly degree of each structural feature point; the confidence anomaly degree calculation formula includes: Among them, YB i represents the confidence abnormality of the i-th structural feature point; exp( ) represents the exponential function with the natural constant e as the base; {PS i } represents the offset degree sequence of the i-th structural feature point; {C i } represents the abnormal score sequence of the i-th structural feature point; SK( ) represents the skewness function; (ΔC i ) l,l-1 Represents the difference in anomaly scores between the lth sampling time and the l-1th sampling time of the i-th structural feature point, (Δn) l,l-1 represents the difference in the number of structural feature points between the lth sampling moment and the l-1th sampling moment of the i-th structural feature point; p represents the sampling times of the i-th structural feature point.

6. According to the intelligent storage and stacking system for plate warehouses as described in claim 5, it is characterized in that: The method for obtaining the abnormal coefficient includes: The abnormal coefficient calculation formula is used to obtain the abnormal coefficient of each structural feature point; the abnormal coefficient calculation formula includes: Among them, C i ' represents the abnormal coefficient of the i-th structural feature point; YB i Indicates the confidence abnormality degree of the i-th structural feature point; Represents the mean of the confidence abnormality of all structural feature points; i } represents the abnormal score sequence of the i-th structural feature point; norm[ ] represents the normalization function; max( ) represents the maximum value function; avg( ) represents the average value function.

7. According to the intelligent storage and stacking system for plate warehouses as described in claim 1, it is characterized in that: The method for obtaining the offset includes: The Euclidean distance of the position information data between the current sampling moment of the structural feature point and the previous adjacent sampling moment is used as the offset of the structural feature point corresponding to the current sampling moment.

8. According to the intelligent storage and stacking system for plate warehouses as described in claim 1, it is characterized in that: The method for acquiring the abnormal structure feature points comprises: The structural feature point whose abnormal coefficient is greater than the preset abnormal warning threshold is taken as an abnormal structural feature point.

9. According to the intelligent storage and stacking system for plate warehouses as claimed in claim 8, it is characterized in that: The preset abnormal warning threshold is 0.

92.

10. The intelligent storage and stacking system for plate warehouse according to claim 1, characterized in that: The method for acquiring the structural feature points comprises: The preset positions are four vertices of the bottom surface of each stacked plate, and the four vertices of the bottom surface of each stacked plate are used as structural feature points.

Images