Internet of Things data acquisition and analysis system for stacking production line

Abstract

The invention belongs to the technical field of industrial automation and Internet of Things, and provides an Internet of Things data acquisition and analysis system for a stacking production line, which comprises the following steps: dividing historical grabbing into normal grabbing and deviation grabbing, determining a deviation threshold value, and performing consistency analysis on deviation characteristics and resonance characteristics of deviation grabbing, so as to obtain a data acquisition result of the stacking production line; whether deviation grabbing is caused by resonance type vibration or not is judged, resonance deviation grabbing is recognized in deviation grabbing, statistical analysis is conducted on the resonance type deviation grabbing, whether the resonance deviation phenomenon exists or not is judged, if yes, multiple resonance deviation risk groups are obtained through clustering, and the relative weight difference of materials and the conveying belt speed critical value are determined; and finally, through real-time comparison and pre-judgment of risks and real-time coordinate correction, accurate recognition, rule quantification and active counteracting of grabbing deviation caused by resonance are achieved, the empty grabbing and deviation grabbing rate of a stacking production line is reduced, and the stacking efficiency and operation stability of an automatic stereoscopic warehouse are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of industrial automation and Internet of Things, and in particular to an Internet of Things data acquisition and analysis system for a palletizing production line. BACKGROUND

[0002] With the rapid development of logistics automation, automated storage and retrieval systems have become the core facilities of modern warehousing due to their high space utilization and fast operation efficiency. As a key link of automated storage and retrieval systems, the operation precision and stability of the palletizing production line directly determine the overall operation efficiency of the warehouse. Currently, the palletizing production line of an automated storage and retrieval system generally adopts a visual positioning sensor + mechanical arm grabbing mode. The visual sensor collects the center coordinates of the materials based on the assumption of uniform motion of the conveyor belt, guiding the mechanical arm to complete grabbing and stacking.

[0003] However, the motion state of the conveyor belt is easily affected by the weight change of the materials in actual operation: when the relative weight difference between adjacent materials is significant, the motor load will change suddenly, causing the conveyor belt to decelerate or accelerate instantaneously; this load change further causes the friction frequency between the conveyor belt and the roller to change, triggering resonance vibration. Unlike ordinary random vibration, this resonance vibration exhibits a coupled motion of forward and backward movement + left and right deviation, rather than up and down shaking, resulting in a deviation between the coordinates collected by the visual sensor and the actual position of the materials.

[0004] The prior art has obvious deficiencies in addressing the above problems: first, there is a lack of accurate identification means for resonance vibration, which often misjudges the deviation caused by resonance as a visual sensor failure or material deformation, leading to incorrect problem positioning; second, there is no correlation between the relative weight difference of the materials, the speed of the conveyor belt and the resonance deviation, making it impossible to provide early warning for high-risk working conditions; third, there is a lack of real-time coordinate correction mechanism based on historical rules, which can only be adjusted passively after grabbing fails, seriously affecting the continuous operation rhythm of the automated storage and retrieval system and increasing the risk of material damage and production line downtime.

[0005] Therefore, the present application provides an Internet of Things data acquisition and analysis system for a palletizing production line. SUMMARY

[0006] To make up for the deficiencies of the prior art and solve at least one of the technical problems raised in the background.

[0007] The technical solution adopted by the present application to solve its technical problems is: an Internet of Things data acquisition and analysis system for a palletizing production line, comprising:

[0008] a deviation threshold determination module: obtaining multiple historical grabbing data of the palletizing production line, dividing the historical grabbing into normal grabbing and deviation grabbing according to the grabbing results, and determining a deviation threshold according to the offset distribution of normal grabbing and deviation grabbing;

[0009] resonance deviation traceability module: by frequency analysis on the deviation capture, it is judged whether the deviation capture exists resonance vibration, if so, the deviation characteristics and resonance characteristics of the deviation capture are analyzed for consistency, it is judged whether the deviation capture is caused by resonance vibration, if so, the deviation capture is marked as resonance deviation capture;

[0010] resonance phenomenon judgment module: statistical analysis is performed on the resonance deviation capture, it is judged whether resonance deviation phenomenon exists when the material is captured by the stacking production line;

[0011] risk law mining module: if so, the relative weight difference of the material of the resonance deviation capture and the conveying belt speed are clustered, a plurality of resonance deviation risk groups are obtained, and the critical value of the relative weight difference of the material and the conveying belt speed that causes the resonance vibration to lead to the deviation reaching the deviation threshold is determined;

[0012] real-time coordinate correction module: by comparing and analyzing the relative weight difference of the current material and the previous material and the critical value of the conveying belt speed, it is judged whether resonance vibration risk exists, if so, the current belonging resonance deviation risk group is determined, and the coordinates collected by the current vision sensor are corrected based on the average deviation of the belonging resonance deviation risk group.

[0013] Further, the determination method of the deviation threshold is:

[0014] For the deviation of normal capture and deviation capture in the conveying belt direction and the direction perpendicular to the conveying belt:

[0015] After sorting the absolute values of the deviation of normal capture from small to large, respectively, the value at the 95% position is taken as the upper limit of safety;

[0016] After sorting the deviation of deviation capture from small to large, the value at the 5% position is taken as the lower limit of failure;

[0017] The candidate threshold range is 95% quantile of normal capture < candidate threshold < 5% quantile of deviation capture;

[0018] For any candidate threshold in the candidate threshold range, the proportion of the deviation amount in the deviation capture greater than or equal to the candidate threshold is calculated to obtain the true positive rate, and the proportion of the deviation amount in the normal capture greater than or equal to the candidate threshold is calculated to obtain the false positive rate;

[0019] Taking the true positive rate as the horizontal axis and the false positive rate as the vertical axis, the ROC curve is drawn, the slope difference of adjacent points is calculated, and the candidate threshold corresponding to the first point meeting the requirement is the deviation threshold.

