Multi-parking-lot distribution path optimization method for electric refrigerator car

Abstract

The invention discloses a multi-parking-lot distribution path optimization method for a pure electric refrigerator car, and the method comprises the following steps: constructing a multi-parking-lot distribution path optimization function of the pure electric refrigerator car by taking the lowest total distribution cost as a target; distributing the distribution points to different parking lots by using a clustering algorithm, and converting the multi-parking-lot distribution path optimization function of the pure electric refrigerator car into a single-parking-lot distribution path optimization function of the pure electric refrigerator car; and the whale optimization algorithm is utilized to solve the single-vehicle-yard distribution path optimization function of the pure electric refrigerator vehicles. According to the method, firstly, a multi-parking-lot distribution path optimization function of the pure electric refrigerator car is constructed with the minimum total distribution cost as the target, then a k-means algorithm is utilized to convert a multi-parking-lot problem into a single-parking-lot problem, finally, the algorithm is adopted to initialize a population, meanwhile, a position updating strategy is added for position updating, and finally, the total distribution cost of the pure electric refrigerator car is optimized. And the distribution path optimization problem of the pure electric refrigerator car can be efficiently optimized and solved.

Description

Technical Field

[0001] The present invention relates to the field of transportation planning and management, and in particular to a method for optimizing a multi-depot distribution path of an electric refrigerated truck. Background Art

[0002] As the general public pays more and more attention to the quality, nutrition and freshness of fresh products such as vegetables and fruits, unprecedented requirements are put forward for the timely and effective distribution of fresh products. In order not to affect the actual perception of customers, enterprises need to ensure product quality and customer service time windows as much as possible. However, in view of the fact that such products must be refrigerated during the actual distribution process, they can be prevented from rotting and deteriorating as much as possible. Therefore, a certain refrigeration cost generated by the use of refrigeration equipment is inevitable. Moreover, during the distribution process, fresh products will also be slightly corrupted due to the influence of time, generating cargo damage costs. Therefore, it is necessary to effectively plan the distribution route to reduce the distribution cost. Among the currently used transportation route planning methods, there is no effective optimization method for the multi-yard distribution path of electric refrigerated trucks. The results obtained by optimizing the multi-yard distribution path of electric refrigerated trucks using the existing transportation route planning method often cannot achieve the goal of minimizing transportation costs. Therefore, it is necessary to propose a more scientific multi-yard distribution path optimization method for electric refrigerated trucks. Summary of the invention

[0003] In order to solve the above technical problems, the present invention provides a multi-yard distribution path optimization method for electric refrigerated trucks based on an improved whale optimization algorithm, which is special in that it includes the following steps:

[0004] S1, with the goal of minimizing the total distribution cost, constructs a multi-depot distribution route optimization function for pure electric refrigerated trucks;

[0005] S2, using clustering algorithm to allocate distribution points to different parking lots, transforming the multi-parking distribution path optimization function of pure electric refrigerated trucks into a single-parking distribution path optimization function of pure electric refrigerated trucks;

[0006] S3, using the whale optimization algorithm to solve the distribution path optimization function of a single parking lot of pure electric refrigerated trucks.

[0007] Furthermore, the whale optimization algorithm in S3 includes the following steps:

[0008] S301, setting the maximum number of iterations of the whale optimization algorithm and initializing its population;

[0009] S302, determine whether the maximum number of iterations has been reached, if so, the calculation ends, if not, jump to step S303;

[0010] S303, using the whale optimization algorithm to calculate the local optimal solution, and substitute it into the pure electric refrigerated vehicle single-park distribution path optimization function, after obtaining the target value, jump to S302,

[0011] Furthermore, the method for initializing the population in S301 includes the following steps:

[0012] S30101, select the customer point closest to the distribution center as the initial point;

[0013] S30102, declare an array dist to store the shortest distance from the initial point to each customer point and a set T of customer points of the shortest path;

[0014] S30103, the path weight of the initial point s is assigned to 0;

[0015] S30104, if there is a customer point m that can be directly reached from the initial point s, then set dist[m] = w, and at the same time set the path distances of the remaining customer points to infinity;

[0016] S30105, select the minimum data from the dist array, then the value is the shortest path from the initial point s to the other delivery point, and add the point to T. At this time, there are two delivery points in T;

[0017] S30106, check whether the newly added customer point can reach other customer points and check whether the path distance from the customer point to other customer points is shorter than the distance from the initial point directly. If so, replace the values ​​of these customer points in dist;

[0018] S30107, continue to find the minimum data from the dist array and repeat the above steps until all customer points are in T.

