Inventory checking method for intelligent vending cabinet

Abstract

The invention discloses an inventory checking method for an intelligent vending cabinet, and belongs to the field of intelligent vending cabinets, and the method comprises the steps: dividing a monitoring region, building a three-dimensional coordinate system, recognizing the coordinates of commodities, building a self-inspection curve chain during the closing period of the cabinet, calculating the Euclidean distance between the chains, and monitoring the inventory state. A commodity quantity inventory mechanism is triggered when abnormity is found, and an early warning instruction is generated to inform operation and maintenance personnel; meanwhile, a self-checking inventory result and an inter-chain Euclidean distance change trend value are recorded, and an adjustment factor is analyzed to dynamically update an inventory interval; during shopping, starting low-frequency coordinate acquisition and comparing the low-frequency coordinate acquisition with a shopping reference chain by taking the last self-inspection curve chain during a cabinet closing interval as the shopping reference chain, switching to high-frequency acquisition when a dynamic Euclidean distance exceeds a threshold value, generating a shopping chain during cabinet closing, comparing and updating a stock, and calculating a selling cabinet rummaging index and a turnover index; analyzing the adjustment demand value and the display priority value to dynamically adjust the shelf; the operation state of the inventory system can be monitored in real time, the inventory interval is dynamically adjusted, the commodity display is optimized, and the inventory management efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent vending cabinets, and particularly relates to an inventory counting method for intelligent vending cabinets. Background Art

[0002] With the rapid development of smart retail, smart vending machines have been widely used in the market due to their convenience and efficiency. However, the existing inventory counting technology of smart vending machines still has many shortcomings.

[0003] 1. Lack of real-time monitoring and timely adjustment mechanism for inventory system anomalies, making it impossible to promptly discover and handle problems that arise during the inventory process;

[0004] 2. The inventory process usually relies on fixed intervals for self-checking and inventory taking. It is unable to dynamically adjust the inventory interval according to the dynamic changes of shelves, resulting in low inventory efficiency and inability to adapt to dynamic changes in sales;

[0005] 3. There is a lack of scientific basis for product display, and it is impossible to dynamically adjust product display; this may cause popular products to be placed in inconspicuous locations, affecting sales efficiency and customer experience. At the same time, it increases the workload of repeated inspection and verification of these products during inventory, reducing inventory efficiency. To this end, we propose an inventory counting method for smart vending machines. Summary of the Invention

[0006] The purpose of the present invention is to provide an inventory counting method for an intelligent vending cabinet to solve the problems raised in the above background technology.

[0007] To achieve the above-mentioned purpose, the present invention provides the following technical solution: an inventory counting method for a smart vending cabinet, comprising the following steps:

[0008] Step 1: Divide the vending machine into a monitoring area, establish a three-dimensional coordinate system for the monitoring area, and identify the coordinates of the products. During the closed-case period, construct a self-check curve chain. Analyze the initial self-check inventory intervals of the vending machine and calculate the Euclidean distance between the self-check curve chains in adjacent intervals to monitor the inventory status. If any suspected anomaly is found, the product quantity inventory mechanism is triggered. Once the anomaly is confirmed, an early warning instruction is generated and the operation and maintenance personnel are notified to handle it.

[0009] Step 2: Record the results of each self-check inventory and the trend value of the Euclidean distance between chains, analyze the adjustment factor based on this, and dynamically update the inventory interval according to the adjustment factor;

[0010] Step 3: Use the self-inspection curve chain of the last self-inspection inventory during the cabinet closing interval as the shopping reference chain. When shopping behavior is detected, start low-frequency coordinate collection, generate a low-frequency collection curve chain and compare it with the shopping reference chain; if the dynamic Euclidean distance exceeds the threshold, switch to high-frequency collection; record the coordinate changes and movement times of the product points; when closing the cabinet, generate a shopping chain and compare it with the shopping reference chain, deduct the sold products to update the inventory; calculate the sales cabinet search index and turnover index, analyze and adjust the demand value and display priority value, and adjust the shelves accordingly.

[0011] Preferably, in step 1: dividing the vending cabinet into monitoring areas, establishing a three-dimensional coordinate system for the monitoring area and identifying the coordinates of the goods, and constructing a self-checking curve chain during the cabinet closing period are as follows:

[0012] Obtain the three-dimensional point cloud data of the product; divide the smart vending cabinet into areas based on the shelf layer to obtain the monitoring area;

[0013] With the lower left corner of the shelf as the coordinate origin, the horizontal rightward direction as the X axis, the vertical upward direction as the Y axis, and the vertical direction from the shelf plane to the inside of the container as the Z axis, a three-dimensional coordinate system is established to obtain the three-dimensional coordinate system of the monitoring area;

[0014] Identify the type of goods in the monitoring area, and use binocular stereo vision technology to calculate the coordinates of the goods in the three-dimensional coordinate system of the monitoring area to obtain the coordinate points of the goods in the monitoring area, which are recorded as the product points;

[0015] For each monitoring area, adjacent commodity points in the coordinate system are connected with curve segments along the direction of the X-axis in the three-dimensional coordinate system of the monitoring area to obtain a self-inspection curve chain.