[0020] Further, the judgment method of whether the deviation capture exists resonance vibration is:

[0021] Calculate the frequency difference between the natural frequency of the conveyor belt and the real-time vibration frequency of the conveyor belt for each deviation. If the frequency difference is less than 1Hz, the frequency matching condition is met.

[0022] The vibration amplitude reference value in the non-resonance state is collected. After the material weight changes, the real-time vibration amplitude is collected, and the ratio of the real-time vibration amplitude to the vibration amplitude reference value is calculated to obtain the amplitude ratio. If the amplitude ratio meets the requirements, it is determined to be an amplitude change.

[0023] Real-time acquisition of conveyor belt speed fluctuations; if the time difference between the moment of weight change, the moment of sudden increase in vibration amplitude, and the moment of sudden change in speed fluctuation all meet the time difference requirement, then the weight change, vibration, and speed fluctuation are synchronized in time.

[0024] If frequency matching, amplitude abrupt change, and time synchronization are all satisfied simultaneously, then resonant vibration exists.

[0025] Furthermore, the method for determining whether the deviation detection is caused by resonant vibration is as follows:

[0026] For any deviation capture that exhibits resonant vibration:

[0027] Calculate the vector pointing from the visual coordinates to the actual position coordinates of the material to obtain the deviation vector. The deviation vector includes the actual position along the conveyor belt's forward direction in the visual coordinates and the actual position to the right of the conveyor belt perpendicular to the visual coordinates.

[0028] Calculate the vibration displacement vector of the conveyor belt when resonance occurs to obtain the resonance vibration vector. After standardizing the deviation vector and the resonance vibration vector into unit vectors, calculate the cosine value of the included angle.

[0029] If the cosine of the included angle is less than the direction similarity threshold, then the directions are determined to be consistent.

[0030] Based on historical resonance deviation data, a linear relationship between the magnitude of the deviation vector and the magnitude of the vibration vector is fitted.

[0031] Calculate the ratio of the magnitude of the deviation vector captured for resonant vibration to the magnitude of the vibration vector. If it is within the range of the historical fitting slope, the magnitude correlation is qualified.

[0032] If the directions are consistent and the magnitudes are correlated, then the deviation in the capture is due to resonant vibration.

[0033] Furthermore, the method for determining whether there is a resonance offset phenomenon during material handling in the palletizing production line is as follows:

[0034] The proportion of resonance-type deviation in historical data is statistically analyzed to obtain the resonance deviation proportion.

[0035] If the resonance deviation ratio meets the requirements, then there is a resonance offset phenomenon.

[0036] Further, the process of clustering the material relative weight difference and the conveyor belt speed is:

[0037] Resonance deviation capture records are obtained as samples, each sample containing two dimensions: the material relative weight difference as the horizontal axis and the conveyor belt speed as the vertical axis;

[0038] The material relative weight difference and the conveyor belt speed of each sample are standardized as sample characteristic values;

[0039] The average silhouette coefficient of different K values is calculated by the silhouette coefficient method, and the K value with the maximum average silhouette coefficient is selected as the clustering number;

[0040] K samples are randomly selected as initial cluster centers, the Euclidean distance of each sample to each initial cluster center is calculated, and the sample is assigned to the cluster group with the nearest Euclidean distance;

[0041] After all samples are assigned to the cluster group, the average value of all sample characteristic values of each cluster group is recalculated as a new cluster center;

[0042] The sample assignment and cluster center update are repeated, the change value of each cluster center is recorded, until the change value meets the requirement, the clustering is completed, and each cluster group is a resonance risk feature group.

[0043] Further, the way of determining the material relative weight difference and the conveyor belt speed critical value is:

[0044] The material relative weight difference and the conveyor belt speed are respectively divided into continuous intervals according to a unit step length, the conveyor belt speed interval and the material relative weight difference interval are fixed respectively, and the material relative weight difference and the conveyor belt speed critical value are determined according to the deviation occurrence rate;

[0045] Finally, the material relative weight difference critical value corresponding to each conveyor belt speed interval and the conveyor belt speed critical value corresponding to each material relative weight difference interval are obtained;

[0046] The material relative weight difference critical values of all conveyor belt speed intervals are summarized, the occurrence frequency of each material relative weight difference critical value is counted, and the mode is taken as the material relative weight difference critical value;

[0047] The conveyor belt speed critical values of all material relative weight difference intervals are summarized, the occurrence frequency of each conveyor belt speed critical value is counted, and the mode is taken as the conveyor belt speed critical value.

