CN110909930B - Goods position distribution method of mobile goods shelf storage system for refrigeration house - Google Patents

Abstract

A goods position distribution method of a mobile goods shelf storage system facing a refrigeration house distributes items with strong correlation to the same sorting roadway, reduces the possibility of opening the sorting roadway for many times, takes the similarity coefficient of order items as the basis of the correlation size, comprehensively considers the item sorting frequency and the goods shelf gravity center, establishes a multi-target goods position distribution optimization model, then solves by adopting an improved invasive weed algorithm to obtain the optimal storage position of goods, generates part of initial populations by adopting a greedy algorithm, then sets a reasonable space diffusion operator, and finally introduces the evolution reversal operation of a genetic algorithm. The invention has stronger overall search and local search capability, obvious optimization effect and effectively improved warehouse picking efficiency and shelf stability.

Description

A kind of mobile rack storage system cargo space allocation method for cold storage

technical field

The invention belongs to the field of storage management, and in particular relates to a storage space allocation method of a mobile rack storage system oriented to a cold storage.

Background technique

In recent years, with the rapid development of the cold chain logistics industry, cold storage has received more and more attention from logistics companies. Warehouse energy consumption, investment cost and efficiency have always been the pain points in cold storage. Therefore, choosing a storage system with compact access space and timely service time has become a new direction for cold storage development. As a new compact storage system, the mobile rack storage system only needs to set aside one picking aisle for the access trolley to operate, and allocate the aisle through the movement of the shelves, and then the access trolley enters the aisle to complete the entry and exit of the goods. The storage system has simple structure, high space utilization and low cost, and has been widely used in major cold storages at home and abroad. The problem of storage space allocation is the key problem of the mobile rack storage system for cold storage, which is directly related to the efficient and stable operation of the warehouse. How to optimize the operation efficiency of the mobile rack storage system through a reasonable storage space strategy Stability has become an urgent problem to be solved.

Commonly used storage space allocation strategies include: positioning storage, random storage, nearby storage, full turnover storage and classified storage. However, in the prior art, there are still few researches on the storage space allocation of the mobile rack storage system for cold storage, and there is no literature to study the specific storage position of each item in the storage system. In view of the logistics characteristics of cold storage logistics such as a wide variety of storage varieties, high timeliness requirements, high costs, and complex technical requirements, adopting a reasonable slot allocation strategy can improve the response speed of cold storage orders, reduce cold storage costs and improve shelf stability.

SUMMARY OF THE INVENTION

In order to overcome the deficiencies of the prior art, the present invention provides a storage space allocation method for a mobile rack storage system oriented to a cold storage, so as to optimize the storage mode of the warehouse and reduce the cost of warehouse management.

The technical scheme adopted by the present invention to solve its technical problems is:

A storage space allocation method for a mobile rack storage system for cold storage, comprising the following steps:

Step 1. Establish a multi-objective inventory allocation optimization model with the goal of improving picking efficiency, improving shelf stability, and improving the correlation of items in the same picking lane;

Calculate the objective function f ₁ to improve the picking efficiency,

First calculate the movement of the access trolley from the I/O position to the cargo position (x, y, z). Assuming that the shelf needs to be moved, the running time in the horizontal direction is:

Among them, y represents the coordinates of the cargo position (x, y, z) along the y direction, w represents the width of the cargo grid, v _y represents the speed of the access trolley along the y direction; t _r represents the time when the rack moves to open the picking lane, and k represents It is necessary to pick the goods at the cargo position (x, y, z) in the picking lane k, d represents the depth of the cargo compartment, l represents the width of the picking lane, v _x represents the speed of the access trolley along the x direction, as can be seen from Figure 2 , the goods on the first row of shelves need to be picked from lane 1, the goods on the second and third rows of shelves are picked from lane 2, ..., and so on;

Then calculate that the access trolley moves from the I/O position to the cargo position (x, y, z), and the running time in the vertical direction is:

Among them, z represents the coordinates of the cargo space (x, y, z) along the z direction, h is the height of the cargo compartment, and v _z represents the speed of the access trolley in the vertical direction;

Continue to calculate that the access trolley returns from the cargo position (x, y, z) to the I/O position. Since the shelf does not move, the running time on the horizontal plane when returning is:

Then calculate the running time in the vertical direction when returning as:

Further, since the movement of the access trolley on the horizontal plane and the movement in the vertical direction are simultaneous, the time used for picking goods is the maximum running time in the two directions, and the picking position (x, y, z) is calculated. The time t _xyz for the goods is:

Among them, t _p is the loading and unloading time of the access trolley;

Then the time taken to pick the item at location (x, y, z) is:

t _xyz ·p _xyz (6)

Among them, p _xyz represents the picking frequency of the item stored at the slot (x, y, z).

