A steel coil storage location allocation method based on multi-objective optimization

Abstract

The application provides a steel coil storage location allocation method based on multi-objective optimization. It belongs to the field of intelligent storage. The steps are as follows: collecting the storage location information of the steel coil warehouse, the crane operation data and the steel coil scheduling history record, establishing a model aiming at improving the steel coil turnover rate; then, the type information of the in-stock steel coil inventory is counted, a model for improving the uniformity of the steel coil storage location allocation is established; the steel coil turnover rate and the uniformity of the steel coil are comprehensively considered, and the model constraint conditions are established according to the actual situation of the warehouse, a multi-objective optimization model is constructed, and the multi-objective model is weighted, so that the multi-objective problem is converted into a single-objective problem; the model is calculated through the hunting algorithm, and the storage location for the steel coil allocation is obtained. The steel coil storage location allocation method based on multi-objective optimization improves the scheduling speed of the high turnover rate steel coil in the warehouse, and also concentrates the same type of steel coil, improves the steel coil scheduling efficiency, and reduces the warehouse management cost.

Description

Technical Field

[0001] This invention belongs to the field of intelligent warehousing, and more specifically, it is a method for allocating storage locations in steel coil warehouses for steel enterprises. Background Technology

[0002] The warehousing and logistics scheduling of finished steel coils is a crucial link in the steel industry's production and sales. As the national strategy for manufacturing transformation and upgrading drives China's manufacturing sector to shift from low-end, labor-intensive to high-end, intelligent, green, and sustainable development, the traditional reliance on manual warehouse allocation carries risks of human error and unreasonable allocation. Furthermore, the need for workers to enter the warehouse's safety perimeter increases safety risks during production.

[0003] Storage location allocation is crucial in the field of warehouse storage, directly impacting warehouse operational efficiency, goods retrieval speed, and space utilization. Effective storage location allocation improves overall warehouse operational efficiency and service levels, positively influencing supply chain management and customer satisfaction. Storage location allocation optimizes internal warehouse space utilization. By rationally planning storage locations and placing goods in the most suitable positions, warehouse storage space can be maximized, reducing inventory buildup and wasted space, thereby increasing storage density and capacity. Rational storage location allocation optimizes the efficiency of goods retrieval. By placing related products in adjacent locations, the distance and time goods travel within the warehouse are reduced, increasing retrieval speed and operational efficiency. This is significant for quickly responding to customer orders, reducing waiting time, and improving order processing efficiency. In conclusion, the importance of storage location allocation in the field of warehouse storage is self-evident. Through rational planning and optimization of storage location allocation, warehouse operational efficiency can be effectively improved, costs reduced, and service levels enhanced, thereby creating greater value and competitive advantage for enterprises.

[0004] A search revealed Chinese patent application number 201711154499.1, published on April 20, 2018, which discloses a method for allocating storage locations in a fishbone-shaped automated warehouse. This method includes setting warehouse storage location and cargo parameters, initializing particle positions and velocities, generating an initial particle swarm within an interval, evaluating the fitness of each particle, updating the individual optimal values ​​of each particle and the global optimal value of the particle swarm, iteratively updating particle velocities and positions, determining if the updated particle positions are valid (if valid, proceed); otherwise, perform a mutation operation on the particles, then return to compare the newly generated particle state with the previous state, selecting the particle with the higher fitness as the initial particle for the next generation update, and then determining if the algorithm has reached its termination iteration count (if valid, proceed); otherwise, proceed to the next step to output the optimal storage location allocation scheme. However, the particle swarm optimization (PSO) algorithm used in this method may converge prematurely during the iteration process, causing the algorithm to stop searching and fail to find the global optimum. This may be due to the particle velocity decreasing too early or local optima in the search space hindering further exploration by the particles.