[0048] Further, the way of determining the material relative weight difference and the conveyor belt speed critical value is:

[0049] Group all samples by the conveyor belt speed interval, extract the relative weight difference of the material and the corresponding offset occurrence rate;

[0050] For each conveyor belt speed interval group, sort the relative weight difference of the material from small to large, calculate the difference value of the offset occurrence rate between the two continuous relative weight differences of the material, and obtain the jump amplitude;

[0051] The relative weight difference of the material that meets the jump amplitude requirement is the relative weight difference critical value of the material in the conveyor belt speed interval;

[0052] Group all samples by the relative weight difference of the material, extract the conveyor belt speed and the corresponding offset occurrence rate;

[0053] For each relative weight difference of the material group, sort the conveyor belt speed from small to large, and calculate the difference value of the offset occurrence rate between the two continuous conveyor belt speeds;

[0054] If the difference value of the offset occurrence rate meets the requirement, the latter conveyor belt speed is taken as the conveyor belt speed critical value.

[0055] Further, the judgment method of whether there is a resonance type vibration risk is:

[0056] Real-time acquisition of the current material weight and the weight of the previous material, and calculation of the relative weight difference:

[0057] Extract the conveyor belt speed from the system real-time data;

[0058] Compare the current conveyor belt speed and the relative weight difference of the material with the critical value respectively, if the conveyor belt speed is greater than or equal to the conveyor belt speed critical value, and the relative weight difference of the material is greater than or equal to the relative weight difference critical value, there is a resonance type vibration risk.

[0059] Further, the process of correcting the coordinates collected by the current vision sensor is:

[0060] Calculate the Euclidean distance between the current relative weight difference of the material and the conveyor belt speed and the center of each resonance risk group, and the resonance risk feature group with the smallest Euclidean distance is the resonance risk group to which the current relative weight difference of the material and the conveyor belt speed belong;

[0061] Based on the average deviation of the resonance risk feature group, correct the coordinates (X_vision, Y_vision) collected by the current vision sensor:

[0062] The corrected coordinates (X_corrected, Y_corrected) are: X_corrected=X_vision+μX, Y_corrected=Y_vision+μY, where μX is the average deviation of the resonance risk feature group in the X direction, and μY is the average deviation of the resonance risk group in the Y direction.

[0063] The beneficial effects of the present application are as follows: the resonance deviation is accurately identified through frequency-amplitude-synchronization three-dimensional verification and vector similarity analysis, non-resonance factors such as visual failure and material deformation are excluded, the resonance deviation identification accuracy is improved, the positioning error problem in the prior art is solved, the material relative weight difference-conveying belt speed critical value model is established to quantize risk early warning, high resonance risk working conditions are identified in advance, passive response to grabbing failure is avoided, the average offset of the historical risk group is used to correct the coordinates to improve the accuracy in real time, the visual coordinate deviation and the empty and partial grabbing rate are reduced, the high-pace operation of the automatic three-dimensional warehouse is adapted, the clustering model and the critical value are automatically updated, the working conditions such as conveying belt aging and material type change are adapted, manual frequent debugging is not required, the dynamic iteration adaptability is strong, the existing visual sensor and mechanical arm control system of the automatic three-dimensional warehouse can be seamlessly connected, the transformation cost is low, and the popularization is strong. BRIEF DESCRIPTION OF DRAWINGS

[0064] The present application will be further described below with reference to the drawings.

[0065] Figure 1 is a step flow chart of the Internet of Things data acquisition and analysis system for the palletizing production line described in the present application;

[0066] Figure 2 is a logic judgment chart of the Internet of Things data acquisition and analysis system for the palletizing production line described in the present application. DETAILED DESCRIPTION

[0067] In order to make the technical means, creative features, purposes and effects achieved by the present application easy to understand, the present application will be further described below with reference to the specific embodiments.

[0068] Please refer to Figure 1 The Internet of Things data acquisition and analysis system for the palletizing production line described in the present application embodiment includes the following modules:

[0069] The deviation threshold determination module: acquires multiple historical grabbing data of the palletizing production line, divides the historical grabbing into normal grabbing and deviation grabbing according to the grabbing result, and determines the deviation threshold according to the offset distribution of the normal grabbing and the deviation grabbing;

[0070] The process of dividing the historical grabbing into normal grabbing and deviation grabbing according to the grabbing result includes:

[0071] The multiple historical grabbing data of the palletizing production line includes: visual coordinates (material center coordinates collected by a visual positioning sensor), actual position coordinates, grabbing results, material IDs and grabbing time stamps;

[0072] For each historical grabbing, if there is empty or partial grabbing, the historical grabbing is marked as deviation grabbing, and if the grabbing is successful, the historical grabbing is marked as normal grabbing;

[0073] Obtain the weight of the material in the previous grab, and calculate the relative weight difference between the current material weight and the weight of the material in the previous grab:

[0074] If the current material weight is heavier: relative weight difference = current material weight / weight of material in the previous grab;

[0075] If the current material is lighter: relative weight difference = weight of material in the previous grab / current material weight;

[0076] According to the visual coordinates and the actual position coordinates Calculate the deviation of the material in the conveying direction X and the vertical conveying direction Y, and obtain the deviation amount of the visual coordinates in the conveying direction X and the vertical conveying direction Y and :

[0077] , ;

[0078] It should be noted that is positive, indicating that the actual position is ahead of the visual coordinates (in the conveying direction), is positive, indicating that the actual position is to the right of the visual coordinates (perpendicular to the conveying direction);

[0079] The process of determining the deviation threshold value by the offset distribution of normal grabbing and deviation grabbing includes:

