Disclosure of Invention
Aiming at the problems of low operation efficiency, high idle load rate and the like of a stacker and an AGV caused by severe requirements on medicine batches of a medicine warehouse, dispersed product distribution and the like, factors such as efficiency matching of an operation system connected in an upper stage and a lower stage are fully considered, the shortest total time of the operation time of an order in and out is taken as a target, and a mixed command sequence operation time model suitable for a combined dispatching system of the stacker and the AGV of the medicine warehouse is established.
Based on an improved immune clone algorithm, completing the optimization solution of a scheduling model to obtain a scheduling scheme;
the practical case of the medicine warehouse is analyzed and compared with other intelligent methods, and the applicability of the method to the medicine logistics warehousing system is verified.
The above aspect and any possible implementation further provides an implementation, where the mathematical model includes:
Min T=max(T1,T2)
the expression shows that the total operation time is determined by one of the two AGVs with smaller operation time, wherein T is the total operation time, and T is the total operation time1And T2AGV operation times No. 1 and No. 2.
Ti=TC+TI+TO
The formula shows that the operation time of each AGV consists of composite cycle operation time, single cycle warehousing operation time and single cycle ex-warehousing operation time, wherein TCTotal time of combined cycle operation, TIAnd TOThe AGV warehousing and ex-warehousing operation time are respectively.
The formula respectively represents the composition of the AGV composite cycle operation time, the AGV single cycle warehousing operation time and the AGV single cycle ex-warehousing operation time. Wherein T isCThe total time of the combined cycle operation; t isLDelivering goods time for the AGV; t isGTravel per unit path time for the AGV; v. ofcThe horizontal running speed of the stacker; v. ofrThe vertical running speed of the stacker; a isCThe horizontal acceleration of the stacker; a isrVertical acceleration of the stacker; hCThe length of the goods grid in the horizontal direction; hrThe length of the goods grid in the vertical direction; t isUThe time for storing and taking the goods by the stacker; a and b are unit path parameters; and c and d are single-cycle task parameters.
The parameters of a, b, c and d take the following values:
Wherein the value range of the warehousing-in/out node number J is (6, 7, 8); the value range of lane number K (1, 2, 3, 4, 5). The value ranges of x and y are shown in constraint 4.
The above-described aspect and any possible implementation further provide an implementation, where the constraint includes the following four constraints:
constraint 1:
constraint 2:
constraint 3:
constraint 4:
the above aspect and any possible implementation manner further provide an implementation manner, and the specific calculation steps of the working time corresponding to the above formula are as follows: :
and a, the AGV receives the goods at the goods loading point and calculates the operation time of the AGV reaching the goods loading point.
Step b, the stacker receives the goods from the AGV, and the stacker is calculated to be put in storage to a goods position (x)1,y1) And storing the working time of the goods.
Step c, calculating the slave (x) of the stacker1,y1) Moving to the delivery location (x) of the composite job order2,y2) And the operation time of inserting and taking the goods.
Step d, calculating the slave (x) of the stacker2,y2) Working time to initial origin (0,0)
And e, the AGV receives the goods from the stacker and runs to a goods placing point to place the goods.
And f, completing the compound circulation order by circulating the above operations.
Step g, calculating the remaining order-placing cycle order operation time
The above aspects and any possible implementations further provide an implementation, and the improved immune cloning algorithm design includes:
this formula represents the fitness function between the antibody and antigen, m is a solution scheme, TmFor the objective function value, G is the penalty weight for each infeasible solution, typically taking a large positive number. MmIs a corresponding infeasible scheme.
The formula represents an affinity function of the antibody and the antibody, and the affinity between the antibodies is judged by adopting a Euclidean distance measuring and calculating method. The larger the value of D, the lower the similarity between the two; when D is 0, both are the same solution.
The above-described aspects and any possible implementation further provide an implementation, and the algorithm solving step includes the following steps:
step a, antigen recognition. And taking the given objective function and constraint conditions as the antigen of the problem to be solved.
And b, initializing the population. Giving various parameters and obtaining 20 pieces of outbound orders and inbound orders.