[0005] Chinese Patent Application No. 202211212469.2, published on December 20, 2020, discloses a method and system for allocating storage locations in an automated warehouse. The method includes: establishing an inbound / outbound efficiency function and its constraints for goods to be allocated in the automated warehouse; establishing a center-of-gravity offset function for shelves after storing goods and its constraints; establishing a center-of-gravity fit function for the same type of goods stored on the shelves and its constraints; calculating the actual values ​​of inbound / outbound efficiency, center-of-gravity offset, and center-of-gravity fit; using a genetic algorithm to solve for the optimal values ​​of inbound / outbound efficiency, center-of-gravity offset, and center-of-gravity fit; determining the evaluation function for each function based on the distance between each actual value and the optimal value; weighting the evaluation functions to obtain a fitness function; and using a genetic algorithm to solve for the fitness function to obtain the allocated storage location of the goods. However, this method uses a traditional genetic algorithm to solve the model. Genetic algorithms typically require a large population to search the solution space, thus consuming significant computational resources. The computational cost of the algorithm becomes even more significant when dealing with large-scale or high-dimensional problems. Summary of the Invention

[0006] To address the problems of low warehouse space utilization and chaotic steel coil storage caused by traditional warehouse location allocation, this invention provides a warehouse location allocation method. By collecting warehouse information, crane operation data, and historical data on steel coil scheduling, a model is established to improve the consistency of steel coil warehouse location allocation and to reduce the scheduling time of high-turnover steel coils. The model is weighted by expert scoring, and finally, an improved hunting algorithm is used to solve the model to obtain the optimal warehouse location allocation scheme.

[0007] To solve the above problems, the present invention adopts the following technical solution.

[0008] A method for allocating storage locations includes the following steps:

[0009] S1. Collect warehouse location information, crane operation data, and historical records of steel coil scheduling in the steel coil warehouse. Analyze the historical records of steel coil scheduling to obtain the turnover rate of different types of steel coils and establish a model aimed at reducing the scheduling time of high-turnover steel coils.

[0010] S2. Collect information on the types of steel coils stored in the warehouse and establish a model to improve the consistency of steel coil storage location allocation;

[0011] S3. Establish model constraints based on the actual warehouse location allocation rules, build a multi-objective steel coil location optimization model, and introduce an expert scoring method to weight the model, transforming the multi-objective problem into a single-objective problem and constructing a single-objective optimization model.

[0012] S4. Using the vacant storage locations in the warehouse as the solution space, the optimization model is solved using the hunting algorithm to obtain the optimal storage location allocation scheme;

[0013] Furthermore, the model for reducing the scheduling time of high-turnover steel coils described in S1 is as follows:

[0014]

[0015] Where (a1, a2) represent the width coordinate range of the storage location, (b1, b2) represent the length coordinate range of the storage location, and (c1, c2) represent the height coordinate range of the storage location; X, Y, and Z represent the storage location coordinates of the steel coil; V x V y V z This represents the speed at which the crane travels along the X, Y, and Z axes; T represents the time it takes for the steel coil to reach the parking space; and P represents the turnover rate of the steel coil.

[0016] Furthermore, the model for improving the uniformity of steel coil warehouse location allocation described in S2 includes the following steps:

[0017] (3.1) Assume that the warehouse stores k families of steel coils, where the i-th family (i = 1, 2, ..., k) contains N... i There are i-th steel coils, where the spatial coordinate vector of the i-th steel coil is [X...]. i ,Y i Z i ], then N of the same family i The coordinate vector of a steel coil can be represented as:

[0018] (3.2) To measure the dispersion within the warehouse, the mean coordinate vector m of each family of steel coils can be calculated. i:

[0019] (3.3) To measure the intra-class dispersion of steel coils in each family, this value can be defined as the sum of the distances d from each steel coil in the family to the class center. i : in

[0020] (3.4) Based on the intra-class dispersion obtained in (3.3), the total intra-class dispersion of the k-family steel coils can be derived as follows:

[0021] (3.5) Let M be the measure of inter-class dispersion, then the coordinates M of the mean vector of the centers of all k-family steel coils are: Where m i (i = 1, 2, ..., k);

[0022] (3.6) Measure the center of the steel coil to M j The dispersion value is defined as the sum of the distances from the center of each class to M, i.e.