[0080] First, the candidate threshold range is determined by statistical quantitative analysis, i.e. the significant dividing point between normal grabbing and deviation grabbing, specifically:

[0081] Sort the absolute values of the conveying direction offset of the normal grabbing from small to large, and take the value at the 95% position as the upper limit of safety;

[0082] Sort the absolute values of the conveying direction offset of the deviation grabbing from small to large, and take the value at the 5% position as the lower limit of failure;

[0083] The candidate threshold range is 95% quantile of normal grabbing < candidate threshold < 5% quantile of deviation grabbing;

[0084] It should be noted that the determination logic of the candidate threshold range is:

[0085] Based on statistical significance and actual risk boundary of the grabbing scene, by quantifying the natural boundary between normal and deviation, the candidate threshold range is ensured to cover the transition interval of safety-risk and avoid threshold failure due to extreme value interference. In normal grabbing, the offset of most samples will not cause failure, so the 95% quantile is taken as the upper limit of safety. In deviation grabbing, the minimum offset is the critical value that just causes failure, so the 5% quantile is taken as the lower limit of failure.

[0086] Secondly, it is necessary to select a specific value from the candidate threshold range and verify it by recognition accuracy to ensure that the deviation threshold can distinguish between normal grabbing and deviation grabbing. Specifically:

[0087] For any candidate threshold in the candidate threshold range, calculate the proportion of deviation grabbing with deviation greater than or equal to the candidate threshold to get the true positive rate, and calculate the proportion of normal grabbing with offset greater than or equal to the candidate threshold to get the false positive rate.

[0088] Draw the ROC curve with the true positive rate as the horizontal axis and the false positive rate as the vertical axis, calculate the slope difference of adjacent points, and the first slope difference that meets the requirements is the balance point of the true positive rate and the false positive rate. The corresponding deviation threshold is the deviation threshold for successful and failed grabbing.

[0089] It can be understood that the balance of the true positive rate and the false positive rate is to find the optimal solution between not missing the deviation grabbing and not misjudging the normal grabbing. It cannot be too high to cause a large number of deviation grabbing to be missed, nor can it be too low to cause a large number of normal grabbing to be misjudged.

[0090] It should be noted that the role of dividing the grabbing type and determining the deviation threshold is to establish the offset mark of successful / failed grabbing through historical data, and to determine how much offset will cause the grabbing to fail, providing a benchmark for subsequent judgment of whether resonance causes excessive offset.

[0091] Resonance deviation tracing module: perform frequency matching analysis on the deviation grabbing to determine whether the deviation grabbing has resonance vibration. If so, perform consistency analysis on the deviation characteristics and resonance characteristics of the deviation grabbing to determine whether the deviation grabbing is caused by resonance vibration. If so, mark the deviation grabbing as resonance deviation grabbing.

[0092] The process of determining whether the deviation grabbing has resonance vibration includes:

[0093] It should be noted that the key difference between resonance vibration and ordinary vibration is that ordinary vibration is random and stable in amplitude, while resonance vibration is triggered by frequency matching, amplified by amplitude mutation, and synchronized with weight difference and speed. Therefore, it needs to be judged comprehensively from three dimensions of frequency, amplitude, and synchronization. Specifically:

[0094] First, the judgment condition of frequency matching is that the vibration frequency is close to the inherent frequency of the system:

[0095] The inherent frequency of the conveying belt system is measured in advance, and when the weight of the material changes, the high-frequency vibration sensor is used to collect the vibration frequency of the conveying belt in real time, and the frequency difference between the inherent frequency of the conveying belt and the real-time collected vibration frequency of the conveying belt is calculated each time;

[0096] If the frequency difference is less than 1Hz, the frequency matching condition is met;

[0097] Second, it is necessary to explain whether the vibration amplitude is significantly mutated and enlarged:

[0098] The vibration amplitude reference value in the non-resonance state is collected, and after the weight of the material changes, the vibration amplitude is collected in real time, and the ratio of the real-time vibration amplitude to the vibration amplitude reference value is calculated to obtain the amplitude ratio;

[0099] The amplitude ratio is compared with the threshold value, and if the amplitude ratio is greater than the threshold value, it is determined that the amplitude is mutated;

[0100] Third, it is necessary to explain the synchronization of vibration-speed fluctuation-weight change:

[0101] The speed fluctuation of the conveying belt is collected in real time, and the time difference between the weight change time, the vibration amplitude sudden increase time and the speed fluctuation mutation time is compared, and if the time difference is less than the preset time difference, the weight change, vibration and speed fluctuation are time synchronized;

[0102] If the frequency matching, amplitude enlargement and time synchronization are met at the same time, it is determined that there is a resonance type vibration;

[0103] Please refer to Figure 2 The process of judging whether the deviation capture is caused by resonance type vibration includes:

[0104] Whether the deviation capture is caused by resonance type vibration is judged by the deviation direction and the deviation amount range, specifically:

[0105] For any deviation capture that exists resonance type vibration:

[0106] The vector from the visual coordinate to the actual position coordinate of the material is calculated to obtain the deviation vector, reflecting the size and direction of the deviation;

[0107] Among them, the deviation vector includes X-axis deviation component and Y-axis deviation component, the direction of the X-axis deviation component is that the actual position is in the visual coordinate along the advancing direction of the conveying belt, and the direction of the Y-axis deviation component is that the actual position is in the visual coordinate perpendicular to the right side of the conveying belt;