And c, classifying the antibodies. And screening out the outbound and inbound orders with the consistent lane number K. And dividing the screened orders into two columns of matrixes in random order. The first column is a warehousing order and the second column is an ex-warehouse order. When multiple warehouse-out orders cannot be matched, the first-column warehouse-in order is X'1,Y′1...X′R-2M,Y′2R-2MThe second column is 0, 0; on the contrary, when the multiple outbound orders cannot be matched, the first column is 0,0, and the second column is X'1,Y′1...X′R-2M,Y′2R-2M。
And d, inoculating the vaccine. And (4) inoculating the established vaccine according to the experience of a principle and the like nearby, and optimizing order sequencing.
And e, calculating the antibody fitness and affinity. The affinity of antibody fitness was calculated by the formulas (3) and (4).
And f, cloning and propagating. Cloning each antibody according to the affinity value according to the size of the affinity, wherein the cloning scale is qiAnd then:
wherein N iscIs the total clone scale of the antibody population, and the higher the antibody fitness, the larger the clone scale.
And g, cloning and mutating. And generating and inputting combinations by respectively and correspondingly exchanging the columns and the layers among each row.
And h, judging the termination condition. A form of hybrid termination is used that defines the number of iterations and cannot be improved over successive iterations.
Detailed Description
For better understanding of the technical solutions of the present invention, the following detailed descriptions of the embodiments of the present invention are provided with reference to the accompanying drawings.
Aiming at the problems of low operation efficiency, high idle load rate and the like of a stacker and an AGV caused by severe requirements on medicine batches of a medicine warehouse, dispersed product distribution and the like, factors such as efficiency matching of an operation system connected in an upper stage and a lower stage are fully considered, the shortest total time of the operation time of an order in and out is taken as a target, and a mixed command sequence operation time model suitable for a combined dispatching system of the stacker and the AGV of the medicine warehouse is established.
On the basis, aiming at the problems of disordered sequencing strategies of single-batch order operation sequences and the like, an improved immune clone algorithm is provided, an improved memory unit is constructed, the defect that only a single optimal antibody is memorized in a genetic algorithm and an optimal solution group cannot be memorized is overcome, the sequencing strategy of the order operation sequences is optimized, and the model and the algorithm are verified and analyzed under the actual case of a certain medicine warehouse through python.
Simulation results show that the model provided by the invention has good applicability to the drug logistics and warehousing system, and meanwhile, the optimization algorithm can reasonably and effectively sequence the operation of grouped orders, thereby effectively improving the warehousing efficiency of the drug warehouse combined operation system.
Fig. 1 is a layout of a logistics system of a drug warehouse. The logistics system consists of two AGVs, five roadway stackers and ten three-dimensional storehouses.
According to the current operation situation of the drug warehouse stacker and AGV combined dispatching system, the following assumptions are provided:
(1) the batch requirements of the drug and goods are strict, so that the goods position information of the warehouse-in and warehouse-out orders is provided by an upper-level system and cannot be randomly modified.
(2) Due to the beat limitation of the preorder film wrapping and packaging work, the AGV has enough time to reach the next task loading point after the operation of entering and exiting the warehouse.
(3) The AGV is running at a constant speed.
(4) The AGV and the stacker have the same no-load and full-load speed.
Based on the assumptions, a mathematical model is established by taking the operation time of the stacker and AGV combined system as a target. The superior system obtains a plurality of ex-warehouse and in-warehouse tasks. The first AGV initial state is at a first order pick-up point and the second AGV initial state is at a second order pick-up point. Each stacker is located at each initial position (0,0) of the roadway.
With this assumption, the present implementation provides the following possible embodiments.
S101, aiming at the shortest total time of the operation time of the outgoing and the warehousing orders, establishing a mixed command sequence operation time model suitable for a stacker and AGV combined scheduling system of a medicine warehouse.
S102, based on an improved immune clone algorithm, completing the optimization solution of a scheduling model to obtain a scheduling scheme;
s103, verifying the applicability of the method to a medicine logistics storage system by analyzing the actual cases of the medicine warehouse and comparing the actual cases with other intelligent methods.