[0023] (3.7) Based on a comprehensive consideration of inter-class and intra-class dispersion, steel coils of the same family should be placed together as much as possible to minimize the intra-class dispersion of steel coils of the same family; at the same time, it is desirable that steel coils of different families are evenly distributed in the steel coil warehouse and that the sum of their distances from the truck parking spaces is minimized, i.e., the inter-class dispersion is maximized. Therefore, based on (3.4) and (3.6), a model for improving the uniformity of steel coil warehouse location allocation is proposed:

[0024] (3.8) Based on S2 and the two warehouse location optimization models described in step (3.7), an overall multi-objective steel coil warehouse location optimization model can be established:

[0025] Furthermore, the single-objective optimization model described in S3 includes the following steps: taking the steel coil turnover rate and the steel coil identity as features, then performing data preprocessing on the features, weighting the data, and summing the results to obtain the final score. The final score is used as the weight values ​​ω1 and ω2, with ω1+ω2=1. The multi-objective model is then linearly weighted to obtain the weighted model: MinF=ω1F1+ω2F2;

[0026] Furthermore, the constraints described in S3 are as follows:

[0027] (a1-P id )*(P iw -b2)*P ip ≤0;

[0028] (Piz -a3)*(P iz -a4)=0;

[0029]

[0030]

[0031] Where: P ib P ic P id P iw P ix P iy P iz , This represents the specifications of the steel coil, including width, height, thickness, net weight, row, column, layer, and coiling condition. i (i = 1, 2, ..., 12) are constants.

[0032] Furthermore, the hunting algorithm steps described in S4 are as follows:

[0033] (6.1) Initialize the population size, fitness, and number of iterations, and set the maximum number of iterations;

[0034] (6.2) Calculate the fitness of the hunting population according to the formula, find the current optimal fitness and retain it;

[0035] (6.3) Calculate the new location of the hunting population by observing the migration and aggression behaviors of the hunting population, and update the hunting location and fitness by comparing the fitness of the old and new hunting locations;

[0036] (6.4) If the current iteration number is less than the maximum iteration number, then jump to (6.3); otherwise, output the optimal solution.

[0037] Furthermore, the attack behavior formula described in (6.3) is: p s (t)=d s (t)*x*y*z+p best (t), where x=r*cosθ, y=r*sinθ, z=r*θ, r=u*e θv In the above formula, r is the radius of the hunting spiral flight, u and v are constants, θ is a random value in [0, 2π], e is the base of the natural number, and p best (t) represents the current globally optimal hunting position. Traditional hunting algorithms perform a global search only after all population iterations are complete, resulting in discontinuous global updates and reducing the algorithm's optimization speed. The optimal hunting position formula introduced by the hunting algorithm is:

[0038]

[0039]

[0040] Furthermore, the migration behavior formula described in (6.3) is: c s (t)=A(t)*p s (t), where c t p represents the new position in a collision-free hunting scenario, where t is the current iteration number; s A(t) describes the current hunting position; A(t) describes the changes in the hunting movement within a certain area, and the formula for A(t) is: The hunting algorithm using cosine-based motion behavior allows for a larger movement distance in the early stages of the hunt, enabling a better global search, where T is the maximum number of iterations.

[0041] Furthermore, the fitness formula described in (6.2) is: m s (t)=B*[z best (t)-p s [t], where B = 2*A 2 *λ and Adding an adaptive weight factor λ to the fitness formula reduces the blindness in calculating hunting positions and better allocates the algorithm's search capabilities in the early and late stages.

[0042] Furthermore, the position update formula described in (6.3) is: d s (t)=|c s (t)+m s (t)|.

[0043] 3. Beneficial effects

[0044] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0045] This invention provides a multi-objective optimization-based method for steel coil storage location allocation, designed with steel coil turnover rate and uniformity in mind. On one hand, to improve inbound and outbound efficiency and shorten picking time, high-turnover goods are placed near the outbound exit, significantly improving hoisting efficiency and reducing hoisting costs. On the other hand, steel coils are allocated storage locations according to type or specification, with similar types or specifications placed in the same storage area. This facilitates warehouse management personnel in managing and inventorying incoming and outbound steel coils, reducing workload.

[0046] The method, by using a hunting algorithm, has the following beneficial effects:

[0047] It has strong global search capabilities and fast convergence, and can effectively find the optimal or near-optimal solution to optimization problems, thereby improving the quality of problem solving.