Further, the main goal of improving the picking efficiency is to minimize the total picking time of the goods, then the objective function expression is:

Among them, f ₁ represents the total picking time of the goods;

Calculate the objective function f ₂ to improve shelf stability,

Among them, f ₂ represents the height of the overall center of gravity of the shelf, and m _xyz represents the weight of the item stored at the cargo position (x, y, z);

Calculate the objective function f ₃ to improve the item correlation of the same picking lane,

The formula for calculating the similarity coefficient proposed by Russel and Rao is as follows:

Among them, a represents the order quantity that contains both item i and item j; b represents the order quantity that only contains item i; c represents the order quantity that only contains item j; d represents that neither item i nor item j is included. the number of orders included;

Storing highly correlated items on the shelves in the same picking lane can effectively reduce the number of opening picking lanes. Further, the objective function can be transformed into the reciprocal of the sum of the item similarity coefficients in the same picking lane as much as possible. small, the objective function expression is:

Among them, f ₃ represents the reciprocal of the sum of the item similarity coefficients of the same picking lane, K represents the number of picking lanes, i and j represent the item number, g represents the number of items, and r _ik =1 represents the storage of item i On the shelf of the picking lane k, otherwise r _ik = 0, r _jk is the same;

Further constraints on the model are as follows:

表示每个品项只能占用一个货位；Among them, 1≤x≤a means the limit on the number of shelf rows; 1≤y≤b means the limit on the number of shelves; 1≤z≤c means the limit on the number of shelf layers;

Indicates that each item can only occupy one slot;

Step 2. Build an evaluation function.

The ideal point method is used to process these three objective functions, and an evaluation function is constructed. First, find the optimal value f _i ^* of each objective function f _i and use it as an ideal point;

Based on the distance between each objective function value and the ideal point, construct the evaluation function of each objective function:

F _i =(f _i -f _i ^* ) ² (12)

Among them, F _i represents the evaluation function value of the ith objective function; f _i represents the function value of the ith objective function; f _i ^* represents the optimal value of the ith objective function;

Further, on the basis of the above formula, the weight coefficient λ _i is introduced, the sum of which is 1, and the multi-objective optimization function is converted into an evaluation function:

Among them, f is the multi-objective optimization evaluation function.

Step 3. Coding design: Number the items and the goods at the same time, and use the natural number arrangement coding method. The length of the code depends on the number of items. When the number of items is N, a code consists of N non-repeating natural numbers. Each natural number corresponds to a slot number; Figure 4 is an example of a coding diagram, that is, item 1 is stored in slot 13, item 2 is stored in slot 3, ..., and so on, until N items are all A slot has been allocated.

Step 4. Generate initial population: initialization algorithm related parameters: initial number of weeds N ₀ , maximum number of weed population N _max , maximum number of iterations iter _max , maximum number S _max and minimum number S of seeds that each weed can produce _min , the nonlinear modulation index n, the initial value of standard deviation σ _init and the final value of standard deviation σ _final when weeds spread in space;

Step 5. Record the objective function value of each weed, the objective function value includes the previous f ₁ , f ₂ , f ₃ and f, and then record the optimal weed individual and the optimal solution.

Step 6. Weed propagation stage: According to the objective function value of each weed, the following formula is used to calculate the number of seeds produced by each weed.

表示向下取整；Among them, f represents the objective function value of the current weeds, f _max and f _min represent the maximum and minimum value of the objective function of the weeds in the current population, respectively,

means round down;

Step 7. Spatial diffusion stage: Weed seeds are scattered around the parent weeds according to a normal distribution with a mean of 0 and a standard deviation of σ. As the number of iterations increases, σ will also decrease from the initial value σ _init to the final value σ _final . For a certain generation, the standard deviation is calculated as follows:

Among them, σ represents the standard deviation value corresponding to the current algebra; iter _max represents the maximum number of iterations; iter represents the current algebra; σ _init represents the initial value of the standard deviation; σ _final represents the final value of the standard deviation;

Step 8. Competitive survival stage: After merging the parent and child individuals in the population into a new population, in order to improve the local search ability of the proposed algorithm, the evolutionary reversal operator of the genetic algorithm is introduced, and two positions are randomly selected, and the The numbers in the two positions are interchanged. If the value of the objective function decreases, the individual will be accepted, otherwise the evolutionary reversal will be invalid. Then the new population is sorted according to the objective function value, leaving excellent individuals with small objective function values, and eliminating weak individuals with large objective function values, and the number of individuals cannot exceed the maximum number N _max of the population;

Step 9: After one iteration is completed, determine whether the maximum iteration times iter _max is reached, if so, output the optimal weed individual; otherwise, return to step 6.

Further, the process of step 4 is as follows:

Step 4.1, first calculate the time t _xyz required to pick the goods at the location (x, y, z), and create a picking time matrix, from which the time t _xyz required to pick the goods of any location can be known;

Step 4.2. Use the greedy algorithm to generate 40% of the initial weed population, and the remaining 60% of the weed population is randomly generated. In the initial weeds generated by the greedy algorithm, two greedy strategies are used. The first is to give priority to short selection times. The second is to give priority to the storage space with a small number of layers.

Further, the process of step 4.2 is as follows:

Step 4.2.1. The method of selecting the location with short picking time is preferred: take the number of each location and the corresponding picking time out to create a matrix, the first column is the number of the location, the second column is the corresponding picking time, and then It is scrambled, and the slot number and picking time of each row are still corresponding, and then they are arranged in ascending order in the second column, and some of the population codes generated at this time correspond to the first column one-to-one;

Step 4.2.2. The method of choosing a small number of layers is preferred: take out the number of each location and the number of layers it is in to create a matrix, the first column is the number of the location, the second column is the corresponding layer, and then mark it. If it is messy, the slot number and layer number of each row are still corresponding, and then they are arranged in ascending order in the second column. At this time, some of the generated population codes are in one-to-one correspondence with the first column.

Further, the process of the step 7 is as follows:

Step 7.1. Using the space diffusion operator based on sliding insertion, first randomly generate two random positions j and k, then use the slot numbers from position j to position k as the queue, the number at position j as the head of the queue, and position k The number of is used as the tail of the queue, and then sequentially deleted from the tail of the queue and inserted into the head of the queue until d insertions are performed. A schematic diagram of the sliding insertion is shown in Figure 5.

作为滑动插入的执行次数d。Step 7.2. The dispersion process of seeds obeys a normal distribution with a mean of 0 and a standard deviation of σ. The size of σ determines the search range of seeds. Inspired by this, let the execution times d of sliding insertion obey N(0, σ ² ), take the absolute value of the random number α and round it up.

d as the number of executions of sliding insertion.

The beneficial effects of the present invention are mainly manifested in: there are few researches on the storage space allocation problem of the mobile rack storage system facing the cold storage, especially the previous research does not consider the specific storage location of each item, and the present invention considers the mobile storage system. To solve the problem that the heavy-duty racks take too long to open the picking lanes, it is proposed to assign the items with strong correlation to the same picking lanes to reduce the possibility of opening the picking lanes multiple times. Considering the frequency of item picking and the center of gravity of the shelf, a multi-objective cargo space allocation optimization model is established, and then the improved invading weed algorithm is used to solve the optimal storage position of the goods. Part of the initial population is generated by using the greedy algorithm, and then a reasonable space diffusion is set. Finally, the evolution reversal operation of the genetic algorithm is introduced. The global search and local search capabilities of the algorithm are strong, the optimization effect is remarkable, and the warehouse picking efficiency and shelf stability are effectively improved.

Description of drawings

Figure 1 is a schematic diagram of a mobile rack storage system, wherein 1 is a movable rack, 2 is a moving track, 3 is I/O, and 4 is an access trolley.

Figure 2 is a top view of the mobile rack storage system.

Figure 3 is a flowchart of the IIWO algorithm.

Figure 4 is a diagram of an example encoding.

Figure 5 is an example diagram of sliding insertion.

Figure 6 is a graph showing the simulation results of improving picking efficiency.

Figure 7 shows the simulation results of improving shelf stability.

Figure 8 is a graph showing the simulation result of improving item correlation in the same picking lane.

Fig. 9 is a multi-objective evaluation function simulation result diagram.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings.

Referring to Figures 1 to 9, a method for allocating cargo spaces in a mobile rack storage system for cold storage includes the following steps:

Step 1. Establish a multi-objective inventory allocation optimization model with the goal of improving picking efficiency, improving shelf stability, and improving the correlation of items in the same picking lane;

Calculate the objective function f ₁ to improve the picking efficiency,

First calculate the movement of the access trolley from the I/O position to the cargo position (x, y, z). Assuming that the shelf needs to be moved, the running time in the horizontal direction is:

Among them, y represents the coordinates of the cargo position (x, y, z) along the y direction, w represents the width of the cargo grid, v _y represents the speed of the access trolley along the y direction; t _r represents the time when the rack moves to open the picking lane, and k represents It is necessary to pick the goods at the cargo position (x, y, z) in the picking lane k, d represents the depth of the cargo compartment, l represents the width of the picking lane, v _x represents the speed of the access trolley along the x direction, as can be seen from Figure 2 , the goods on the first row of shelves need to be picked from lane 1, the goods on the second and third rows of shelves are picked from lane 2, ..., and so on;

Then calculate that the access trolley moves from the I/O position to the cargo position (x, y, z), and the running time in the vertical direction is:

Among them, z represents the coordinates of the cargo space (x, y, z) along the z direction, h is the height of the cargo compartment, and v _z represents the speed of the access trolley in the vertical direction;

Continue to calculate that the access trolley returns from the cargo position (x, y, z) to the I/O position. Since the shelf does not move, the running time on the horizontal plane when returning is:

Then calculate the running time in the vertical direction when returning as:

Further, since the movement of the access trolley on the horizontal plane and the movement in the vertical direction are simultaneous, the time used for picking goods is the maximum running time in the two directions, and the picking position (x, y, z) is calculated. The time t _xyz for the goods at is:

Among them, t _p is the loading and unloading time of the access trolley.

Then the time taken to pick the item at location (x, y, z) is:

t _xyz ·p _xyz (6)

Among them, p _xyz represents the picking frequency of the item stored at the slot (x, y, z).

Further, the main goal of improving the picking efficiency is to minimize the total picking time of the goods, then the objective function expression is:

Among them, f ₁ represents the total picking time of the goods.

Calculate the objective function f ₂ to improve shelf stability,

Among them, f ₂ represents the height of the overall center of gravity of the shelf, and m _xyz represents the weight of the item stored at the cargo position (x, y, z);

Calculate the objective function f ₃ to improve the item correlation of the same picking lane,

The formula for calculating the similarity coefficient proposed by Russel and Rao is as follows:

Among them, a represents the order quantity that contains both item i and item j; b represents the order quantity that only contains item i; c represents the order quantity that only contains item j; d represents that neither item i nor item j is included. The number of orders included.

Storing highly correlated items on the shelves in the same picking lane can effectively reduce the number of opening picking lanes. Further, the objective function can be transformed into the reciprocal of the sum of the item similarity coefficients in the same picking lane as much as possible. small, the objective function expression is:

Among them, f ₃ represents the reciprocal of the sum of the item similarity coefficients of the same picking lane, K represents the number of picking lanes, i and j represent the item number, g represents the number of items, and r _ik =1 represents the storage of item i On the shelf of the picking lane k, otherwise r _ik = 0, and the same is true for r _jk .

Further constraints on the model are as follows:

表示每个品项只能占用一个货位。Among them, 1≤x≤a means the limit on the number of shelf rows; 1≤y≤b means the limit on the number of shelves; 1≤z≤c means the limit on the number of shelf layers;

Indicates that each item can only occupy one slot.

Step 2. Build an evaluation function.

The present invention selects the ideal point method to process the three objective functions and constructs the evaluation function. First, find the optimal value f _i ^* of each objective function f _i and take it as the ideal point.

Further, based on the distance between each objective function value and the ideal point, the evaluation function of each objective function is constructed:

F _i =(f _i -f _i ^* ) ² (12)

Among them, F _i represents the evaluation function value of the ith objective function; f _i represents the function value of the ith objective function; f _i ^* represents the optimal value of the ith objective function.

Further, on the basis of the above formula, the weight coefficient λ _i is introduced, the sum of which is 1, and the multi-objective optimization function is converted into an evaluation function:

Among them, f is the multi-objective optimization evaluation function.

Step 3. Coding design: Number the items and the goods at the same time, and use the natural number arrangement coding method. The length of the code depends on the number of items. When the number of items is N, a code consists of N non-repeating natural numbers. Each natural number corresponds to a slot number. Figure 4 is an example diagram of a code, that is, item 1 is stored in slot 13, item 2 is stored in slot 3, ..., and so on, until N items have been allocated slots.

Step 4. Generate initial population: initialization algorithm related parameters: initial number of weeds N ₀ , maximum number of weed population N _max , maximum number of iterations iter _max , maximum number S _max and minimum number S of seeds that each weed can produce _min , the nonlinear modulation index n, the initial value of standard deviation σ _init and the final value of standard deviation σ _final when weeds are spreading in space.

Step 4.1, first calculate the time t _xyz required to pick the goods at the location (x, y, z), and create a picking time matrix, from which the time t _xyz required to pick the goods of any location can be known;

Step 4.2, use the greedy algorithm to generate 40% of the initial weed population, and the remaining 60% of the weed population is randomly generated. In the initial weeds generated by the greedy algorithm, two greedy strategies are adopted. The first is to preferentially select the slot with short picking time; the second is to preferentially select the slot with a small number of layers.

Step 4.2.1. The method of selecting the location with short picking time is preferred: take the number of each location and the corresponding picking time out to create a matrix, the first column is the number of the location, the second column is the corresponding picking time, and then It is scrambled, and the slot number and picking time of each row are still corresponding, and then they are arranged in ascending order in the second column, and some of the population codes generated at this time correspond to the first column one-to-one;

Step 4.2.2. The method of choosing a small number of layers is preferred: take out the number of each location and the number of layers it is in to create a matrix, the first column is the number of the location, the second column is the corresponding layer, and then mark it. Random, the slot number and layer number of each row are still corresponding, and then they are arranged in ascending order in the second column, and some of the population codes generated at this time correspond to the first column one-to-one;

Step 5. Record the objective function value of each weed, the objective function value includes the previous f ₁ , f ₂ , f ₃ and f, and then record the optimal weed individual and the optimal solution.

Step 6. Weed propagation stage: According to the objective function value of each weed, the following formula is used to calculate the number of seeds produced by each weed.

表示向下取整。Among them, f represents the objective function value of the current weeds, f _max and f _min represent the maximum and minimum value of the objective function of the weeds in the current population, respectively,

Indicates rounded down.

Step 7. Spatial diffusion stage: Weed seeds are scattered around the parent weeds according to a normal distribution with a mean of 0 and a standard deviation of σ. As the number of iterations increases, σ will also decrease from the initial value σ _init to the final value σ _final . For a certain generation, the standard deviation is calculated as follows:

Among them, σ represents the standard deviation value corresponding to the current algebra; iter _max represents the maximum number of iterations; iter represents the current algebra; σ _init represents the initial value of the standard deviation; σ _final represents the final value of the standard deviation.

Step 7.1. Using the space diffusion operator based on sliding insertion, first randomly generate two random positions j and k, then use the slot numbers from position j to position k as the queue, the number at position j as the head of the queue, and position k The number of is used as the tail of the queue, and then sequentially deleted from the tail of the queue and inserted into the head of the queue until d insertions are performed. A schematic diagram of the sliding insertion is shown in Figure 5.

作为滑动插入的执行次数d。Step 7.2. The dispersion process of seeds obeys a normal distribution with a mean of 0 and a standard deviation of σ. The size of σ determines the search range of seeds. Inspired by this, let the execution times d of sliding insertion obey N(0, σ ² ), take the absolute value of the random number α and round it up.

d as the number of executions of sliding insertion.

Step 8. Competitive survival stage: After merging the parent and child individuals in the population into a new population, in order to improve the local search ability of the proposed algorithm, the evolutionary reversal operator of the genetic algorithm is introduced. Two positions are randomly selected, and the numbers of the two positions are exchanged. If the value of the objective function decreases, the individual is accepted, otherwise the evolutionary reversal is invalid. Then the new population is sorted according to the objective function value, leaving excellent individuals with a small objective function value, and eliminating weak individuals with a large objective function value, and the number of individuals cannot exceed the maximum number N _max of the population.

Step 9: After one iteration is completed, determine whether the maximum iteration times iter _max is reached, if so, output the optimal weed individual; otherwise, return to step 6.

Taking the actual cold storage mobile shelf storage system of an enterprise as the research object, programming and simulation in matlab. The basic parameters of the simulation of the mobile rack storage system studied in the present invention are shown in Table 1.

Table 1 shows the basic parameters of the mobile rack storage system simulation

Simulation parameters value Cargo width w 1.3m cargo height h 1.4m cargo depth d 1.1m The total number of rows a 6 total number of columns b 10 Layer c 4 The speed v_x of the access trolley in the x direction on the horizontal plane 2m/s The speed v_y of the access trolley in the y direction on the horizontal plane 2m/s The vertical speed of the access cart is v_z 1m/s The loading and unloading time of the access trolley is t_p 5s Picking aisle width l 4.3m Shelf moving speed v_r 4m/min The time for the rack to move to open the picking lane is t_r 64.5s

Table 1

In a certain slot allocation optimization task in this warehouse, 200 items need to be stored in the warehouse. The storage system has a total of 240 slots, and each slot can only store one item. The weights are known, and the similarity coefficient between the two items has also been calculated by formula (9) based on historical orders. In order to make the warehouse run efficiently and stably, the present invention adopts the improved invasive weed algorithm to simulate and analyze the optimization of the warehouse's cargo space allocation.

The parameter settings of IIWO algorithm are shown in Table 2. Table 2 shows the parameters of the IIWO algorithm.

Table 2

Improved Picking Efficiency Simulation Results:

Figure 6 is the iterative curve of the improved invasive weed algorithm. The ideal point of objective function 1 is 582.43, which is 0.7% compared to the initial 586.5.

Improve shelf center of gravity simulation results:

Figure 7 shows the iterative curve of the improved invasive weed algorithm. The ideal point of objective function 2 is 2.59, which is 14.5% compared to the initial 3.03.

Improve the simulation results of item correlation in the same picking lane:

Figure 8 shows the iterative curve of the improved invasive weed algorithm. The ideal point of objective function 3 is 0.11, compared with the initial 0.133, the optimization effect reaches 17.3%.

Multi-objective evaluation function simulation results:

Considering the three objective functions comprehensively, the company focuses on improving shelf stability and item correlation in the same picking lane. Since it takes a long time to open the picking lane due to shelf movement, the improvement of picking efficiency largely depends on the same picking lane. Item correlation, so the importance of improving the picking efficiency is the least, that is, (λ ₁ , λ ₂ , λ ₃ ) takes the value of (0.2, 0.4, 0.4), which is substituted into the multi-objective optimization evaluation function for simulation solution.

Figure 9 shows the iterative curve of the improved invasive weed algorithm. The optimal value of the multi-objective optimization evaluation function is 0.24, which is 87.4% compared to the initial 1.91.

The simulation results of the above four objective functions can be verified, and it can be concluded that the method proposed in the present invention can better solve the optimization problem of cargo space allocation.

Claims (4)

1. A goods position distribution method of a mobile goods shelf warehousing system facing a refrigeration house is characterized by comprising the following steps:

step 1, establishing a multi-target goods space allocation optimization model for the purpose of improving the picking efficiency, improving the shelf stability and improving the relevance of the items of the same picking roadway;

calculating an objective function f for improving the sorting efficiency₁，

The movement of the storage trolley from the I/O position to the cargo space (x, y, z) is calculated, assuming that the pallet needs to be moved, the horizontal running time is:

wherein y represents the coordinate of the cargo space (x, y, z) in the y-direction, w represents the cargo compartment width, v_yRepresenting the speed of the access trolley in the y direction; t is t_rIndicating the time of the shelf moving to open the picking roadway, k indicating the goods to be picked at the goods position (x, y, z) on the picking roadway k, d indicating the depth of the goods grid, l indicating the width of the picking roadway, v_xRepresenting the speed of the access trolley in the x direction;

and then calculating the running time of the access trolley moving from the I/O position to the cargo space (x, y, z) in the vertical direction as follows:

wherein z represents the coordinate of the cargo space (x, y, z) along the z direction, h is the cargo grid height, v_zThe speed of the access trolley in the vertical direction is shown;

continuing to calculate the return of the access cart from the cargo space (x, y, z) to the I/O location, the return time on the horizontal plane is:

the run time in the vertical direction on the return of the recalculation is:

further, since the movement of the storage trolley in the horizontal plane and the movement in the vertical direction are simultaneous, the time taken to pick the goods is the maximum value of the running times in the two directions, and the time t taken to pick the goods at the goods location (x, y, z) is calculated_xyzComprises the following steps:

wherein, t_pThe time for loading and unloading goods for the storage trolley;

the time taken to pick the item at the cargo space (x, y, z) is then:

t_xyz·p_xyz (6)

wherein p is_xyzRepresenting a picking frequency of items deposited at the cargo space (x, y, z);

further, the goal of improving the picking efficiency is to minimize the total picking time of the goods, and the expression of the objective function is as follows:

wherein f is₁Representing the total picking time of the goods;

calculating an objective function f for improving shelf stability₂，

Wherein f is₂Indicates the height of the center of gravity of the goods shelf as a whole, m_xyzRepresenting the weight of an item stored at the cargo space (x, y, z);

calculating and improving item correlation objective function f of same picking roadway₃，

The calculation formula of the similarity coefficient proposed by Russel and Rao is as follows:

wherein A represents the order quantity simultaneously containing item i and item j; b represents the order quantity containing only item i; c represents the order quantity containing only item j; d represents the order quantity that neither item i nor item j contains;

the method has the advantages that items with strong correlation are stored on the shelf of the same sorting roadway, so that the times of opening the sorting roadway can be effectively reduced, furthermore, the reciprocal of the sum of the similarity coefficients of the items of the same sorting roadway converted from an objective function is as small as possible, and the expression of the objective function is as follows:

wherein f is₃Expressing the reciprocal of the sum of item similarity coefficients of the same picking lane, K is the number of the picking lane, i and j represent item numbers, g represents the number of the items, r_ikItem i is stored on the rack of the sorting lane k when the value 1 indicates that item i is not stored, otherwise r_ik＝0；r_jkDepositing item j as 1On the shelf of lane k, otherwise r_jk＝0；

Further constraints of the model are as follows:

wherein x is more than or equal to 1 and less than or equal to a represents the limitation of the number of rows of the goods shelves; y is more than or equal to 1 and less than or equal to b, which represents the limitation of the number of rows of the goods shelves; z is more than or equal to 1 and less than or equal to c represents the limitation of the number of the shelf layers;

indicating that each item can only occupy one cargo space;

step 2, establishing an evaluation function

Processing the three objective functions by an ideal point method to construct an evaluation function, and firstly, finding each objective function f_iOptimum value of f_i ^*And taking it as an ideal point;

based on the distance between each objective function value and the ideal point, an evaluation function of each objective function is constructed:

F_i＝(f_i-f_i ^*)² (12)

wherein, F_iAn evaluation function value representing an ith objective function; f. of_iA function value representing an ith objective function; f. of_i ^*An optimal value representing the ith objective function;

further, on the basis of the above formula, a weight coefficient lambda is introduced_iAnd if the sum is 1, converting the multi-objective optimization function into an evaluation function:

wherein f is a multi-objective optimization evaluation function;

step 3, coding design: numbering the items and the goods positions simultaneously, adopting a natural number arrangement coding mode, wherein the coding length depends on the number of the items, when the number of the items is N, one code consists of N unrepeated natural numbers, and each natural number corresponds to a goods position number;

step 4, generating an initial population: initializing algorithm related parameters: initial number of weeds N₀Maximum number of weed population N_maxMaximum number of iterations iter_maxMaximum number of seeds S that can be produced per weed_maxAnd minimum value S_minNonlinear modulation index n, initial value of standard deviation σ of weeds in spatial diffusion_initSum standard deviation final value σ_final；