In the embodiment of the present invention, a job scheduling problem is modeled first, and the mathematical model in step S101 includes:
Min T=max(T1,T2)
the expression shows that the total operation time is determined by one of the two AGVs with smaller operation time, wherein T is the total operation time, and T is the total operation time1And T2AGV operation times No. 1 and No. 2.
Ti=TC+TI+TO
The formula shows that the operation time of each AGV consists of composite cycle operation time, single cycle warehousing operation time and single cycle ex-warehousing operation time, wherein TCTotal time of combined cycle operation, TIAnd TOThe AGV warehousing and ex-warehousing operation time are respectively.
The formula respectively represents the composition of the AGV composite cycle operation time, the AGV single cycle warehousing operation time and the AGV single cycle ex-warehousing operation time. Wherein T isCThe total time of the combined cycle operation; t isLDelivering goods time for the AGV; t isGTravel per unit path time for the AGV; v. ofcThe horizontal running speed of the stacker; v. ofrThe vertical running speed of the stacker; a isCThe horizontal acceleration of the stacker; a isrVertical acceleration of the stacker; hcThe length of the goods grid in the horizontal direction; hrThe length of the goods grid in the vertical direction; t isUThe time for storing and taking the goods by the stacker; a and b are unit path parameters; and c and d are single-cycle task parameters.
The parameters of a, b, c and d take the following values:
Wherein the value range of the warehousing-in/out node number J is (6, 7, 8); the value range of lane number K (1, 2, 3, 4, 5). The value ranges of x and y are shown in constraint 4.
In addition, based on the above objective function, the constraint conditions for establishing the model are as follows:
constraint 1:
constraint 2:
constraint 3:
constraint 4:
the specific calculation steps of the working time corresponding to step S101 are as follows:
and a, the AGV receives the goods at the goods loading point and calculates the operation time of the AGV reaching the goods loading point.
Step b, the stacker receives the goods from the AGV, and the stacker is calculated to be put in storage to a goods position (x)1,y1) And storing the working time of the goods.
Step c, calculating the slave (x) of the stacker1,y1) Moving to the delivery location (x) of the composite job order2,y2) And the operation time of inserting and taking the goods.
Step d, calculating the slave (x) of the stacker2,y2) Working time to initial origin (0,0)
And e, the AGV receives the goods from the stacker and runs to a goods placing point to place the goods.
And f, completing the compound circulation order by circulating the above operations.
And g, calculating the remaining work time of the order placing circulation order.
FIG. 2 depicts a coding scheme for an immune cloning algorithm
And when the order warehousing-out node number set is J ═ {6, 7, 8 and 9}, when J ═ 9, the order is a warehousing-out order, and the rest are warehousing orders. The order warehouse-in/warehouse-out target roadway number set is K ═ 1, 2, 3, 4, 5, and this indicates that the order is warehouse-in/warehouse-out operated by the stacker of the roadway.
The line number and the layer number set of the order target goods space are respectively X ═ 1, 2, 3.
In the set of P ═ {0, 1}, 0 indicates that the order goods position is in the stereo library on the left side of the roadway, and 1 indicates the right side.
The improved immune cloning algorithm design in step S102 includes:
this formula represents the fitness function between the antibody and antigen, m is a solution scheme, TmFor the objective function value, G is the penalty weight for each infeasible solution, typically taking a large positive number. MmIs a corresponding infeasible scheme.
The formula represents an affinity function of the antibody and the antibody, and the affinity between the antibodies is judged by adopting a Euclidean distance measuring and calculating method. The larger the value of D, the lower the similarity between the two; when D is 0, both are the same solution.
FIG. 3 is a schematic flow chart of the improved immunoconclone provided by the implementation of the present invention, which specifically comprises the following steps:
step a, antigen recognition. And taking the given objective function and constraint conditions as the antigen of the problem to be solved.
And b, initializing the population. Giving various parameters and obtaining 20 pieces of outbound orders and inbound orders.