[0048] It has a fast convergence speed and good adaptability, and can find a satisfactory solution in a short time, thereby accelerating the problem-solving process and saving time and resource costs.

[0049] It has strong global search capabilities and can effectively handle complex, multi-peak optimization problems, providing an effective solution for solving complex problems encountered in practice. Attached Figure Description

[0050] Figure 1 This is a flowchart of the storage location allocation method for a steel coil warehouse according to an embodiment of the present invention;

[0051] Figure 2 This is a flowchart of the hunting algorithm in an embodiment of the present invention. Detailed Implementation

[0052] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0053] This invention proposes a steel coil warehouse location allocation method based on multi-objective optimization. To facilitate the establishment and solution of the mathematical model in this invention, the following assumptions are made about the mathematical model based on the actual operation of the various entities in the steel coil warehouse:

[0054] All steel coils entering and leaving the warehouse meet the warehouse location specifications, and there will be no situation where warehouse locations cannot be allocated.

[0055] The crane moves at a constant speed along each spatial axis;

[0056] The specifications of the storage locations and the distance between them are the same;

[0057] The coordinates of the parking space are assumed to be a single coordinate point;

[0058] Example 1

[0059] A method for steel coil storage location allocation based on multi-objective optimization includes the following steps, detailed below. Figure 1 As shown.

[0060] Step 1: Based on historical records of steel coil scheduling, establish a model aimed at reducing scheduling time for high-turnover steel coils. The calculation method is as follows:

[0061] 11) Analyze and process the structural information of the steel coil warehouse and the historical data of steel coil scheduling to obtain the basic information of the steel coil warehouse and the information on steel coil entry and exit. The basic information of the warehouse includes the length, width, and height coordinate range of the steel coil storage location, and the operating speed of the crane on the X, Y, and Z axes during operation. The information on steel coil entry and exit includes the type, quality, and frequency of entry and exit of the steel coil.

[0062] 12) Based on historical orders and inventory data of steel coil scheduling, obtain the turnover rate of different types of steel coils, and establish a mathematical model according to preset principles. The steps are as follows:

[0063] Based on the turnover rate of steel coils, crane operating speed, and warehouse location coordinate range, a model can be established to reduce the scheduling time of high-turnover steel coils:

[0064]

[0065] Specifically, (a1, a2) represent the width coordinate range of the storage location, (b1, b2) represent the length coordinate range of the storage location, and (c1, c2) represent the height coordinate range of the storage location; X, Y, and Z represent the storage location coordinates of the steel coil; V x V y V z This represents the speed at which the crane travels along the X, Y, and Z axes; T represents the time it takes for the steel coil to reach the parking space; and P represents the turnover rate of the steel coil.

[0066] Step 2: Collect statistics on the types of steel coils stored in the warehouse, perform cluster analysis based on correlation, and establish a model to improve the uniformity of steel coil storage location allocation. The specific method is as follows:

[0067] 21) The warehouse now stores k families of steel coils, where the i-th family (i = 1, 2, ..., k) contains N... i There are i-th steel coils, where the spatial coordinate vector of the i-th steel coil is [X...]. i ,Y i Z i ], then N of the same family i The coordinate vector of a steel coil can be represented as:

[0068] 22) To measure the dispersion within the warehouse, the mean coordinate vector m of each family of steel coils can be calculated. i :

[0069] 23) To measure the intra-class dispersion of steel coils in each family, this value can be defined as the sum of the distances d from each steel coil in the family to the class center. i : in

[0070] 24) Based on the intra-class dispersion obtained in 23), the total intra-class dispersion of family k steel coils can be calculated as follows:

[0071]

[0072] 25) Let M be the measure of inter-class dispersion, then the coordinates M of the mean vector of the centers of all k-family steel coils are: Where m i(i = 1, 2, ..., k);

[0073] 26) Measure the center of the steel coil to M j The dispersion value is defined as the sum of the distances from the center of each class to M, i.e.