[0108] The vibration displacement vector of the conveying belt when resonance occurs is calculated to obtain the resonance vibration vector;

[0109] Calculate the cosine of the angle after normalizing the deviation vector and the resonance vibration vector into unit vectors;

[0110] Based on historical resonance deviation data, set a direction similarity threshold, and if the cosine of the angle is less than the direction similarity threshold, determine that the directions are consistent;

[0111] It should be noted that the deviation caused by resonance is essentially the deviation of the material due to the vibration of the conveying belt, so the size of the deviation vector should be proportional to the size of the vibration vector;

[0112] Based on historical resonance deviation data, fit the linear relationship between the size of the deviation vector and the size of the vibration vector;

[0113] Calculate the ratio of the size of the deviation vector to the size of the vibration vector for the deviation capture caused by resonance vibration, and if it is within the reasonable range of historical fitting k, then the size is consistent;

[0114] If the directions are consistent and the sizes are consistent, the deviation capture is caused by resonance vibration, and the deviation capture is marked as resonance deviation capture;

[0115] It should be noted that the role of identifying deviation capture caused by resonance is to filter out samples caused by resonance vibration from all deviation captures, exclude non-resonance factors such as sensor failure and material deformation, and ensure that subsequent analysis focuses on the specific problem of resonance;

[0116] Resonance phenomenon judgment module: statistically analyze resonance deviation captures to determine whether there is a resonance deviation phenomenon when the palletizing production line captures materials;

[0117] The process of determining whether there is a resonance deviation phenomenon includes:

[0118] Statistically analyze the proportion of resonance deviation captures in historical captures to obtain the resonance deviation proportion;

[0119] Compare the resonance deviation proportion with the preset proportion, and if the resonance deviation proportion is greater than the preset proportion, there is a resonance deviation phenomenon;

[0120] It should be noted that the role of determining whether there is a resonance deviation phenomenon is to verify whether the deviation caused by resonance is a systematic problem that exists universally, rather than an occasional case, and to determine whether resources need to be invested for targeted optimization;

[0121] Risk law mining module: if so, cross-correlate the relative weight difference of the material, the conveying belt speed, and the deviation amount of the resonance deviation capture, and cluster the relative weight difference of the material and the conveying belt speed to obtain multiple resonance deviation risk groups, and determine the critical values of the relative weight difference of the material and the conveying belt speed that cause resonance vibration to cause the deviation amount to reach the deviation threshold;

[0122] The process of clustering the relative weight difference of materials and the conveyor belt speed includes:

[0123] The resonance deviation capture records are used as samples. Each sample contains two dimensions: relative weight difference and conveyor belt speed. The samples are converted into coordinate points. Each sample corresponds to a point in a two-dimensional coordinate system, with relative weight difference as the horizontal axis and conveyor belt speed as the vertical axis.

[0124] The relative weight difference and conveyor belt speed of each sample are standardized and used as the sample feature values;

[0125] The average silhouette coefficient for different K values ​​is calculated using the silhouette coefficient method, and the K with the largest average silhouette coefficient is selected as the number of clusters.

[0126] Initially, K samples are randomly selected as initial cluster centers. The Euclidean distance from each sample to each initial cluster center is calculated, and the sample is assigned to the cluster group with the closest Euclidean distance.

[0127] After all samples are assigned to cluster groups, the average value of all sample feature values ​​in each cluster group is recalculated and used as the new cluster center.

[0128] The distribution of duplicate samples and the updating of cluster centers are recorded. The change value of the cluster centers is recorded each time and compared with a preset threshold. When the change value is less than the preset threshold, the clustering is completed.

[0129] Each cluster group is a resonant risk feature group;

[0130] For each resonance risk feature group, calculate the number of resonance deviation captures within the resonance risk feature group. and The arithmetic mean and the offset occurrence rate, where the offset occurrence rate is the ratio of the number of resonance deviation captures in the resonance risk feature group to the total number of captures in the resonance risk feature group;

[0131] The process of determining the relative weight difference of the material and the critical value of the conveyor belt speed that trigger resonant vibration and cause the offset to reach the offset threshold includes:

[0132] The critical value is located by the point where the offset occurrence rate jumps. Specifically:

[0133] All resonance deviation samples were extracted from historical records, and the relative weight difference, conveyor belt speed, and corresponding ( The total number of grasps and resonance deviations of the V) combination;

[0134] For each ( By combining V and V, the proportion of resonance deviations in the total number of grabs is calculated to obtain the offset occurrence rate;

[0135] The relative weight difference of the material and the conveying belt speed are respectively divided into continuous intervals according to a unit step length, the conveying belt speed interval and the relative weight difference interval are fixed respectively, the critical values of the relative weight difference of the material and the conveying belt speed are determined according to the deviation occurrence rate, and the specific process is as follows:

[0136] Firstly, by analyzing the relationship between the relative weight difference of the material and the deviation occurrence rate in different conveying belt speed intervals, the relative weight difference value of the material at which the deviation occurrence rate suddenly jumps is found, and the specific process is as follows:

[0137] All samples are grouped according to the conveying belt speed interval, and only samples in the conveying belt speed interval are retained in each group, and the relative weight difference of the material and the corresponding deviation occurrence rate are extracted;

[0138] For each conveying belt speed interval group, the relative weight difference of the material is sorted from small to large, and a broken line graph is drawn with the relative weight difference of the material as the horizontal axis and the deviation occurrence rate as the vertical axis;