[0016] Preferably, in step 1, the specific process of analyzing the initial self-check inventory interval of the vending machine and calculating the Euclidean distance between the self-check curve chains in adjacent intervals to monitor the inventory status is as follows:

[0017] For each monitoring area, perform the first inventory based on the initial self-inspection inventory interval corresponding to the current cabinet closure interval. Obtain the coordinate set of each commodity point on the self-inspection curve chain at the current moment, and simultaneously retrieve the coordinate set of each commodity point during the previous self-inspection inventory during the same cabinet closure interval. Calculate the deviation value of the self-inspection curve chain between two adjacent self-inspection inventories using the Euclidean distance calculation method to obtain the Euclidean distance between the chains.

[0018] Compare the Euclidean distance between chains at the current moment with the preset threshold. If the Euclidean distance between chains at the current moment is less than the threshold, it is determined that the self-inspection inventory is normal.

[0019] On the contrary, if the Euclidean distance between chains corresponding to the current moment is greater than or equal to the corresponding threshold, the inventory is determined to be a suspected abnormal inventory.

[0020] Preferably, in step 1, if a suspected abnormality is found, the commodity quantity inventory mechanism is triggered; after the abnormality is confirmed, an early warning instruction is generated and the operation and maintenance personnel are notified to handle the specific process as follows:

[0021] A coordinate point-product type-inventory quantity mapping table is constructed. When there is a suspected abnormal inventory, the product quantity inventory mechanism is triggered. At this time, the actual inventory quantity of each product in the coordinate point-product type-inventory quantity mapping table is calculated; the theoretical inventory quantity of each product is obtained. If the actual inventory quantity of a product is greater or less than the theoretical inventory quantity of the product, the suspected abnormal inventory is determined to be an abnormal inventory of the vending machine. At this time, an inventory abnormality warning instruction is triggered and sent to the operation and maintenance terminal, which dispatches staff to inspect and repair the vending machine.

[0022] If the actual inventory quantity of all commodities is equal to the theoretical inventory quantity of the corresponding commodities, the suspected abnormal inventory is determined to be the movement of commodity coordinates. At this time, the coordinates of each commodity point are remarked and the self-check curve chain is updated.

[0023] Preferably, in step 2, the specific process of constructing an inventory record table to record the results of each self-inspection inventory and the trend value of the Euclidean distance between chains is as follows:

[0024] Construct an inventory record table and record the results of each self-inspection inventory as binary data, where 0 indicates normal and 1 indicates suspected abnormality;

[0025] Calculate the ratio of the difference between the Euclidean distances of two adjacent inventory chains to the Euclidean distance of the previous chain to obtain the trend value of the Euclidean distance change between chains.

[0026] Preferably, in step 2, the specific process of analyzing the adjustment factor and dynamically updating the inventory interval according to the adjustment factor is as follows:

[0027] For each monitoring area, use the formula: Get the adjustment factor F; where i is the number of each self-check inventory; i = 1, 2, ..., n; n represents the total number of inventories up to the current moment; S i is the result of the i-th self-inspection inventory, W si The preset weight assigned to the inventory result of the i-th self-inspection inventory, T i is the trend value of the Euclidean distance between chains in the i-th self-inspection inventory, W ti The preset weight assigned to the trend value of the Euclidean distance change between chains is stored for the i-th self-check;

[0028] The updated inventory interval f is obtained using the formula: f = fo × (1 + F); where fo is the initial self-inspection inventory interval corresponding to the shelf's closed cabinet interval;

[0029] The update inventory interval is used as the time interval for the next self-inspection inventory corresponding to the current time in the monitoring area.

[0030] Preferably, in step 3, during shopping, the self-check curve chain of the last self-check inventory during the closed cabinet interval is used as the shopping reference chain. When shopping behavior is detected, low-frequency coordinate collection is started, a low-frequency collection curve chain is generated and compared with the shopping reference chain. If the dynamic Euclidean distance exceeds the threshold, the specific process of switching to high-frequency collection is as follows:

[0031] When the smart vending machine door is opened for shopping, the self-inspection curve chain corresponding to the last self-inspection inventory of each monitoring area during the closed cabinet interval is obtained as the shopping reference chain;

[0032] Lay out infrared tube arrays at the cargo space boundaries of each monitoring area; initialize the status of the infrared tube arrays in each monitoring area, and mark them as inactive areas by default;

[0033] When the infrared tube in the monitoring area is blocked, the monitoring area is marked as a suspected active shopping area and the preset low-frequency coordinate collection is started; a low-frequency collection curve chain is generated for each collection time; the dynamic Euclidean distance between the low-frequency collection curve chain and the shopping reference chain at each collection time is calculated; if the dynamic Euclidean distance corresponding to the collection time is greater than or equal to the corresponding threshold, the monitoring area is marked as an active area and the preset high-frequency coordinate system is switched to collect the coordinate changes of each product point;

[0034] If the dynamic Euclidean distance corresponding to the collection time is always less than the preset dynamic Euclidean distance threshold and the occlusion signal disappears, the suspected active shopping area mark of the monitored area is cancelled and restored to an inactive area.

[0035] Preferably, in step 3, when closing the cabinet, a shopping chain is generated and compared with a shopping reference chain, and the sold goods are deducted. The specific process of updating the inventory is as follows:

[0036] For each active shopping area, the coordinates of each product point in the shopping sub-area at the time of collection are recorded in real time; when the smart vending machine is closed, the curved chain formed by each product point is recorded as the shopping chain;

[0037] The shopping chain at the time of cabinet closure is compared with the shopping reference chain of the last self-inspection inventory during the cabinet closure interval through a dynamic contour matching algorithm to identify the commodity points that exist in the reference chain but disappear in the shopping chain. In the coordinate point-commodity type-inventory quantity mapping table, the corresponding inventory quantity of the commodity type corresponding to the disappeared commodity point is deducted according to the actual disappeared quantity; the inventory quantity of each commodity after the deduction is calculated and used as the theoretical inventory quantity of each commodity during this cabinet closure interval.

[0038] Preferably, in step 3, the specific process of calculating the commodity obstruction index and turnover index is:

[0039] Establish a product point movement count database. For each product point, calculate the Euclidean distance between the coordinates of two adjacent collection times, which is recorded as the moving Euclidean distance. If the moving Euclidean distance is greater than the corresponding threshold, it is determined that the product point has moved significantly, and the movement count of the product point in the product point movement count database is increased by one.

[0040] For each product, the total number of times the product point moves during each smart vending machine opening is counted, which is recorded as the product opening movement value. At the same time, the duration of the smart vending machine opening is counted, which is recorded as the opening duration. The product occlusion index YD is obtained by dividing the product opening movement value by the opening duration.

[0041] Taking the most recent smart vending machine shelf adjustment as the starting moment, calculate the average of all commodity movement indices up to the current moment, and record it as the vending machine search index FS;

[0042] Calculate the sales quantity XS, inventory quantity KC, and average sales time XT of the product; use the formula: SZ = XS / (KC×XT) to obtain the product turnover index SZ;

[0043] Taking the most recent time when the smart vending machine adjusted the shelves as the starting moment, calculate the average turnover index of all commodities up to the current moment, and record it as the vending machine turnover index ZS.

[0044] Preferably, in step 3, the specific process of analyzing and adjusting the demand value and the display priority value and adjusting the shelf accordingly is as follows:

[0045] After normalizing the sales cabinet search index FS and the sales cabinet turnover index ZS, the adjusted demand value TQ is obtained using the formula: TQ = FS × a1 + (1-ZS) × a2; where a1 and a2 are preset weight coefficients;

[0046] If the current adjustment demand value is greater than the corresponding threshold, a shelf adjustment signal is generated and the product display priority sequence of each monitoring area is retrieved and sent to the personnel terminal. The personnel terminal dispatches the staff to make shelf adjustments;

[0047] The analysis process of the product display priority sequence in the monitoring area is as follows:

[0048] For each product, the sales frequency SP of the product is obtained according to the corresponding single-unit sales time XT. After normalizing the sales frequency SP of the product, the product obscuration index YD and the product turnover index SZ, the formula is used: Get the product display priority value CY; where b1, b2, and b3 are preset weight coefficients;

[0049] For each monitoring area, the products are sorted according to the product popularity values ​​corresponding to the product types to obtain the product sales popularity sequence.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] (1) This is an inventory counting method for smart vending cabinets. During the cabinet closing period, by constructing a self-checking curve chain and calculating the Euclidean distance between chains, the operating status of the inventory system is monitored in real time to ensure data reliability; the monitoring area is divided according to the shelf layer, and a self-checking curve chain connecting the product points is generated based on the three-dimensional coordinate system. The curve chain coordinate differences between adjacent inventory intervals are regularly compared, and the deviation value between chains is calculated by the Euclidean distance; if the deviation value exceeds the preset threshold, the product quantity inventory mechanism is triggered to verify the actual inventory and the theoretical inventory: if the inventory quantity is consistent, it is determined that the system coordinate collection is abnormal, and the curve chain is automatically marked and updated; if the inventory quantity is inconsistent, it is determined that the hardware is abnormal, and an early warning is immediately sent to the operation and maintenance terminal; this mechanism focuses on system stability monitoring, and distinguishes system errors from real changes through curve chain comparison + inventory verification, providing a reliable benchmark for inventory management after the cabinet is opened.