[0074] 27) Based on a comprehensive consideration of inter-class and intra-class dispersion, steel coils of the same family should be grouped together as much as possible to minimize intra-class dispersion. Simultaneously, it is desirable to evenly distribute non-family steel coils throughout the warehouse and minimize the sum of their distances from truck parking spaces, i.e., maximize inter-class dispersion. Therefore, based on 24) and 26), a model for improving the uniformity of steel coil warehouse location allocation is proposed:

[0075] 28) Based on the two objective models completed in 12) and 27), an overall multi-objective steel coil storage location optimization model can be established:

[0076] 29) Using expert scoring (as shown in Table 1), with steel coil turnover rate and steel coil consistency as features, the features are preprocessed, weighted, and summed to obtain the final score. The final score is used as the weight values ​​ω1 and ω2, and ω1+ω2=1. The multi-objective model is linearly weighted to obtain the weighted model: MinF=0.7F1+0.3F2;

[0077] Table 1 Expert Scoring Sheet

[0078]

[0079] Step 3: Further constrain the steel coil allocation model based on the actual warehouse location allocation rules (as shown in Table 2):

[0080] (0.5-P id )*(P iw -10)*P ip ≤0;

[0081] (P iz -1)*(P iz -2)=0

[0082]

[0083]

[0084] Where: P ib P ic P id P iw P ix Piy P iz , This represents the specifications of the steel coil, including width, height, thickness, net weight, row, column, layer, and coiling condition. i (i = 1, 2, ..., 12) are constants.

[0085] Table 2 Storage Location Allocation Rules

[0086]

[0087]

[0088] Step 4: Solve the model obtained in Step 3 using the hunting algorithm to obtain the optimal storage location allocation scheme.

[0089] like Figure 2 As shown, the specific algorithm steps are as follows:

[0090] 41) Initialize the population size, fitness, and number of iterations, and set the maximum number of iterations;

[0091] 42) Calculate the fitness of the hunting population according to the formula, find the current optimal fitness and retain it;

[0092] 43) Calculate the new location of the hunting population by observing its migration and aggression behaviors, and update the hunting location and fitness by comparing the fitness of the old and new hunting locations;

[0093] 44) If the current iteration number is less than the maximum iteration number, jump to 43); otherwise, output the optimal solution.

[0094] Step 5: Based on the actual storage location allocation and the model storage location allocation, Table 3.5 is summarized.

[0095] Table 3.5 Storage Location Allocation Statistics

[0096]

[0097] By comparing the results of the manual allocation mode and the model allocation mode, it can be seen that the warehouse space utilization rate of the storage location allocated by the model is higher than that of the manual allocation mode. It not only improves the placement rate of steel coils on the second floor, but also strengthens the clustering of steel coils of the same specification.

Claims

1. A method for allocating steel coil storage locations based on multi-objective optimization, characterized in that, Includes the following steps: S1. Collect warehouse location information, crane operation data, and historical records of steel coil scheduling in the steel coil warehouse. Analyze the historical records of steel coil scheduling to obtain the turnover rate of different types of steel coils and establish a model aimed at reducing the scheduling time of high-turnover steel coils. S2. Collect information on the types of steel coils stored in the warehouse and establish a model to improve the consistency of steel coil storage location allocation; S3. Establish model constraints based on the actual warehouse location allocation rules, build a multi-objective steel coil location optimization model, and introduce an expert scoring method to weight the model, transforming the multi-objective problem into a single-objective problem and constructing a single-objective optimization model. S4. Using the vacant storage locations in the warehouse as the solution space, the optimization model is solved using the hunting algorithm to obtain the optimal storage location allocation scheme; The single-objective optimization model described in S3 includes the following steps: taking the steel coil turnover rate and the steel coil identity as features, then performing data preprocessing on the features, weighting the data, and summing the results to obtain the final score. The final score is used as the weight values ​​ω1 and ω2, and ω1+ω2=1. The multi-objective model is then linearly weighted to obtain the weighted model: MinF=ω1F1+ω2F2. The constraints described in S3 are: (a1-P id )*(P iw -b2)*P ip ≤0; (P iz -a3)*(P iz -a4)=0; (j,k are the lower volumes of i); (j,k are the lower volumes of i); Where: P ib P ic P id P iw P ix P iy P iz , This represents the specifications of the steel coil, including width, height, thickness, net weight, row, column, layer, and coiling condition. i (i = 1, 2, ..., 12) are constants; The hunting algorithm steps described in S4 are as follows: Step (6.1) initializes the population size N and fitness m. s (t) and the number of iterations t, and set the maximum number of iterations T; Step (6.2) calculates the fitness m of the hunting algorithm population according to the formula. s (t), find the current optimal fitness and retain it; Step (6.3) calculates the new location c of the hunting population based on the hunters' migration and aggression behaviors. t By comparing the hunters' fitness in new and old locations, the hunting location p s (t) and fitness m s (t) is used to update; In step (6.4), if the current iteration number is less than the maximum iteration number, then jump to step (6.3); otherwise, output the optimal solution.

2. The method for allocating steel coil storage locations based on multi-objective optimization according to claim 1, characterized in that: The model for reducing the scheduling time of high-turnover steel coils described in S1 is as follows: Where (a1, a2) represent the width coordinate range of the storage location, (b1, b2) represent the length coordinate range of the storage location, and (c1, c2) represent the height coordinate range of the storage location; X, Y, and Z represent the storage location coordinates of the steel coil; V x V y V z This represents the speed at which the crane travels along the X, Y, and Z axes; T represents the time it takes for the steel coil to reach the parking space; and P represents the turnover rate of the steel coil.

3. The method for allocating steel coil storage locations based on multi-objective optimization according to claim 1, characterized in that: In S2, the model for improving the consistency of steel coil warehouse location allocation includes the following steps: Suppose the warehouse stores k families of steel coils, where the i-th family (i = 1, 2, ..., k) contains N... i There are i-th steel coils, where the spatial coordinate vector of the i-th steel coil is [X...]. i ,Y i Z i ], then N of the same family i The coordinate vector of a steel coil is represented as: To measure the dispersion within the warehouse, the mean coordinate vector m of each group of steel coils is calculated. i : To measure the intra-class dispersion of steel coils in each family, the intra-class dispersion of a steel coil is defined as the sum of the distances d from each steel coil in that family to the class center. i : in Based on the obtained intra-class dispersion, the total intra-class dispersion of family k steel coils can be calculated as follows: Let M be the measure of inter-class dispersion, then the coordinates M of the mean vector of the centers of all k-family steel coils are: Where m i (i = 1, 2, ..., k); Measuring the center of steel coils to M j The dispersion value is defined as the sum of the distances from the center of each class to M, i.e. Identify a model to improve the consistency of steel coil warehouse location allocation: Establish a holistic, multi-objective steel coil storage location optimization model: St X, Y, Z are integers.

4. The method for allocating steel coil storage locations based on multi-objective optimization according to claim 1, characterized in that: In step (6.3), the attack behavior formula is: p s (t)=d s (t)*x*y*z+p best (t), where x=r*cosθ, y=r*sinθ, z=r*θ, In the above formula, r is the radius of the hunting spiral flight, u and v are constants, θ is a random value in [0, 2π], e is the base of a natural number, and p best (t) represents the current position of the globally optimal hunter. The formula for the optimal hunting position introduced by the hunting algorithm is:

5. The method for allocating steel coil storage locations based on multi-objective optimization according to claim 1, characterized in that: In step (6.3), the formula for hunter migration behavior is: c s (t)=A(t)*p s (t), where c t p represents the new position of the hunter group in a collision-free situation, where t is the current iteration number; s A(t) describes the hunter's current position; A(t) describes the changes in the hunter's movement within a certain area, and the formula for A(t) is: The hunting algorithm using cosine-based motion behavior allows for a larger movement distance in the early stages of the hunt, enabling a better global search, where T is the maximum number of iterations.

6. The method for allocating steel coil storage locations based on multi-objective optimization according to claim 1, characterized in that: In step (6.2), the fitness formula is: m s (t)=B*[z best (t)-p s [t], where B = 2*A 2 *λ and 7. The method for allocating steel coil storage locations based on multi-objective optimization according to claim 1, characterized in that: The position update formula mentioned in step (6.3) is: d s (t)=|c s (t)+m s (t)|.

Images