[0139] The difference between the deviation occurrence rates of two continuous relative weight differences of the material is calculated to obtain the jump amplitude, and the relative weight difference of the material whose jump amplitude is greater than the preset amplitude is the critical value of the relative weight difference of the material in the conveying belt speed interval;

[0140] Finally, the critical value of the relative weight difference of the material corresponding to each conveying belt speed interval is obtained;

[0141] Secondly, by analyzing the relationship between the conveying belt speed and the deviation occurrence rate in different relative weight difference intervals of the material, the conveying belt speed value at which the deviation occurrence rate suddenly jumps is found, and the specific process is as follows:

[0142] All samples are grouped according to the relative weight difference interval of the material, and only samples in the relative weight difference interval of the material are retained in each group, and the conveying belt speed and the corresponding deviation occurrence rate are extracted;

[0143] For each relative weight difference interval group of the material, the conveying belt speed is sorted from small to large, and a broken line graph is drawn with the conveying belt speed as the horizontal axis and the deviation occurrence rate as the vertical axis;

[0144] The difference between the deviation occurrence rates of two continuous conveying belt speeds is calculated, and if the difference between the deviation occurrence rates is greater than or equal to the preset amplitude, the next conveying belt speed is taken as the critical value;

[0145] Finally, the critical value of the conveying belt speed corresponding to each relative weight difference interval of the material is obtained;

[0146] Thirdly, all critical values are integrated, and the mode is taken to determine the final critical value, and the specific process is as follows:

[0147] The critical values of the relative weight difference of the material in all conveying belt speed intervals are summarized, the number of occurrences of each value is counted, and the mode is taken as the critical value of the relative weight difference of the material;

[0148] Aggregate the critical values of the conveyor belt speed for all material relative weight difference intervals, count the number of occurrences of each value, and take the mode as the critical value of the conveyor belt speed;

[0149] It should be noted that the role of mining resonance risk law and determining the critical value is to quantify the correlation between weight difference-speed combination and resonance deviation, and to determine which working conditions the resonance risk will significantly increase, providing an operable critical standard for real-time risk judgment;

[0150] Real-time coordinate correction module: through the comparison and analysis of the relative weight difference between the current material and the previous material and the critical value of the conveyor belt speed, it is determined whether there is a resonance vibration risk, if there is, the current belongs to the resonance deviation risk group is determined, and based on the average deviation of the resonance deviation risk group, the coordinates collected by the current vision sensor are corrected;

[0151] The process of determining whether there is a resonance vibration risk includes:

[0152] Real-time acquisition of the weight of the current material and the weight of the previous material, and calculation of the relative weight difference:

[0153] If the current material is heavier: weight difference = current material weight / previous material weight;

[0154] If the current material is lighter: weight difference = previous material weight / current material weight;

[0155] Extract the conveyor belt speed from the system real-time data;

[0156] According to the current conveyor belt speed and the relative weight difference of the material, it is determined whether there is a resonance vibration risk:

[0157] Compare the current conveyor belt speed and the relative weight difference of the material with the critical value respectively, if the conveyor belt speed is greater than or equal to the critical value of the conveyor belt speed, and the relative weight difference of the material is greater than or equal to the critical value of the relative weight difference, then there is a resonance vibration risk, and the coordinate correction is triggered;

[0158] The process of correcting the coordinates collected by the current vision sensor includes:

[0159] According to the current material relative weight difference and the conveyor belt speed, find the closest resonance risk feature group, and based on the average deviation of the resonance risk feature group, correct the coordinates (X_vision, Y_vision) collected by the current vision sensor:

[0160] Corrected coordinates = vision coordinates + historical average deviation;

[0161] X_corrected = X_vision + μX

[0162] Y_corrected = Y_vision + μY

[0163] It should be noted that the correction logic is that the visual deviation caused by resonance is that the visual coordinates are ahead of or lag behind the actual position ( ), deviate to the left or right ( ), and the historical average deviation μX, μY reflects the systematic deviation direction and size under the working condition, and the offset after adding the correction amount can offset the deviation.

[0164] It should be noted that the functions of real-time risk judgment and coordinate correction are to apply historical laws to actual production, offset the resonance deviation by real-time correction of visual coordinates, upgrade from passive failure handling to active failure prevention, and directly improve the success rate of grabbing.

[0165] The technical scheme and advantages of the embodiment of the application are: obtaining multiple historical grabbing data of the stacking production line, dividing the historical grabbing into normal grabbing and deviation grabbing according to the grabbing results, determining a deviation threshold according to the offset amount distribution of the normal grabbing and the deviation grabbing, judging whether the deviation grabbing exists resonance vibration by frequency analysis on the deviation grabbing, if so, performing consistency analysis on the deviation characteristics and resonance characteristics of the deviation grabbing, judging whether the deviation grabbing is caused by resonance vibration, if so, marking the deviation grabbing as resonance deviation grabbing, performing statistical analysis on the resonance deviation grabbing, judging whether there is resonance offset phenomenon when the stacking production line is grabbing materials, if so, clustering the relative weight difference of the materials and the conveying belt speed of the resonance deviation grabbing, obtaining multiple resonance offset risk groups, and determining the critical values of the relative weight difference of the materials and the conveying belt speed that cause the offset amount to reach the offset threshold due to resonance vibration, judging whether there is resonance vibration risk by comparison and analysis of the relative weight difference of the current material and the previous material and the critical values of the conveying belt speed, if so, determining the current belonging resonance offset risk group, and correcting the coordinates collected by the current vision sensor based on the average deviation of the belonging resonance offset risk group. The application divides the historical grabbing into normal grabbing and deviation grabbing, determines a deviation threshold, and performs consistency analysis on the deviation characteristics and resonance characteristics of the deviation grabbing to judge whether the deviation grabbing is caused by resonance vibration. In the deviation grabbing, resonance deviation grabbing is identified, statistical analysis is performed on the resonance deviation grabbing to judge whether there is resonance offset phenomenon, if so, multiple resonance offset risk groups are obtained by clustering, and the critical values of the relative weight difference of the materials and the conveying belt speed are determined. Finally, the risk is predicted in real time by real-time comparison, and real-time coordinate correction is performed to realize accurate identification, law quantification and active offset of grabbing deviation caused by resonance, reduce the empty grabbing and deviation rate of the stacking production line, and improve the stacking efficiency and operation stability of the automated stereoscopic warehouse.

[0166] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above-mentioned embodiments, and the above-mentioned embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. An Internet of Things data acquisition and analysis system for a palletizing production line, characterized in that: Comprises: Deviation threshold determination module: obtain multiple historical grabbing data of the palletizing production line, divide the historical grabbing into normal grabbing and deviation grabbing according to the grabbing results, and determine the deviation threshold according to the offset distribution of normal grabbing and deviation grabbing; Resonance deviation tracing module: judge whether the deviation grabbing exists resonance vibration by frequency analysis, if so, analyze the consistency of the deviation characteristics and the resonance characteristics of the deviation grabbing, judge whether the deviation grabbing is caused by resonance vibration, if so, mark the deviation grabbing as resonance deviation grabbing; Resonance phenomenon judgment module: statistically analyze the resonance deviation grabbing, and judge whether there is resonance deviation phenomenon when the palletizing production line is grabbing materials; Risk law mining module: if so, cluster the relative weight difference of the material and the conveying belt speed of the resonance deviation grabbing, obtain multiple resonance deviation risk groups, and determine the critical value of the relative weight difference of the material and the conveying belt speed that causes the offset to reach the offset threshold due to resonance vibration; Real-time coordinate correction module: judge whether there is resonance vibration risk by comparing and analyzing the relative weight difference of the current material and the previous material and the critical value of the conveying belt speed, if so, determine the current belonging resonance deviation risk group, and correct the coordinates collected by the current vision sensor based on the average deviation of the belonging resonance deviation risk group.

2. The Internet of Things data acquisition and analysis system for the palletizing production line according to claim 1, characterized in that: The determination method of the deviation threshold is: For the offset of normal grabbing and deviation grabbing in the conveying belt direction and the direction perpendicular to the conveying belt: Sort the absolute values of the offset of normal grabbing in ascending order, and take the value at the 95% position as the upper limit of safety; Sort the offset of deviation grabbing in ascending order, and take the value at the 5% position as the lower limit of failure; The candidate threshold range is 95% quantile of normal grabbing < candidate threshold < 5% quantile of deviation grabbing; For any candidate threshold in the candidate threshold range, calculate the proportion of the deviation amount in the deviation grabbing that is greater than or equal to the candidate threshold to obtain the true positive rate, and calculate the proportion of the offset in the normal grabbing that is greater than or equal to the candidate threshold to obtain the false positive rate; Draw the ROC curve with the true positive rate as the horizontal axis and the false positive rate as the vertical axis, calculate the slope difference of adjacent points, and the candidate threshold corresponding to the first point that meets the requirement is the deviation threshold.

3. The Internet of Things data acquisition and analysis system for the palletizing production line according to claim 1, characterized in that: The judgment method of whether the deviation grabbing exists resonance vibration is: Calculate the frequency difference between the inherent frequency of the conveying belt and the real-time collected vibration frequency of the conveying belt, if the frequency difference is less than 1 Hz, then the frequency matching condition is met; Collect the vibration amplitude reference value in the non-resonance state, after the weight of the material changes, collect the real-time vibration amplitude, calculate the ratio of the real-time vibration amplitude to the vibration amplitude reference value to obtain the amplitude ratio, if the amplitude ratio meets the requirements, then it is determined that the amplitude has suddenly changed; Real-time acquisition of the speed fluctuation of the conveying belt. If the time difference of the weight change moment, the vibration amplitude sudden increase moment and the speed fluctuation mutation moment all meet the time difference requirement, the weight change, vibration and speed fluctuation are time-synchronized; If the frequency matching, amplitude mutation and time synchronization are all met, there is a resonance vibration.

4. The Internet of Things data acquisition and analysis system for the palletizing production line according to claim 3, characterized in that: The judgment method of whether the deviation capture is caused by the resonance vibration is: For any deviation capture with resonance vibration: Calculate the vector from the visual coordinate to the actual position coordinate of the material to obtain the deviation vector, wherein the deviation vector includes the actual position in the visual coordinate along the conveying direction and the actual position in the visual coordinate perpendicular to the right side of the conveying belt; Calculate the vibration displacement vector of the conveying belt when the resonance occurs to obtain the resonance vibration vector. Calculate the cosine of the included angle after the deviation vector and the resonance vibration vector are standardized to unit vectors; If the cosine of the included angle is less than the direction similarity threshold, the directions are determined to be consistent; Based on the historical resonance deviation data, a linear relationship between the size of the deviation vector and the size of the vibration vector is fitted; Calculate the ratio of the size of the deviation vector to the size of the vibration vector for the deviation capture with resonance vibration. If it is within the range of the historical fitting slope, the size is consistent. If the directions are consistent and the sizes are consistent, the deviation capture is caused by the resonance vibration.

5. The Internet of Things data acquisition and analysis system for the palletizing production line according to claim 1, characterized in that: The judgment method of whether the palletizing production line has resonance deviation phenomenon during material capture is: Calculate the proportion of the resonance deviation capture in the historical capture to obtain the resonance deviation proportion; If the resonance deviation proportion meets the requirement, there is a resonance deviation phenomenon.

6. The Internet of Things data acquisition and analysis system for the palletizing production line according to claim 1, characterized in that: The process of clustering the relative weight difference of the material and the conveying belt speed is: Obtain resonance deviation capture records as samples, each sample contains two dimensions: the relative weight difference of the material and the conveying belt speed. The relative weight difference of the material is the horizontal axis, and the conveying belt speed is the vertical axis; Standardize the relative weight difference of the material and the conveying belt speed of each sample as sample characteristic values; Calculate the average contour coefficient of different K values using the contour coefficient method, and select the K value with the maximum average contour coefficient as the clustering number; Initially randomly select K samples as initial clustering centers, calculate the Euclidean distance of each sample to each initial clustering center, and assign the sample to the clustering group with the nearest Euclidean distance; After all samples are assigned to the clustering group, recalculate the average of all sample characteristic values in each clustering group as the new clustering center; Repeat the sample assignment and clustering center update, record the change value of each clustering center, until the change value meets the requirement, the clustering is completed, and each clustering group is a resonance risk feature group.

7. The Internet of Things data acquisition and analysis system for the palletizing production line according to claim 6, characterized in that: The method of determining the critical value of the relative weight difference of the material and the conveying belt speed is: The relative weight difference of the material and the conveying belt speed are respectively divided into continuous intervals according to a unit step length, the conveying belt speed interval and the relative weight difference interval are fixed respectively, and the critical values of the relative weight difference of the material and the conveying belt speed are determined according to the shift occurrence rate; Finally, the critical value of the relative weight difference of the material corresponding to each conveying belt speed interval and the critical value of the conveying belt speed corresponding to each relative weight difference interval of the material are obtained; The critical values of the relative weight difference of the material of all the conveying belt speed intervals are summarized, the occurrence frequency of each critical value of the relative weight difference of the material is counted, and the mode is taken as the critical value of the relative weight difference of the material; The critical values of the conveying belt speed of all the relative weight difference intervals of the material are summarized, the occurrence frequency of each critical value of the conveying belt speed is counted, and the mode is taken as the critical value of the conveying belt speed.

8. The Internet of Things data acquisition and analysis system for the palletizing production line according to claim 7, characterized in that: The critical values of the relative weight difference of the material and the conveying belt speed are in the form of: Group all samples according to the conveying belt speed interval, extract the relative weight difference of the material and the corresponding shift occurrence rate; For each conveying belt speed interval group, sort the relative weight difference of the material from small to large, calculate the difference value of the shift occurrence rate between two continuous relative weight differences of the material, and obtain the jump amplitude; The relative weight difference of the material that meets the requirements of the jump amplitude is the critical value of the relative weight difference of the material in the conveying belt speed interval; Group all samples according to the relative weight difference interval of the material, extract the conveying belt speed and the corresponding shift occurrence rate; For each relative weight difference interval group of the material, sort the conveying belt speed from small to large, and calculate the difference value of the shift occurrence rate between two continuous conveying belt speeds; If the difference value of the shift occurrence rate meets the requirements, the latter conveying belt speed is taken as the critical value of the conveying belt speed.

9. The Internet of Things data acquisition and analysis system for the palletizing production line according to claim 1, characterized in that: The judgment method of whether there is a resonance type vibration risk is: Real-time acquisition of the current material weight and the weight of the previous material, and calculation of the relative weight difference: Extract the conveying belt speed from the real-time data of the system; Compare the current conveying belt speed and the relative weight difference of the material with the critical values respectively, if the conveying belt speed is greater than or equal to the critical value of the conveying belt speed, and the relative weight difference of the material is greater than or equal to the critical value of the relative weight difference, then there is a resonance type vibration risk.

10. The Internet of Things data acquisition and analysis system for the palletizing production line according to claim 9, characterized in that: The process of correcting the coordinates collected by the current vision sensor is: Calculate the Euclidean distance between the current relative weight difference of the material and the conveying belt speed and the centers of each resonance risk group, and the resonance risk feature group with the smallest Euclidean distance is the resonance risk group to which the current relative weight difference of the material and the conveying belt speed belong; Based on the average deviation of the resonance risk feature group, correct the coordinates (X_vision, Y_vision) collected by the current vision sensor: The corrected coordinates (X_corrected, Y_corrected) are: X_corrected = X_visual + μX, Y_corrected = Y_visual + μY, where μX is the average deviation of the resonance risk feature group in the X direction, and μY is the average deviation of the resonance risk group in the Y direction.

Images