Step 5, recording an objective function value of each weed, wherein the objective function value comprises f₁，f₂，f₃And f, then recording the optimal weed individuals and the optimal solution;

step 6, a weed propagation stage: calculating the number of seeds generated by each weed according to the objective function value of each weed by using the following formula;

wherein f represents the objective function value of the current weed, f_maxAnd f_minRepresenting the maximum and minimum values of the objective function representing the weeds in the current population respectively,

represents rounding down;

step 7, a spatial diffusion stage: weed seeds are scattered around the parent weeds according to normal distribution with the mean value of 0 and the standard deviation of sigma, and the sigma can also be distributed from the initial value sigma as the iteration number is increased_initDecrease to the final value σ_finalSpecifically, the standard deviation is calculated as follows when a certain generation is reached:

wherein, σ represents a standard deviation value corresponding to the current algebra; iter_maxRepresenting the maximum number of iterations; iter represents the current algebra; sigma_initRepresents an initial value of standard deviation; sigma_finalRepresents the final value of the standard deviation;

step 8, competition survival stage: after the parent individuals and the offspring individuals in the population are combined into a new population, in order to improve the local search capability of the algorithm, an evolution reversal operator of the genetic algorithm is introduced, two positions are randomly selected, the numbers of the two positions are interchanged, if the objective function value is reduced, the individuals are accepted, otherwise, the evolution reversal is invalid, then the new population is sorted according to the objective function value, excellent individuals with small objective function values are left, the disadvantaged individuals with large objective function values are eliminated, and the number of the individuals cannot exceed the maximum number N of the population_max；

Step 9, completing the iteration once, and judging whether the maximum iteration number iter is reached_maxAnd if so, outputting the optimal weed individuals, otherwise, returning to the step 6.

2. The method for allocating the cargo space of the mobile goods shelf warehouse system facing the cold storage according to claim 1, wherein the process of the step 4 is as follows:

step 4.1, firstly, the time t required for picking the goods at the goods position (x, y, z) is calculated_xyzCreating a picking time matrix from which the time t required to pick a good at any location can be known_xyz；

4.2, using a greedy algorithm to generate 40% of initial weed populations, randomly generating the rest 60% of the weed populations, and using two greedy strategies in the initial weeds generated by the greedy algorithm, wherein the first method is to preferentially select a goods space with short picking time; the second is to preferentially select a cargo space with a small number of layers.

3. The method for allocating the cargo space of the mobile goods shelf warehouse system facing the cold storage according to claim 2, wherein the process of the step 4.2 is as follows:

step 4.2.1, preferentially selecting the goods location method with short picking time: taking out the serial number of each goods position and the corresponding picking time to create a matrix, wherein the first column is the goods position serial number, the second column is the corresponding picking time, then the goods position serial number and the picking time of each row are still corresponding, then the goods position serial number and the picking time are arranged according to the ascending order of the second column, and at the moment, the generated partial population codes correspond to the first column one by one;

step 4.2.2, preferentially selecting the method for the goods location with the small layer number: taking out each goods position number and the number of layers where the goods position number is located to create a matrix, wherein the first column is the goods position number, the second column is the corresponding layer, then the goods position number and the number of layers of each row are still corresponding, then the goods position number and the number of layers are arranged according to the ascending order of the second column, and at the moment, the generated partial population codes correspond to the first column one by one.

4. The goods location distribution method of the mobile goods shelf storage system facing the cold storage according to any one of claims 1 to 3, wherein the process of the step 7 is as follows:

step 7.1, adopting a space diffusion operator based on sliding insertion, firstly randomly generating two random positions j and k, then taking the goods space numbers from the position j to the position k as a queue, taking the number of the position j as the head of the queue, taking the number of the position k as the tail of the queue, then sequentially deleting the numbers from the tail of the queue, and inserting the numbers into the head of the queue until d times of insertion are executed;

step 7.2, the dispersion process of the seeds obeys normal distribution with the mean value of 0 and the standard deviation of sigma, the size of sigma determines the search range of the seeds, and the execution times d of the sliding insertion obeys N (0, sigma)²) The absolute value of the random number alpha is taken and then is rounded up

As the number of execution times d of the slide insertion.

Images