And c, classifying the antibodies. Screening laneAnd (5) taking out and putting in the order with the consistent track number K. And dividing the screened orders into two columns of matrixes in random order. The first column is a warehousing order and the second column is an ex-warehouse order. When multiple warehouse-out orders cannot be matched, the first-column warehouse-in order is X'1,Y′1...X′R-2M,Y′2R-2MThe second column is 0, 0; on the contrary, when the multiple outbound orders cannot be matched, the first column is 0,0, and the second column is X'1,Y′1...X′R-2M,Y′2R-2M。
FIG. 4 is a graphical representation of the antigen classification groupings provided by the practice of the present invention.
And d, inoculating the vaccine. And (4) inoculating the established vaccine according to the experience of a principle and the like nearby, and optimizing order sequencing.
And e, calculating the antibody fitness and affinity. The affinity of antibody fitness was calculated by the formulas (3) and (4).
And f, cloning and propagating. Cloning each antibody according to the affinity value according to the size of the affinity, wherein the cloning scale is qiAnd then:
wherein N iscIs the total clone scale of the antibody population, and the higher the antibody fitness, the larger the clone scale.
And g, cloning and mutating. And generating and inputting combinations by respectively and correspondingly exchanging the columns and the layers among each row.
FIG. 5 is a diagram of clonal variation rules provided in the practice of the present invention.
And h, judging the termination condition. A form of hybrid termination is used that defines the number of iterations and cannot be improved over successive iterations.
And carrying out example analysis on the actual out-of-warehouse and in-warehouse orders of the drug warehouse stacker and AGV combined scheduling system, and verifying the applicability and effectiveness of the established model and the optimization method.
The medicine logistics warehouse comprises five roadways, ten stereoscopic warehouses, two AGV and other equipment, the warehouse entry and exit operation is carried out on finished inner and outer bags of medicines, the operation is carried out for two shifts, the working time of each shift is 8 hours, and the whole transfer process is automatically operated.
In the operation process, the stacker and the AGV carry out warehouse-in and warehouse-out operation according to the order operation sequence arranged by the superior system. Therefore, the overall work efficiency of the system is greatly influenced by the different arrangement orders of the warehouse-in and warehouse-out work orders.
The step S103 of analyzing the actual cases of the drug warehouse and comparing the actual cases with other intelligent methods mainly includes:
calculating by an immune clone algorithm, selecting a population with the size of 200, the iteration times of 100, the cross probability of 0.8, the individual vaccination probability of 0.5 and the infeasible punishment weight of 100. The total operation time of the combined system of the stacker crane and the AGV becomes shorter and shorter along with the continuous optimization of the population, and tends to be stable after 75 iterations.
FIG. 6 is a graph of the results of evolutionary iterations of the improved immune cloning algorithm.
The result shows that the scheduling efficiency of the optimized stacker and AGV combined operation system is improved by 4.2%.
FIG. 7 is a comparison of results before and after optimization.
On the basis, all the warehouse-in and warehouse-out orders of 8 hours of work completed at a certain time are simulated, and the orders come from the warehouse-in and warehouse-out order list before the intelligent logistics system of the medicine warehouse is not modified. The operation results are as follows
FIG. 8 is an iterative graph of an immune cloning algorithm for a shift of tasks.
Convergence is completed around 65 iterations, the total job time is reduced from 28004 seconds to 25960 seconds, and the total job time is reduced by about 7.3%.
Meanwhile, by using the genetic algorithm and the comparison of the optimization solution of the model, the convergence of the genetic algorithm and the artificial immune algorithm is finished in about 250 generations and 150 generations respectively, the time for the warehouse-in and warehouse-out execution is 26825 seconds and 26875S respectively, and the optimization rate is about 4.2 percent and 4 percent respectively.
FIG. 9 is an iterative graph of genetic algorithm for a shift of tasks.
FIG. 10 is an iterative graph of the artificial immune algorithm under a certain shift task.
Therefore, the immune clone algorithm has great advantages in convergence speed and optimized warehouse-in and warehouse-out execution time, and has better applicability to the problem of stacker-AGV combined scheduling optimization of the intelligent logistics system of the medicine warehouse.
In conclusion, the stacker and AGV combined scheduling model established by the invention is suitable for the intelligent logistics system of the medicine warehouse, and can greatly improve the operation efficiency of the stacker-AGV combined scheduling system on the premise of not changing the order and goods allocation.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention.