A method for allocating cargo space in a pharmaceutical warehouse
Through the multi-target cargo space optimization model and simulated annealing genetic algorithm, the problem of cargo space optimization in pharmaceutical warehouses is solved, the picking efficiency and cargo space balance are improved, and the picking congestion caused by small quantity of medical orders but many orders are solved.
Patent Information
- Application Number
- CN202210828985.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-07-15
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2042-07-15
AI Technical Summary
The existing technology is difficult to effectively optimize the location of the pharmaceutical warehouse, resulting in congestion in picking when the quantity of pharmaceutical orders is small but the orders are too high, affecting efficiency and user experience.
The multi-target cargo space optimization model is adopted, and the property information of the goods to be in stock and the warehouse partition layout information are obtained, and the cargo space is allocated in batches, which optimizes the cargo turnover rate, cargo correlation and shelf stability, and combines the simulated annealing genetic algorithm to solve it. Achieve shelf load balancing and improve picking efficiency.
It effectively reduces the cost of cargo handling, improves picking efficiency, solves the problem of picking congestion, and ensures the rationality of cargo space distribution.
Smart Images

Figure CN115115256B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of pharmaceutical storage, and in particular relates to a method for allocating cargo spaces in a pharmaceutical warehouse. Background Art
[0002] With the development of modern mechanical technology and information technology, the pharmaceutical distribution industry increasingly relies on modern logistics equipment and logistics information management systems, and optimizes the acceptance, storage, picking, distribution and other operation processes in the warehousing and distribution links by effectively integrating upstream and downstream resources of marketing channels, aiming to shorten inventory and turnover time, reduce logistics costs, and achieve automation, informatization and efficiency. Among them, the optimization of the storage space for pharmaceutical goods has attracted more and more attention from corporate managers.
[0003] The optimization management of cargo space pursues the characteristics of different equipment and shelf types, product grouping, cargo space planning, built-in labor costs and other factors to achieve the best cargo space layout, which can effectively grasp the changes in goods and maximize cost savings. The optimization management of cargo space is to explore the efficiency and cost of warehouses in operation, and to prepare for the key management before the operation of a distribution center or warehouse under construction. However, about 80% of distribution centers or warehouses are currently unable to perform correct cargo space optimization, especially for third-party logistics companies of pharmaceutical companies. The existing research on cargo space optimization is mainly conducted on ordinary warehouses, which does not take into account the particularity of pharmaceutical warehouses. For example, considering the cargo turnover rate alone for the situation where the pharmaceutical order volume is small but the order volume is large, it is easy to cause congestion during picking. At the same time, different from other goods, for pharmaceutical warehouses, different batches of the same goods need different cargo spaces to store goods to avoid confusion, which increases the difficulty of cargo space allocation. Therefore, the traditional warehouse cargo space optimization method cannot fully solve the problem of cargo space optimization in the pharmaceutical warehouse scenario, which may lead to unsatisfactory optimization results, affect the order picking speed, and thus affect the user experience. Summary of the invention
[0004] In order to solve the deficiencies in the prior art, the present invention provides a method for allocating cargo spaces in a pharmaceutical warehouse, which solves the problem of picking congestion that is easily caused when the pharmaceutical order volume is small but there are many orders.
[0005] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a method for allocating cargo space in a pharmaceutical warehouse, comprising: obtaining attribute information of goods to be stored and partition layout information of the pharmaceutical warehouse; allocating the goods to be stored to the corresponding partitions of the pharmaceutical warehouse according to the storage characteristics in the attribute information of the goods to be stored; for the goods in each partition allocated to the pharmaceutical warehouse, obtaining the storage location allocation results of the first batch of goods by solving the first batch multi-objective cargo space optimization model; taking the storage location allocation results of the first batch of goods as constraints, obtaining the cargo space allocation results of the second batch of goods by solving the second batch multi-objective cargo space optimization model; wherein, the first batch multi-objective cargo space optimization model is constructed and obtained with cargo turnover rate, cargo correlation and shelf stability as optimization objectives; the second batch multi-objective cargo space optimization model is constructed and obtained with shelf stability, shelf load balance and aisle load balance as optimization objectives.
[0006] Furthermore, the attribute information of the goods to be stored includes not only storage characteristics, but also: variety, specification, origin, specification packaging, and batch of the goods to be stored; wherein, the storage characteristics include storage at room temperature, refrigeration, constant temperature and constant humidity.
[0007] Furthermore, the method for constructing the first batch of multi-objective cargo location optimization models includes: taking cargo turnover rate, cargo correlation and shelf stability as optimization objectives respectively, and constructing corresponding objective functions;
[0008] Assign weights β1, β2, β3 to the cargo turnover rate objective function, cargo correlation objective function and shelf stability objective function, β1+β2+β3=1 respectively;
[0009] The first batch multi-objective cargo location optimization model is constructed according to the cargo turnover rate objective function, cargo correlation objective function, shelf stability objective function and the weights of each objective function; wherein the cargo turnover rate objective optimization function is:
[0010]
[0011] Among them, F1 is the target optimization function constructed based on the cargo turnover rate, z is the position coordinate of the zth row of shelves in the warehouse where the goods are located, x and y are the position coordinates of the length and height of the shelves; a is the total number of rows of shelves in the warehouse where the goods are located, c is the total number of mezzanines in each row of shelves in the warehouse where the goods are located, and b is the total number of grids on each shelf layer in the warehouse where the goods are located; t x,y,z v is the time it takes for the picking machine to run from the cargo location (z, x, y) to the shelf entrance, x 、v y are the horizontal and vertical running speeds of the picking machine in an aisle, respectively, and v z is the speed of the picking machine running between lanes; qz,x,y is the turnover rate of the goods at the cargo location (z, x, y). When there is no goods at the cargo location (z, x, y), q z,x,y =0; L, H, D are the length and height of the unit cargo space and the distance between two adjacent lanes;
[0012] The cargo correlation objective optimization function is:
[0013]
[0014] Among them, F2 is the overall cargo relevance objective optimization function, Φ represents the preset community value, and F′ i Optimize the function for the relevance objective of goods in the i-th community subgraph;
[0015] The shelf stability objective optimization function is:
[0016]
[0017] Among them, F3 is the shelf stability objective optimization function, W z,x,y is the unit weight of the goods placed at the cargo location (z, x, y), n z,x,y is the quantity of goods placed at the cargo location (z, x, y).
[0018] Furthermore, the method for constructing the cargo correlation objective optimization function includes:
[0019] The community discovery algorithm is used to cluster medical goods, achieving strong correlation of goods attributes within the same community and weak correlation of goods between different communities;
[0020] According to the mobile robot fulfillment system, the historical order data of medicines in the system maps the correlation between each product into an undirected graph; the product sample m i As a node v in an undirected graph i , we can get the node set V of the undirected graph = {v1, v2, ..., v N}, where the subscript N indicates that there are N types of drugs; establish the feature vector of medical goods, including drug efficacy, expiration date and batch number, and each node is composed of attribute features {u i1 ,…,u iT} indicates that the attribute characteristics are obtained by identifying the information on the drug packaging through optical character recognition technology when the drug goods are first put into the warehouse; at the same time, the components of each dimension are standardized so that each dimension satisfies the standard normal distribution; the edge set of the undirected graph is defined as E, the weight matrix is W, and the element w in the matrix ij Represents node v i With v j The size of the goods correlation is calculated as follows:
[0021]
[0022] Among them, i Indicates cargo m i Does it exist in order o? If it exists in order o, then o i =1, otherwise 0; O represents the historical order set; For cargo m i and cargo j The proportion of orders that appear in one order at the same time to the total orders, Indicates cargo m i and cargo j The similarity difference of the attributes; α and β are parameter correction factors, making the weight w ij Keep it within the range greater than the threshold τ. If the calculated edge weight is less than the threshold τ, set the edge weight to 0, that is, remove the edge that connects the two nodes in the undirected graph; if there is an independent node in the undirected graph, delete it from the undirected graph;
[0023] Considering that different drugs have different drug properties and are prone to react with other special types of drugs, affecting the drug properties, the weight of the edge between the nodes corresponding to the two pharmaceutical goods that react with each other is set to 0 when the model is established, that is, this edge is removed in the undirected graph;
[0024] Use the Louvain algorithm to divide the undirected graph into communities until the number of communities formed reaches the preset community subgraph set value;
[0025] Based on the community subgraph set S generated above, S = {s1, s2, ..., s Φ}, where each subset consists of the subscripts of the goods sets with greater correlation, and the subscript Φ represents the preset community value. The node with the largest degree in the community subgraph is taken as the central goods, and other goods in its set are placed near the central goods in the set. Then the optimization function of the goods correlation objective in a single community subgraph set is:
[0026]
[0027] Among them, F′ i is the i-th community subgraph s i The correlation objective optimization function of the goods in j is the i-th community subgraph s i The shelf ranking of other goods that are not the central goods, x j is the i-th community subgraph s i The grid position of the mezzanine where other goods that are not the central goods are located, y j is the i-th community subgraph s iThe coordinates of the interlayer where the other goods that are not the central goods are located (a i , b i , c i ) is the community subset s i The center cargo coordinates of;
[0028] The cargo relevance objective optimization function is obtained according to the cargo relevance objective optimization function in a single community subgraph set.
[0029] Furthermore, according to the storage location allocation results of the first batch of goods, the load balance of each row of shelves is obtained:
[0030]
[0031] Among them, G z is the load balancing value of each row of shelves after the first batch of goods is allocated. It is only affected by the allocation of the first batch of goods. z,x,y is the status mark of the cargo location (z, x, y), S z,x,y =1 means that the cargo has been placed at the cargo location (z, x, y), S z,x,y = 0 means that the cargo location (z, x, y) is idle, R z is the type of goods that have been allocated to the zth row of shelves in the first batch. If S z,x,y =0, then q z,x,y =0, if S z,x,y =1, then q z,x,y is the turnover rate of the goods stored at the cargo location (z, x, y);
[0032] The load balance of each row of shelves obtained by the storage location allocation of the second batch of goods is expressed as:
[0033]
[0034] Among them, G z ′ is the load balancing value of each row of shelves after the second batch of goods is allocated, which is only affected by the allocation of the first batch of goods. ! is the inversion operation. If S z,x,y =1, then! S z,x,y =0, if S z,x,y =0, then! S z,x,y =1; R′ z The types of goods in the second batch to be allocated to the zth row of shelves.
[0035] Furthermore, an aisle contains two rows of shelves on the left and right sides, and the load balance of the aisle is expressed as:
[0036]
[0037] Among them, GX pThe load balance value of the pth aisle after the first batch of goods is allocated, which includes the load mean of the left shelf G 2·p-1 , and the load balancing value G of the right shelf 2·p ; GX′ p is the load balance value of the pth aisle after the second batch of goods is allocated, which includes the load mean of the left shelf G′ 2·p-1 , and the load balance value G′ of the right shelf 2·p ;
[0038] The optimization objective function of lane load balance is:
[0039]
[0040] Among them, F4 is the optimization objective function of lane load balance, is the maximum value of the load balance degree of each lane.
[0041] Furthermore, the first batch multi-objective cargo space optimization model is constructed as follows:
[0042]
[0043] Among them, F is the cargo location optimization objective function of the first batch of cargo constructed based on cargo-related attributes;
[0044] Taking the storage location allocation result of the first batch of goods as a constraint, the storage location allocation result of the second batch of goods is obtained by solving the second batch multi-objective storage location optimization model, including: assigning weights γ1 and γ2 to the shelf stability objective function and the aisle load balancing objective function respectively, converting the multi-objective optimization problem into a single-objective optimization problem, and then the second batch multi-objective storage location optimization model is constructed as follows:
[0045]
[0046] Among them, F′ is the cargo location optimization objective function of the second batch of cargo constructed based on cargo-related attributes, γ1+γ2=1.
[0047] Furthermore, F and F′ are used as objective functions respectively, and the model is solved by twice simulated annealing genetic algorithm to obtain the optimal solution of cargo location.
[0048] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention warehouses the goods to be stored in batches, allocates appropriate cargo locations for the goods to be stored according to the cargo turnover rate of the goods to be stored, the storage characteristics of the goods to be stored and the weight of the goods to be stored, and allocates cargo locations for the second batch of goods on the basis of the allocation of storage locations for the first batch of goods in order to achieve the load balancing goal of the shelves in the warehouse, thereby reducing the cost of cargo handling, improving the picking efficiency, ensuring that the distribution of cargo locations is in a relatively reasonable state, and solving the problem of picking congestion that is easily caused when the pharmaceutical order volume is small but there are many orders. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Figure 1 It is a flow chart of a method for allocating cargo space in a pharmaceutical warehouse provided by an embodiment of the present invention;
[0050] Figure 2 It is an undirected graph formed by mapping the correlation between each product according to the historical order data of the medicine in the RMFS system in the embodiment of the present invention;
[0051] Figure 3 Schematic diagram of the simulated annealing genetic algorithm used in the embodiment of the present invention. DETAILED DESCRIPTION
[0052] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.
[0053] like Figure 1 As shown, a method for allocating cargo space in a pharmaceutical warehouse comprises: obtaining attribute information of goods to be stored and partition layout information of the pharmaceutical warehouse; allocating the goods to be stored to the corresponding partitions of the pharmaceutical warehouse according to the storage characteristics in the attribute information of the goods to be stored; for the goods in each partition allocated to the pharmaceutical warehouse, obtaining the storage location allocation results of the first batch of goods by solving the first batch multi-objective cargo space optimization model; taking the storage location allocation results of the first batch of goods as constraints, obtaining the cargo space allocation results of the second batch of goods by solving the second batch multi-objective cargo space optimization model; wherein, the first batch multi-objective cargo space optimization model is constructed and obtained with cargo turnover rate, cargo correlation and shelf stability as optimization objectives; the second batch multi-objective cargo space optimization model is constructed and obtained with shelf stability, shelf load balance and aisle load balance as optimization objectives.
[0054] (1): Obtain the attribute information of the goods to be stored and the partition layout information of the pharmaceutical warehouse.
[0055] (1-1): Scan the information code of the goods to be stored and obtain the associated information (attribute information), including the name, specification, origin, quantity, width, unit weight, production date, turnover rate and other coded information of the goods to be stored, the quantity information of the goods to be stored, and the batch information of the goods to be stored. For example, the following goods to be stored are scanned:
[0056] [1, Ibuprofen, Boxed, Guangdong, 4 boxes, Width 55cm, 10kg, January 2022, 0.8]
[0057] [2, Ibuprofen, Boxed, Guangdong, 3 boxes, Width 55cm, 10kg, December 2021, 0.8]
[0058] [3, montmorillonite powder, boxed, Shandong, 2 boxes, width 40cm, 8kg, February 2022, 0.6]
[0059] [4, Jiangzhong Jianwei Xiaoshi Tablets, boxed, Jiangxi, 5 boxes, width 45cm, 6kg, January 2022, 0.4]
[0060] [5, Children's cough syrup, boxed, Guangdong, 2 boxes, width 50cm, 15kg, March 2022, 0.5]
[0061] [6, Liuwei Mingmu Pills, boxed, Qinghai, 3 boxes, width 55cm, 8kg, December 2021, 0.6]
[0062] [7, Taurine granules, boxed, Shandong, 6 boxes, width 40cm, 6kg, November 2021, 0.6]
[0063] [8, Siwei Piwei Shu Granules, boxed, Guangxi, 2 boxes, width 55cm, 14kg, January 2022, 0.7]
[0064] (1-2): According to the requirements of GPS (Global Positioning System), warehouse area information may include the shelf location information, shelf type, shelf load threshold range and cargo position information corresponding to the warehouse for inspection, qualified product warehouse, delivery warehouse, unqualified product warehouse, return warehouse and special warehouse for weighing Chinese herbal medicine pieces. Considering the storage characteristics of pharmaceuticals, the qualified product warehouse is divided into general warehouse, insulation warehouse, refrigeration warehouse, constant temperature and humidity warehouse, dangerous goods warehouse and controlled atmosphere warehouse. The storage areas with different storage functions are regionally coded to realize the specialized storage of goods to be stored.
[0065] (1-3): Check the incoming goods information, that is, check the packaging, quantity, batch and other information of the incoming goods. The incoming goods that meet the requirements can enter the warehouse for storage, and the incoming goods that do not meet the requirements will enter the return storage area for the return process.
[0066] (2): Allocate the goods to be stored to the corresponding partition of the pharmaceutical warehouse according to the storage characteristics in the attribute information of the goods to be stored.
[0067] (2-1): Combined with the first storage characteristics of the goods to be stored, that is, the storage conditions, such as whether the drugs to be stored need to be stored at low temperatures, whether they need to be stored in a cool environment, whether they are flammable, explosive, toxic, or radioactive goods, whether they need to be stored under a specific gas concentration, etc., select the best storage partition for the goods to be stored in the qualified product warehouse.
[0068] The storage conditions of this batch of goods show that the drugs to be stored do not need to be stored at low temperatures, but only need to be stored in a cool environment. They are not flammable, explosive, toxic, or radioactive goods, and do not need to be stored under specific gas concentrations. In the qualified product warehouse, a ventilated and dry general storage partition is selected. There are 4 steel shelves in the general storage partition, forming two lanes. Each shelf has two layers and two rows, that is, the number of available cargo spaces on one shelf is 4, and the shelf load threshold is 150kg.
[0069] (2-2): Calculate the minimum number of cargo spaces required for the goods to be stored based on the cargo space attributes of the partition selected in (2-1). It should be noted that for a pharmaceutical warehouse, different batches of the same goods require different cargo spaces for storage.
[0070] Taking into account the aisle-oriented principle in warehouse management, that is, placing the product labels facing the aisle, not considering stacking the goods in the depth direction, and considering the loss of goods caused by stacking and squeezing of goods, consider discharging the goods in the length direction of the cargo space.
[0071] Suppose the set of goods to be stored is M, and the index value is i, then the goods to be stored m i The minimum number of cargo spaces required is expressed as:
[0072]
[0073] Among them, L i Indicates the goods to be stored m i The minimum number of cargo spaces required, according to L i Allocate cargo space in batches for incoming goods. i Indicates the goods to be stored m i Width, Q i Indicates the goods to be stored m i The quantity is represented by L, which represents the length of the unit storage space.
[0074] Calculate the minimum number of cargo spaces required for this batch of goods to be stored based on the cargo space attributes. The unit cargo space length is 2m; the number of cargo spaces required for this batch of goods is: L1 = 2, L2 = 1, L3 = 1, L4 = 2, L5 = 1, L6 = 1, L7 = 2, L8 = 1.
[0075] It should be noted that if the number of storage locations required for the goods to be stored exceeds one, it is for goods with higher popularity. In order to ensure safety inventory, more than one storage location must be stored at one time. For other goods with average or lower popularity, one storage location can be arranged to store the goods to meet the storage needs. Therefore, the process of allocating storage locations for the second batch of goods will not consume more computing power and time, that is, it will not delay the storage location allocation process for the goods to be stored too much.
[0076] (3): For the goods allocated to each partition of the pharmaceutical warehouse, a comprehensive analysis of multiple storage location optimization rules is performed to allocate the storage location for the first batch of goods.
[0077] According to the number of cargo spaces required for the goods to be stored, the set of cargo spaces allocated in the first batch is i∈{1, 2, 3, 4, 5, 6, 7, 8}.
[0078] (3-1): Construct the cargo turnover rate target optimization function.
[0079] Based on the turnover rate as the basic rule, the higher the turnover rate of the goods to be stored, the closer they should be to the entrance and exit. The target optimization function of the goods turnover rate with the cargo location as the research object is expressed as:
[0080]
[0081] F1 is the target optimization function constructed based on the cargo turnover rate, x and y are the position coordinates of the length and height of the shelf, z is the position coordinates of the zth row of the shelf in the warehouse where the goods are located; a is the total number of rows of shelves in the warehouse where the goods are located, c is the total number of mezzanines in each row of shelves in the warehouse where the goods are located, and b is the total number of grids in each shelf layer in the warehouse where the goods are located; t x,y,z v is the time it takes for the picking machine to run from the cargo location (z, x, y) to the shelf entrance, x 、v y 、v z are the horizontal running speed, vertical running speed and running speed between lanes of the picking machine in an aisle; q z,x,y is the turnover rate of the cargo at the cargo location (z, x, y); L, H, D are the length, height and distance between two adjacent lanes of the unit cargo location; when there is no cargo at the cargo location (z, x, y), q z,x,y =0.
[0082] For this batch of goods to be stored, x∈{1,2}, y∈{1,2}, z∈{1,2,3,4} are the length, height, and row position coordinates of the shelf; L=2m, H=1m, D=3m are the length, height, and distance between two adjacent lanes of the unit storage space, v x =0.5m / s,v y =0.4m / s,v z =0.5m / s are the horizontal and vertical running speeds of the picking machine in an aisle.
[0083] It should be noted that the cargo turnover rate is calculated based on the historical order data of medicines in the RMFS (Robotic Fulfillment Systems) system, that is, the total number of goods shipped out / average inventory.
[0084] (3-2): Construct cargo correlation objective optimization function
[0085] The community discovery problem is to use the local characteristics of single nodes in the network and their mutual correlation to transform part of the network into a highly correlated internal subset, where the nodes between different communities are only sparsely linked.
[0086] Here, the community discovery algorithm is used to realize the clustering of medical goods, to achieve strong correlation of goods attributes within the same community and weak correlation of goods between different communities.
[0087] According to RMFS (Robotic Fulfillment Systems), the historical order data of medicines in the system maps the correlation between each product into an undirected graph. i As a node v in an undirected graph i , we can get the node set V of the undirected graph = {v1, v2, ..., v N}, where the subscript N indicates that there are N types of drugs. Establish the feature vector of pharmaceutical goods, including drug efficacy, expiration date, batch number, etc. Each node is composed of attribute features {u i1 ,…,u iT} indicates that the attribute characteristics are obtained by identifying the information on the drug packaging through OCR (Optical Character Recognition) technology when the drug goods are first put into the warehouse. At the same time, the components of each dimension are standardized so that each dimension satisfies the standard normal distribution. Define the edge set of the undirected graph as E, the weight matrix as W, and the element w in the matrix ij Represents node v i With v j The size of the goods correlation is calculated as follows:
[0088]
[0089] o i Indicates cargo m i Does it exist in order o? If it exists in order o, then o i =1, otherwise 0. O represents the historical order set. The first term of the formula is the goods m i and cargo j The number of orders that appear in one order at the same time accounts for the proportion of the total orders. The latter formula represents the number of goods m i and cargo j The similarity difference of the attributes. α and β are parameter correction factors, making the weight w ij Keep it within the range greater than the threshold τ. If the calculated edge weight is less than the threshold τ, in order to ensure the accuracy of the subsequent community division, the edge weight is set to 0, that is, the edge connecting the two nodes is removed in the undirected graph. If there is an independent node in the undirected graph, it will be deleted from the graph.
[0090] For this batch of goods to be stored, the undirected correlation graph between the goods is as follows Figure 2 shown.
[0091] The Louvain algorithm is used to divide the undirected graph into communities until the number of communities formed reaches a preset set value Φ. In this embodiment, the number of communities formed is set to be no more than 3, that is, Φ = 3. The first element of each set is the central element in the community. The three sets are: s1 = {6, 1, 5}, s2 = {2, 7}, s3 = {3, 4, 8}.
[0092] Based on the community subgraph set generated above, the node with the largest degree in the community subgraph is taken as the central cargo, and other cargoes in the set are placed near the central cargo of the set. Then the correlation objective optimization function of a single cargo set is:
[0093]
[0094] Or written as:
[0095]
[0096] Among them, F′ i is the i-th community subgraph s i The correlation objective optimization function of the goods in j is the i-th community subgraph s i The shelf ranking of other goods that are not the central goods, x j is the i-th community subgraph s i The grid position of the mezzanine where other goods that are not the central goods are located, y j is the i-th community subgraph si The coordinates of the interlayer where the other goods that are not the central goods are located (a i , b i , c i ) is the community subset s i The center cargo coordinates of .
[0097] Specifically, for this batch of goods to be stored, the correlation objective optimization function of a single set of goods is:
[0098]
[0099] Among them, the coordinates (a6, b6, c6) are the central cargo coordinates of community set s1.
[0100]
[0101] Among them, the coordinates (a2, b2, c2) are the central cargo coordinates of community set s2.
[0102]
[0103] Among them, the coordinates (a3, b3, c3) are the center cargo coordinates of community set s3.
[0104] The objective optimization function for constructing cargo correlation is:
[0105]
[0106] Among them, F2 is the overall cargo relevance objective optimization function.
[0107] Specifically, for this batch of goods to be stored, the goods correlation objective optimization function constructed is:
[0108]
[0109] (3-3): Construct shelf stability objective optimization function.
[0110] According to the storage weight characteristic law, the height of the storage location is determined according to the weight of the items. Heavy items should be stored in the lower layer of the shelf, while light items should be stored in the upper layer of the shelf, so the shelf stability target is equivalent to the lowest center of gravity target of the goods:
[0111] Construct the shelf stability objective optimization function:
[0112]
[0113] Among them, F3 is the shelf stability objective optimization function, W z,x,y is the unit weight of the goods placed at the cargo location (z, x, y), nz,x,y The quantity of goods placed at the cargo location (z, x, y).
[0114] Specifically, for this batch of goods to be stored, the constructed shelf stability objective optimization function is:
[0115]
[0116] It should be noted that n z,x,y and Q i There is a difference, Q i For cargo m i The total number of z,x,y is the quantity of goods stored at the current cargo location (z, x, y), i.e. Q i ≥n z,x,y .
[0117] (4): After the storage location allocation for the first batch of goods is completed, the storage location allocation for the second batch of goods is carried out based on load balancing.
[0118] (4-1): Step (3) realizes the location allocation of the first batch of goods, that is, each type of goods is considered to have one cargo location for location allocation.
[0119] Considering that areas with high turnover rates are prone to congestion when sorting using AGV carts, balanced cargo load is considered when allocating storage locations for a small number of second batches of goods.
[0120] Specifically, for this batch of goods to be stored, select the second batch of goods to be stored m i′ For the set {1, 4, 7}, the required storage space is 1.
[0121] (4-2): According to the first batch of goods location distribution, the load balance of each row of shelves is obtained, which is expressed as:
[0122]
[0123] Among them, G z is the load balancing value of each row of shelves after the first batch of goods is allocated. It is only affected by the allocation of the first batch of goods. z,x,y S is the status mark of the cargo location. z,x,y =1 means that the cargo has been placed at the cargo location (z, x, y), S z,x,y = 0 means that the cargo location (z, x, y) is idle, that is, there is no cargo on the cargo location (z, x, y). z is the type of goods that have been allocated to the zth row of shelves in the first batch. If S z,x,y =0, then q (z,x,y) =0, if S z,x,y =1, then qz,x,y is the turnover rate of the goods stored at the cargo location (z, x, y).
[0124] Specifically, for this batch of goods to be stored, the shelf load balance is expressed as:
[0125]
[0126] Based on the allocation of storage locations for the first batch of goods, the second batch of goods will be allocated in the unoccupied storage locations for the first batch of goods.
[0127] The load balance of each row of shelves obtained by the storage location allocation of the second batch of goods is expressed as:
[0128]
[0129] Among them, G′ z is the load balancing value of each row of shelves after the second batch of goods is allocated, which is only affected by the allocation of the first batch of goods! is the inversion operation, if S z,x,y =1, then! S z,x,y =0, if S z,x,y =0, then! S z,x,y =1; R′ z The types of goods in the second batch to be allocated to the zth row of shelves.
[0130] Specifically, for this batch of goods to be stored, the load balance of each row of shelves obtained by allocating the second batch of goods can be expressed as:
[0131]
[0132] It should be noted that an aisle contains two rows of shelves on the left and right sides. For example, aisle 1 can access shelves 1 and 2. The load balancing degree of the aisle is expressed as:
[0133]
[0134] Among them, GX p The load balance value of the pth aisle after the first batch of goods is allocated, which includes the load mean of the left shelf G 2·p-1 , and the load balancing value G of the right shelf 2·p ; GX′ p is the load balance value of the pth aisle after the second batch of goods is allocated, which includes the load mean of the left shelf G′ 2·p-1 , and the load balance value G′ of the right shelf 2·p ;
[0135] Then the optimization objective function of lane load balance is:
[0136]
[0137] Among them, F4 is the optimization objective function of lane load balance, is the maximum value of the load balance degree of each lane.
[0138] (5): After the second batch of goods storage location allocation is completed, a multi-objective storage location optimization model is constructed and the simulated annealing genetic algorithm is used to solve the model to complete the storage location optimization of the pharmaceutical warehouse.
[0139] (5-1): According to step (3), the first batch of goods is allocated storage space, and weights β1, β2, and β2 are assigned to the goods turnover rate optimization function, the goods correlation optimization function, and the shelf stability optimization function, respectively, to represent the importance of each optimization goal. In this way, the multi-objective optimization problem is converted into a single-objective optimization problem. The final optimized objective function is:
[0140]
[0141] Among them, F is the cargo location optimization objective function of the first batch of cargo constructed based on cargo-related attributes, β1+β2+β3=1.
[0142] Specifically, for this batch of goods to be stored, the parameters are set to β1=0.4, β2=0.3, and β3=0.3.
[0143] The second batch of goods is allocated based on the allocation of the storage locations of the first batch of goods to meet the load balance of each shelf. The weights γ1 and γ2 are assigned to the shelf stability objective function and the aisle load balance objective function respectively. The multi-objective optimization problem is converted into a single-objective optimization problem. The storage location optimization objective function of the second batch of goods is:
[0144]
[0145] Among them, F′ is the cargo location optimization objective function of the second batch of cargo constructed based on cargo-related attributes, γ1+γ2=1.
[0146] Specifically, for this batch of goods to be stored, the parameters are set to γ1=0.5, γ2=0.5.
[0147] (5-2): According to the constructed cargo location optimization model, the model is solved by using twice simulated annealing genetic algorithm, with F and F′ as the objective functions respectively, to obtain the optimal solution for cargo location.
[0148] The flowchart of simulated annealing genetic algorithm is as follows Figure 3 shown.
[0149] The encoding method adopted is integer arrangement method, and the specific method is as follows: each chromosome represents a cargo location. The information contained in the cargo location includes the aisle number, column number, layer number and other information of the cargo location. The number of the initial population has been determined, and the value of the gene is the coordinate of the cargo location represented by an integer, such as 1 (2, 3, 4) means that Class 1 cargo is stored in the 2nd aisle, the third column, and the third layer.
[0150] The population represents the number of goods to be stored. Roulette selection is used, with two individuals in a group, one as the father and one as the mother for replication.
[0151] The crossover probability selected in the crossover operation is 0.5, and the crossover operation is performed using the sequential crossover method. Let a gene position of the chromosomes of the father and mother selected according to the crossover probability be exchanged, such as the position (1, 2, 3) and the position (2, 3, 4) randomly exchange two gene positions in the same position, thus forming two new individuals (1, 2, 4) and (2, 3, 3).
[0152] The mutation probability selected in the mutation operation is 0.05, and the mutation operation is performed using the exchange mutation method. The fitness values of the child individuals generated by selection, crossover, and mutation are compared with those of the parent generation.
[0153]
[0154] Among them, E(x new ) is the fitness value of the offspring, E(x old ) is the fitness value of the parent generation, and T is the maximum evolutionary generation.
[0155] If the fitness value of the child individual is less than that of the parent, the child individual is accepted unconditionally; if it is greater than that of the parent individual,
[0156]
[0157] Receive sub-individual, when the probability P t If the value is greater than [0, 1), the individual is accepted; if not, the parent individual is retained. Set the initial temperature T to 1000, the temperature drop rate δ to 0.8, and perform cyclic annealing operations on the individual and the parent individual until a new individual is found. Select the maximum number of iterations to be 400. If the maximum number of iterations is not reached, set the temperature T = T × δ. When the termination number is 400, the cycle will end and the calculation will be stopped. The most suitable value, that is, the optimal solution or satisfactory solution, will be output.
[0158] Specifically, for this batch of goods to be stored, the storage location allocation results of the first batch of goods are shown in Table 1, and the storage location allocation results of the second batch of goods are shown in Table 2.
[0159] Table 1 The results of cargo space allocation for the first batch of goods
[0160] goods Row layer List 1 1 1 1 2 2 1 1 3 3 1 1 4 3 2 2 5 1 1 2 6 1 2 1 7 2 1 2 8 3 1 2
[0161] G1=2.48, G2=2.15, G3=5.98, G4=0
[0162] Table 2 The results of cargo space allocation for the second batch of goods
[0163]
[0164]
[0165] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for allocating cargo space in a pharmaceutical warehouse. Its characteristics are: include: Obtain the attribute information of the goods to be stored and the partition layout information of the pharmaceutical warehouse; Allocate the goods to be stored to the partitions corresponding to the pharmaceutical warehouse according to the storage characteristics in the attribute information of the goods to be stored; For the goods allocated to each partition of the pharmaceutical warehouse, the first batch of goods storage location allocation results are obtained by solving the first batch multi-objective cargo location optimization model; Taking the storage location allocation result of the first batch of goods as a constraint, the storage location allocation result of the second batch of goods is obtained by solving the multi-objective storage location optimization model of the second batch; The first batch of multi-objective cargo location optimization models are constructed and obtained by taking cargo turnover rate, cargo correlation and shelf stability as optimization targets; the second batch of multi-objective cargo location optimization models are constructed and obtained by taking shelf stability, shelf load balance and aisle load balance as optimization targets; The method for constructing the first batch of multi-objective cargo location optimization models includes: Taking goods turnover rate, goods correlation and shelf stability as optimization targets, the corresponding objective functions are constructed; Assign weights to the cargo turnover rate objective function, cargo relevance objective function, and shelf stability objective function respectively , ; The first batch multi-objective cargo location optimization model is constructed based on the cargo turnover rate objective function, cargo correlation objective function, shelf stability objective function and the weights of each objective function; Among them, the cargo turnover rate objective optimization function is: (2) in, is the objective optimization function constructed based on the cargo turnover rate, The warehouse where the goods are located Shelf location coordinates, , is the location coordinates of shelf length and height; is the total number of shelves in the warehouse where the goods are located, is the total number of mezzanines in each row of shelves in the warehouse where the goods are located, The total number of shelves on each shelf level in the warehouse where the goods are located; For picking machines from the goods to be shipped out of the warehouse The time it takes to run to the shelf entrance and exit, , are the horizontal and vertical running speeds of the picking machine in an aisle, respectively. The speed at which the picking machine moves between aisles; For cargo space The turnover rate of goods on When there is no cargo on the cargo space, ; , , The length, height and distance between two adjacent aisles of the unit cargo space; The cargo correlation objective optimization function is: (5) in, Optimize the function for the overall cargo relevance objective, Indicates the preset community value. For the The relevance objective optimization function of the goods in the community subgraph; The shelf stability objective optimization function is: (6) in, Optimize the function for the shelf stability objective, To place in the cargo space The unit weight of the goods, To place in the cargo space The quantity of goods.
2. The method for allocating cargo space in a pharmaceutical warehouse according to claim 1, Its characteristics are: The attribute information of the goods to be stored includes not only storage characteristics but also: the variety, specification, origin, specification packaging, and batch of the goods to be stored; wherein the storage characteristics include storage at room temperature, refrigeration, constant temperature and constant humidity.
3. The method for allocating cargo space in a pharmaceutical warehouse according to claim 1, Its characteristics are: The method for constructing the cargo correlation objective optimization function includes: The community discovery algorithm is used to cluster medical goods, achieving strong correlation of goods attributes within the same community and weak correlation of goods between different communities; According to the mobile robot fulfillment system, the historical order data of medicines in the system maps the correlation between each product into an undirected graph; As a node in an undirected graph , thus the node set of the undirected graph is , where the subscript Indicates shared Class drugs; establish the feature vector of medical goods, including drug efficacy, expiration date and batch number, and each node is composed of attribute features Indicates that the attribute characteristics are obtained by identifying the information on the drug packaging through optical character recognition technology when the drug goods are first put into the warehouse; at the same time, the components of each dimension are standardized so that each dimension satisfies the standard normal distribution; the edge set of the undirected graph is defined as , the weight matrix is , the elements in the matrix Representation Node and The size of the goods correlation is calculated as follows: (3) in, Indicates goods Is there an order? , if present in the order ,but , otherwise 0; Represents a set of historical orders; For goods and goods The proportion of orders that appear in one order at the same time to the total orders, Indicates goods and goods Similarity differences of attributes; , is the parameter correction factor, making the weight Keep above the threshold If the calculated edge weight is less than the threshold , set the edge weight to 0, that is, remove the edge that connects the two nodes in the undirected graph; if there is an independent node in the undirected graph, delete it from the undirected graph; Considering that different drugs have different drug properties and are prone to react with other special types of drugs, affecting the drug properties, the weight of the edge between the nodes corresponding to the two pharmaceutical goods that react with each other is set to 0 when the model is established, that is, this edge is removed in the undirected graph; Use the Louvain algorithm to divide the undirected graph into communities until the number of communities formed reaches the preset community subgraph set value; Based on the generated community subgraph collection , where each subset consists of the subscripts of the sets of goods with greater correlation, Represents the preset community value. The node with the largest degree in the community subgraph is taken as the central cargo, and other cargoes in its collection are placed near the central cargo of the collection. Then the cargo correlation objective optimization function in a single community subgraph collection is: (4) in, For the Community subgraph The correlation objective optimization function of the goods in For the Community subgraph The shelf ranking of other goods that are not the central goods. For the Community subgraph The grid position of the mezzanine where other goods that are not the central goods are located, For the Community subgraph The coordinates of the mezzanine where the other goods that are not the central goods are located Community subset The center cargo coordinates of; The cargo relevance objective optimization function is obtained according to the cargo relevance objective optimization function in a single community subgraph set.
4. The method for allocating cargo space in a pharmaceutical warehouse according to claim 1, Its characteristics are: According to the storage location allocation results of the first batch of goods, the load balance of each row of shelves is obtained: (7) in, It is the load balancing value of each row of shelves after the first batch of goods is allocated. It is only affected by the allocation of the first batch of goods. For cargo space Status indicator, Indicates the cargo location The goods have been placed. It indicates the cargo location In idle state, The first batch has been allocated to The types of goods on the shelves, if ,but ,like ,but For storage in cargo space The turnover rate of goods on the The load balance of each row of shelves obtained by the storage location allocation of the second batch of goods is expressed as: (8) in, It is the load balancing value of each row of shelves after the second batch of goods is allocated. It is only affected by the allocation of the first batch of goods. For the negation operation, if ,but ,like ,but ; The second batch is to be allocated to The types of goods on the shelves.
5. The method for allocating cargo space in a pharmaceutical warehouse according to claim 4, Its characteristics are: An aisle contains two rows of shelves on the left and right sides. The load balance of the aisle is expressed as: (9) in, After the first batch of goods is allocated, The load balancing value of each aisle, which includes the load average of the rack on the left. , and the load balancing value of the right shelf ; After the allocation of the cargo space for the second batch of goods is completed The load balancing value of each aisle, which includes the load average of the rack on the left. , and the load balancing value of the right shelf ; The optimization objective function of lane load balance is: (10) in, is the optimization objective function of lane load balance, is the maximum value of the load balance degree of each lane.
6. The method for allocating cargo space in a pharmaceutical warehouse according to claim 5, Its characteristics are: The first batch of multi-objective cargo space optimization models are constructed as follows: (11) in, The objective function for optimizing the cargo location of the first batch of cargo is constructed based on cargo-related attributes; Taking the storage location allocation result of the first batch of goods as a constraint, the storage location allocation result of the second batch of goods is obtained by solving the multi-objective storage location optimization model of the second batch, including: assigning weights to the shelf stability objective function and the aisle load balancing objective function respectively , transforming the multi-objective optimization problem into a single-objective optimization problem, the second batch of multi-objective cargo location optimization model is constructed as follows: (12) in, The objective function for optimizing the location of the second batch of goods based on the relevant attributes of the goods is constructed. .
7. The method for allocating cargo space in a pharmaceutical warehouse according to claim 6, Its characteristics are: Respectively and As the objective function, twice simulated annealing genetic algorithm is used to solve the model and obtain the optimal solution for cargo location.