[0052] (2) The inventory counting method of the intelligent vending machine calculates the dynamic value of the shelf and matches the corresponding initial self-inspection inventory interval; the inventory cycle of the shelf with high dynamic value is automatically shortened, and the cycle of the shelf with low dynamic value is extended, so as to realize on-demand inventory and avoid the waste of resources of fixed interval inventory; the adjustment factor is calculated by formula based on the historical inventory results and the trend of the change of the Euclidean distance between chains, and the subsequent inventory interval is dynamically adjusted; the interval of the shelf with frequent abnormalities is automatically shortened, and the interval of the stable shelf is extended, so as to further improve the inventory efficiency and system adaptability.

[0053] (3) This method of inventory counting for smart vending machines not only achieves the synchronization of inventory data and sales behavior through a closed loop of real-time tracking of shopping behavior → dynamic update of inventory → display optimization, but also reduces the causes of inventory abnormalities from the source through user behavior analysis and display strategy adjustment; for example, obstruction, unsold goods, and backlogs, so that inventory counting can be transformed from post-error correction to pre-emptive prevention + real-time management, thereby comprehensively improving the timeliness, accuracy, and operational efficiency of smart vending machine inventory management. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] Example 1

[0057] See also Figure 1 The present invention provides an inventory counting method for an intelligent vending cabinet, comprising:

[0058] Step 1: Divide the vending machine into a monitoring area, establish a three-dimensional coordinate system for the monitoring area, and identify the coordinates of the products. During the closed period, construct a self-check curve chain. Analyze the initial self-check inventory intervals of the vending machine, calculate the Euclidean distance between the self-check curve chains in adjacent intervals to monitor the inventory status, and trigger the product inventory mechanism if any suspected anomalies are found. After confirming the anomaly, generate an early warning instruction and notify the operation and maintenance personnel to handle it. The specific process is as follows:

[0059] An infrared sensor matrix is ​​installed at the bottom of each shelf of the smart vending machine to detect the presence of goods, and a wide-angle camera is installed on the top to capture product images. Depth cameras are installed on both sides of the shelf to obtain three-dimensional point cloud data of the goods. The smart vending machine is divided into areas based on the shelf layer to obtain monitoring areas.

[0060] With the lower left corner of the shelf as the coordinate origin, the horizontal rightward direction as the X axis, the vertical upward direction as the Y axis, and the vertical direction from the shelf plane to the inside of the container as the Z axis, a three-dimensional coordinate system is established to obtain the three-dimensional coordinate system of the monitoring area;

[0061] A wide-angle camera and a depth camera are combined with a target detection algorithm to identify the type of goods in the monitoring area. At the same time, binocular stereo vision technology is used to calculate the coordinates of the goods in the three-dimensional coordinate system of the monitoring area. The coordinate points of the goods in the monitoring area are obtained and recorded as the product points. The product points include the type of goods and the coordinates of the goods in the monitoring area.

[0062] For each monitoring area, along the X-axis direction of the three-dimensional coordinate system of the monitoring area, the adjacent commodity points in the coordinate system are connected with curve segments in turn to obtain a self-inspection curve chain;

[0063] The period between the closing and the next opening of the smart vending machine door is defined as the closed interval. At the beginning of each closed interval, transaction and inventory data of each shelf layer are obtained. Transaction and inventory data include: sales quantity, sales amount, total amount of shelf goods, and remaining quantity of shelf goods.

[0064] Assign different weight coefficients to each data in transaction and inventory data, normalize them, and use weighted analysis method to perform comprehensive calculations to obtain shelf dynamic values;

[0065] Set multiple shelf dynamic value intervals, each shelf dynamic value interval corresponds to an initial self-check inventory interval during the cabinet closing interval; match the shelf dynamic value of each shelf with all shelf dynamic value intervals, and output the initial self-check inventory interval of each shelf during the cabinet closing interval;

[0066] For each monitoring area, perform the first inventory based on the initial self-inspection inventory interval corresponding to the current cabinet closure interval. Obtain the coordinate set of each commodity point on the self-inspection curve chain at the current moment, and simultaneously retrieve the coordinate set of each commodity point during the previous self-inspection inventory during the same cabinet closure interval. Calculate the deviation value of the self-inspection curve chain between two adjacent self-inspection inventories using the Euclidean distance calculation method to obtain the Euclidean distance between the chains.

[0067] A threshold value of the inter-chain Euclidean distance is preset. The current inter-chain Euclidean distance is compared with the preset threshold value. If the current inter-chain Euclidean distance is less than the threshold value, the self-check inventory is considered normal.

[0068] It should be noted that if the Euclidean distance between chains is less than the corresponding threshold, it means that the quantity of each product is consistent with the previous inventory, and the coordinates have not changed, indicating that there is no increase, decrease, or displacement of the product. This further indicates that there is no data deviation in the inventory process of the smart vending machine camera recognition, coordinate calculation, curve chain generation, etc., and the smart vending machine inventory system is in a stable and reliable normal working state.

[0069] On the contrary, if the Euclidean distance between chains at the current moment is greater than or equal to the corresponding threshold, the inventory is judged as a suspected abnormal inventory;

[0070] A coordinate point-product type-inventory quantity mapping table is constructed. When there is a suspected abnormal inventory, the product quantity inventory mechanism is triggered. At this time, the actual inventory quantity of each product in the coordinate point-product type-inventory quantity mapping table is calculated; the theoretical inventory quantity of each product is obtained. If the actual inventory quantity of a product is greater or less than the theoretical inventory quantity of the product, the suspected abnormal inventory is determined to be an abnormal inventory of the vending machine. At this time, an inventory abnormality warning instruction is triggered and sent to the operation and maintenance terminal, which dispatches staff to inspect and repair the vending machine.

[0071] If the actual inventory quantity of all products is equal to the theoretical inventory quantity of the corresponding products, the suspected abnormal inventory is determined to be a product coordinate movement. At this time, the coordinates of each product point are remarked and the self-check curve chain is updated;

[0072] Step 2: Construct an inventory record table to record the results of each self-inspection inventory and the trend value of the Euclidean distance between chains. Based on this, analyze the adjustment factor and dynamically update the inventory interval according to the adjustment factor to optimize the subsequent inventory schedule. The specific process is as follows:

[0073] Construct an inventory record table and record the results of each self-inspection inventory as binary data, where 0 indicates normal and 1 indicates suspected abnormality; at the same time, record the trend value of the Euclidean distance between chains;

[0074] The calculation process of the Euclidean distance change trend value is as follows: the ratio of the difference between the Euclidean distances of two consecutive inventory chains to the Euclidean distance of the previous chain is used to measure it. A positive ratio indicates that the Euclidean distance between chains is increasing, and a negative ratio indicates that it is decreasing.

[0075] For each monitoring area, use the formula: Get the adjustment factor F; where i is the number of each self-check inventory; i = 1, 2, ..., n; n represents the total number of inventories up to the current moment; S i is the result of the i-th self-inspection inventory, W si The preset weight assigned to the inventory result of the i-th self-inspection inventory, T i is the trend value of the Euclidean distance between chains in the i-th self-inspection inventory, W ti The preset weight assigned to the trend value of the Euclidean distance change between chains is stored for the i-th self-check;

[0076] Use the formula: f = fo × (1 + F) to obtain the update inventory interval f; where fo is the initial self-inspection inventory interval corresponding to the shelf's closed cabinet interval; the update inventory interval is the time interval for the next self-inspection inventory corresponding to the current time in the monitored area;

[0077] Step 3: During shopping, the self-inspection curve chain of the last self-inspection inventory during the cabinet closing interval is used as the shopping reference chain. When shopping behavior is detected, low-frequency coordinate acquisition is started, a low-frequency acquisition curve chain is generated, and compared with the shopping reference chain; if the dynamic Euclidean distance exceeds the threshold, high-frequency acquisition is switched to; the coordinate changes and movement times of the product points are recorded; when the cabinet is closed, a shopping chain is generated and compared with the shopping reference chain, the sold products are deducted, and the inventory is updated; the sales cabinet search index and turnover index are calculated, the demand value and product display priority value are analyzed and adjusted, and the shelf display is dynamically adjusted. The specific process is as follows:

[0078] When a customer is shopping with the smart vending machine door open, the self-inspection curve chain corresponding to the last self-inspection inventory of each monitoring area during the closed cabinet interval is obtained as a shopping reference chain;

[0079] Lay out infrared tube arrays at the cargo space boundaries of each monitoring area; initialize the status of the infrared tube arrays in each monitoring area, and mark them as inactive areas by default;

[0080] When the infrared pair of tubes in the monitoring area is blocked, the monitoring area is marked as a suspected active shopping area and preset low-frequency coordinate collection is started; a low-frequency collection curve chain is generated for each collection moment; the dynamic Euclidean distance between the low-frequency collection curve chain and the shopping reference chain at each collection moment is calculated; a dynamic Euclidean distance threshold is preset, and if the dynamic Euclidean distance corresponding to the collection moment is greater than or equal to the preset dynamic Euclidean distance threshold, the monitoring area is marked as an active area and the preset high-frequency coordinate system is switched to collect the coordinate changes of each product point;

[0081] If the dynamic Euclidean distance corresponding to the collection time is always less than the preset dynamic Euclidean distance threshold and the occlusion signal disappears, the suspected active shopping area mark of the monitored area is cancelled and restored to an inactive area;

[0082] For each active shopping area, the coordinates of each product point in the shopping sub-area at the time of collection are recorded in real time; when the smart vending machine is closed, the curved chain formed by each product point is recorded as the shopping chain;

[0083] The shopping chain at the time of cabinet closure is compared with the shopping reference chain from the last self-inspection inventory during the cabinet closure interval using a dynamic contour matching algorithm to identify product points that existed in the reference chain but disappeared from the shopping chain. In the coordinate point-product type-inventory quantity mapping table, the corresponding inventory quantity of the product type corresponding to the disappeared product point is deducted according to the actual disappeared quantity. The inventory quantity of each product after the deduction is calculated and used as the theoretical inventory quantity of each product at the time of the cabinet closure interval.

[0084] Establish a product point movement count database. For each product point, calculate the Euclidean distance between the coordinates of two adjacent collection times, which is recorded as the moving Euclidean distance. Preset a moving Euclidean distance threshold. If the moving Euclidean distance is greater than the corresponding threshold, it is determined that the product point has moved significantly, and the movement count of the product point in the product point movement count database is increased by one.

[0085] For each product, the total number of times the product point moves during each smart vending machine opening is counted, which is recorded as the product opening movement value. At the same time, the duration of the smart vending machine opening is counted, which is recorded as the opening duration. The product occlusion index YD is obtained by dividing the product opening movement value by the opening duration.

[0086] Taking the most recent smart vending machine shelf adjustment as the starting moment, calculate the average of all commodity movement indices up to the current moment, and record it as the vending machine search index FS;

[0087] Calculate the sales quantity XS, inventory quantity KC, and average sales time XT of each item. The average sales time is the average sales time of each item from the last time the smart vending machine adjusted the shelves to the current time. Use the formula: SZ = XS / (KC × XT) to obtain the product turnover index SZ.

[0088] Taking the most recent smart vending machine shelf adjustment as the starting moment, calculate the average turnover index of all commodities up to the current moment, and record it as the vending machine turnover index ZS;

[0089] After normalizing the sales cabinet search index FS and the sales cabinet turnover index ZS, the adjusted demand value TQ is obtained using the formula: TQ = FS × a1 + (1-ZS) × a2; where a1 and a2 are preset weight coefficients;

[0090] A preset adjustment demand threshold is set. If the current adjustment demand value is greater than the corresponding threshold, a shelf adjustment signal is generated and the product display priority sequence of each monitoring area is retrieved and sent to the personnel terminal. The personnel terminal dispatches the staff to make shelf adjustments.

[0091] The analysis process of the product display priority sequence in the monitoring area is as follows:

[0092] For each product, the sales frequency SP of the product is obtained according to the corresponding single-unit sales time XT. After normalizing the sales frequency SP of the product, the product obscuration index YD and the product turnover index SZ, the formula is used: Get the product display priority value CY; where b1, b2, and b3 are preset weight coefficients;

[0093] For each monitoring area, the products are sorted according to the product popularity values ​​corresponding to the product types to obtain the product sales popularity sequence.

[0094] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An inventory counting method for an intelligent vending machine, characterized by: Includes the following steps: Step 1: Divide the vending machine into a monitoring area, establish a three-dimensional coordinate system for the monitoring area, and identify the coordinates of the products. During the closed-case period, construct a self-check curve chain. Analyze the initial self-check inventory intervals of the vending machine and calculate the Euclidean distance between the self-check curve chains in adjacent intervals to monitor the inventory status. If any suspected anomaly is found, the product quantity inventory mechanism is triggered. Once the anomaly is confirmed, an early warning instruction is generated and the operation and maintenance personnel are notified to handle it. Step 2: Record the results of each self-check inventory and the trend value of the Euclidean distance between chains, analyze the adjustment factor based on this, and dynamically update the inventory interval according to the adjustment factor; Step 3: Use the self-inspection curve chain of the last self-inspection inventory during the cabinet closing interval as the shopping reference chain. When shopping behavior is detected, start low-frequency coordinate collection, generate a low-frequency collection curve chain and compare it with the shopping reference chain; if the dynamic Euclidean distance exceeds the threshold, switch to high-frequency collection; record the coordinate changes and movement times of the product points; when closing the cabinet, generate a shopping chain and compare it with the shopping reference chain, deduct the sold products to update the inventory; calculate the sales cabinet search index and turnover index, analyze and adjust the demand value and display priority value, and adjust the shelves accordingly.

2. The method for counting inventory of a smart vending machine according to claim 1, wherein: In step 1, the monitoring area of ​​the vending machine is divided, a three-dimensional coordinate system of the monitoring area is established, and the coordinates of the products are identified. During the closed cabinet period, the specific process of constructing the self-inspection curve chain is as follows: Obtain the three-dimensional point cloud data of the product; divide the smart vending machine into areas based on the shelf layer to obtain the monitoring area; With the lower left corner of the shelf as the coordinate origin, the horizontal rightward direction as the X axis, the vertical upward direction as the Y axis, and the vertical direction from the shelf plane to the inside of the container as the Z axis, a three-dimensional coordinate system is established to obtain the three-dimensional coordinate system of the monitoring area; Identify the type of goods in the monitoring area, and use binocular stereo vision technology to calculate the coordinates of the goods in the three-dimensional coordinate system of the monitoring area to obtain the coordinate points of the goods in the monitoring area, which are recorded as the product points; For each monitoring area, adjacent commodity points in the coordinate system are connected with curve segments along the direction of the X-axis in the three-dimensional coordinate system of the monitoring area to obtain a self-inspection curve chain.

3. The method for counting inventory of a smart vending machine according to claim 2, wherein: In step 1, the specific process of analyzing the initial self-check inventory interval of the vending machine and calculating the Euclidean distance between the self-check curve chains in adjacent intervals to monitor the inventory status is as follows: For each monitoring area, perform the first inventory based on the initial self-inspection inventory interval corresponding to the current cabinet closure interval. Obtain the coordinate set of each commodity point on the self-inspection curve chain at the current moment, and simultaneously retrieve the coordinate set of each commodity point during the previous self-inspection inventory during the same cabinet closure interval. Calculate the deviation value of the self-inspection curve chain between two adjacent self-inspection inventories using the Euclidean distance calculation method to obtain the Euclidean distance between the chains. Compare the Euclidean distance between chains at the current moment with the preset threshold. If the Euclidean distance between chains at the current moment is less than the threshold, it is determined that the self-inspection inventory is normal. On the contrary, if the Euclidean distance between chains corresponding to the current moment is greater than or equal to the corresponding threshold, the inventory is determined to be a suspected abnormal inventory.

4. The method for counting inventory of a smart vending machine according to claim 3, wherein: In step 1, if a suspected anomaly is found, the product inventory mechanism is triggered. After the anomaly is confirmed, an early warning instruction is generated and the operation and maintenance personnel are notified to handle it. The specific process is as follows: Construct a coordinate point-product type-inventory quantity mapping table. When there is a suspected abnormal inventory, the product quantity inventory mechanism is triggered. At this time, the actual inventory quantity of each product in the coordinate point-product type-inventory quantity mapping table is calculated. Obtain the theoretical inventory quantity of each product. If the actual inventory quantity of a product is greater or less than the theoretical inventory quantity, the suspected abnormal inventory is determined to be an inventory abnormality of the vending machine. At this time, an inventory abnormality warning instruction is triggered and sent to the operation and maintenance terminal. The operation and maintenance terminal dispatches staff to inspect and repair the vending machine; If the actual inventory quantity of all commodities is equal to the theoretical inventory quantity of the corresponding commodities, the suspected abnormal inventory is determined to be the movement of commodity coordinates. At this time, the coordinates of each commodity point are remarked and the self-check curve chain is updated.

5. The inventory counting method for a smart vending cabinet according to claim 4, characterized in that: In step 2, the specific process of constructing an inventory record table to record the results of each self-inspection inventory and the trend value of the Euclidean distance between chains is as follows: Construct an inventory record table and record the results of each self-inspection inventory as binary data, where 0 indicates normal and 1 indicates suspected abnormality; Calculate the ratio of the difference between the Euclidean distances of two adjacent inventory chains to the Euclidean distance of the previous chain to obtain the trend value of the Euclidean distance change between chains.

6. The method for counting inventory of a smart vending machine according to claim 5, characterized in that: In step 2, the specific process of analyzing the adjustment factor and dynamically updating the inventory interval based on the adjustment factor is as follows: For each monitoring area, use the formula: Get the adjustment factor F; where i is the number of each self-check inventory; i = 1, 2, ..., n; n represents the total number of inventories up to the current moment; S i is the result of the i-th self-inspection inventory, W si The preset weight assigned to the inventory result of the i-th self-inspection inventory, T i is the trend value of the Euclidean distance between chains in the i-th self-inspection inventory, W ti The preset weight assigned to the trend value of the Euclidean distance change between chains is stored for the i-th self-check; The updated inventory interval f is obtained using the formula: f = fo × (1 + F); where fo is the initial self-inspection inventory interval corresponding to the shelf's closed cabinet interval; The update inventory interval is used as the time interval for the next self-inspection inventory corresponding to the current time in the monitoring area.

7. The method for counting inventory of a smart vending machine according to claim 6, characterized in that: In step 3, during shopping, the self-check curve chain of the last self-check inventory during the closed counter interval is used as the shopping reference chain. When shopping behavior is detected, low-frequency coordinate collection is started, a low-frequency collection curve chain is generated and compared with the shopping reference chain. If the dynamic Euclidean distance exceeds the threshold, the specific process of switching to high-frequency collection is as follows: When the smart vending machine door is opened for shopping, the self-inspection curve chain corresponding to the last self-inspection inventory of each monitoring area during the closed cabinet interval is obtained as the shopping reference chain; An infrared tube array is laid at the cargo space boundaries of each monitoring area; Initialize the infrared tube array status of each monitoring area, and mark it as an inactive area by default; When the infrared tube in the monitoring area is blocked, the monitoring area is marked as a suspected active shopping area and the preset low-frequency coordinate collection is started; a low-frequency collection curve chain is generated for each collection time; the dynamic Euclidean distance between the low-frequency collection curve chain and the shopping reference chain at each collection time is calculated; if the dynamic Euclidean distance corresponding to the collection time is greater than or equal to the corresponding threshold, the monitoring area is marked as an active area and the preset high-frequency coordinate system is switched to collect the coordinate changes of each product point; If the dynamic Euclidean distance corresponding to the collection time is always less than the preset dynamic Euclidean distance threshold and the occlusion signal disappears, the suspected active shopping area mark of the monitored area is cancelled and restored to an inactive area.

8. The method for counting inventory of a smart vending machine according to claim 7, characterized in that: In step 3, when closing the cabinet, a shopping chain is generated and compared with the shopping reference chain, and the sold goods are deducted. The specific process of updating the inventory is as follows: For each active shopping area, the coordinates of each product point in the shopping sub-area at the time of collection are recorded in real time; when the smart vending machine is closed, the curved chain formed by each product point is recorded as the shopping chain; Compare the shopping chain at the time of cabinet closure with the shopping reference chain from the last self-inspection inventory during the cabinet closure interval using a dynamic contour matching algorithm to identify product points that existed in the reference chain but disappeared from the shopping chain. In the coordinate point-product type-inventory quantity mapping table, deduct the corresponding inventory quantity for the product type corresponding to the disappeared product point based on the actual disappeared quantity. The inventory quantity of each commodity after deduction is calculated and used as the theoretical inventory quantity of each commodity during this cabinet closing interval.

9. The method for counting inventory of a smart vending machine according to claim 8, characterized in that: In step 3, the specific process of calculating the product obstruction index and turnover index is as follows: Establish a product point movement count database. For each product point, calculate the Euclidean distance between the coordinates of two adjacent collection times, which is recorded as the moving Euclidean distance. If the moving Euclidean distance is greater than the corresponding threshold, it is determined that the product point has moved significantly, and the movement count of the product point in the product point movement count database is increased by one. For each product, the total number of times the product point moves during each smart vending machine opening is counted, which is recorded as the product opening movement value. At the same time, the duration of the smart vending machine opening is counted, which is recorded as the opening duration. The product occlusion index YD is obtained by dividing the product opening movement value by the opening duration. Taking the most recent smart vending machine shelf adjustment as the starting moment, calculate the average of all commodity movement indices up to the current moment, and record it as the vending machine search index FS; Calculate the sales quantity XS, inventory quantity KC, and average sales time XT of the product; use the formula: SZ = XS / (KC×XT) to obtain the product turnover index SZ; Taking the most recent time when the smart vending machine adjusted the shelves as the starting moment, calculate the average turnover index of all commodities up to the current moment, and record it as the vending machine turnover index ZS.

10. The method for counting inventory of a smart vending machine according to claim 9, wherein: In step three, the specific process of analyzing and adjusting the demand value and display priority value and adjusting the shelves accordingly is as follows: After normalizing the sales cabinet search index FS and the sales cabinet turnover index ZS, the adjusted demand value TQ is obtained using the formula: TQ = FS × a1 + (1-ZS) × a2; where a1 and a2 are preset weight coefficients; If the current adjustment demand value is greater than the corresponding threshold, a shelf adjustment signal is generated and the product display priority sequence of each monitoring area is retrieved and sent to the personnel terminal. The personnel terminal dispatches the staff to make shelf adjustments; The analysis process of the product display priority sequence in the monitoring area is as follows: For each product, the sales frequency SP of the product is obtained according to the corresponding single-unit sales time XT. After normalizing the sales frequency SP of the product, the product obscuration index YD and the product turnover index SZ, the formula is used: Get the product display priority value CY; where b1, b2, and b3 are preset weight coefficients; For each monitoring area, the products are sorted according to the product popularity values ​​corresponding to the product types to obtain the product sales popularity sequence.

Images