[0019] Furthermore, in the calculation of the local optimal solution using the whale optimization algorithm in S303, the update of the whale position can randomly select the spiral update strategy or the encirclement update strategy. If the spiral update strategy is selected, the update formula of the whale position is:

[0020]

[0021] D=|CX * 9t)-X(t)| (2)

[0022]

[0023] C=2r2 (4)

[0024]

[0025] Where D is the distance between the whale individual and the optimal individual, t is the current iteration number, X * (t) represents the position of the optimal individual, X(t) represents the current position of the whale individual, X(t+1) represents the updated position of the whale, r2 is a random number between [0, 1], l represents a random number between [-1, 1], and T is the maximum number of iterations;

[0026] If the encirclement update strategy is selected, the update formula for the whale position is

[0027]

[0028] D=|CX * (t)-X(t)| (7)

[0029] D rand =|CX rand (t)-X(t)| (8)

[0030] A=2αr1-α (9)

[0031] C=2r2 (10)

[0032]

[0033] X rand (t) is the individual position of the random whale, r1 and r2 are random numbers between [0, 1], D is the distance between the individual whale and the optimal individual, t is the current iteration number, X * (t) represents the position of the optimal individual, X(t) represents the current position of the whale individual, X(t+1) represents the updated whale position, l represents a random number between [-1, 1], T is the maximum number of iterations, and α is the linear convergence factor.

[0034] Furthermore, the multi-depot distribution path optimization function of pure electric refrigerated trucks in S1 is:

[0035]

[0036]

[0037]

[0038]

[0039]

[0040]

[0041]

[0042]

[0043] Among them, M is the parking lot set, N is the distribution point set, F is the charging station set, K is the vehicle set, is the time when vehicle k arrives at node i, is the travel time of vehicle k from node i to node j, is the service time at customer i, is the time when vehicle k departs from distribution center m, is the charging time required for vehicle k at the charging station, The time when all services are completed for vehicle k, SOC max is the maximum battery state of charge, SOC min The lowest safe state of charge. is the state of charge of vehicle k arriving at node i, P1 is the unit value of fresh products, P2 is the unit refrigeration cost during the distribution process, P ^ is the unit cooling cost during service, P4 is the unit charging cost during fast charging, P5 is the unit charging cost during slow charging, d ij is the distance from node i to node j, c f is the fixed cost of vehicle travel, c d is the unit distance travel cost, c dt is the penalty cost for late arrival per unit, TL2 is the heat load when the door is open, h is the battery power consumption rate, Q is the maximum load of the vehicle, B is the maximum capacity of the vehicle battery, θ1 is the corruption coefficient of fresh products during delivery, θ2 is the corruption coefficient of fresh products during service, and q i is the cargo demand of customer i, r is the charging rate of the vehicle battery, The quality of the remaining fresh products after the k-th vehicle has served customer point i, [ET i ,LT i ] is the time window of customer i, is a 0-1 variable, that is It means the kth vehicle passes through the road section ij, is a 0-1 variable, that is indicates that vehicle k serves customer i, is a 0-1 variable, that is indicates that vehicle k is charging at charging station i, r mk is a 0-1 variable, i.e. r mk =1 indicates that vehicle k is involved in fresh food delivery.

[0044] The multi-depot distribution path optimization function of pure electric refrigerated trucks aims to minimize the cost. Certain model assumptions and parameter descriptions are made to supplement the conditions required by the model, and the fixed cost, variable cost, charging cost, refrigeration cost, cargo damage cost and time window penalty cost in the distribution process are analyzed. Finally, based on the above analysis, the multi-depot distribution path optimization of pure electric refrigerated trucks with the minimum total distribution cost as the objective function is established.

[0045] Furthermore, the S2, using a clustering algorithm to allocate the delivery points to different parking lots, wherein the clustering algorithm is a K-means algorithm, comprises the following steps:

[0046] S201, obtaining location information of X customer points;

[0047] S202, randomly selecting Y customers from the X customers as cluster centers, and determining the number of categories;

[0048] S203, calculating the distance between each customer and Y cluster centers, and then assigning the customer to the category to which the cluster center with the smallest distance belongs, repeating X times to form Y customer groups;

[0049] S204, recalculating the cluster center for the obtained customer group;

[0050] S205, return to S202, when the position of the cluster center no longer changes, the clustering ends and the corresponding result is output.

[0051] The K-means algorithm is used to cluster customers according to their distance from each parking lot. The centroid data after clustering is obtained through iterative calculation. The distance between each centroid and each parking lot is calculated, and the nearest parking lot is selected as the delivery parking lot for this type of customer, thereby realizing the specific allocation of customers.

[0052] The present invention uses a two-stage hybrid algorithm to solve the multi-yard distribution path optimization model of pure electric refrigerated trucks. First, the multi-yard problem is transformed into a single-yard problem using the k-means algorithm, and then the population is initialized using the improved whale optimization algorithm. At the same time, the improved position update strategy is added to update the position, so that the pure electric refrigerated truck distribution path optimization problem can be efficiently optimized and solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The drawings in the specification, which constitute a part of this application, are used to provide a further understanding of the present invention. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0054] Figure 1 It is a schematic diagram of the process of the multi-depot distribution path optimization method of a pure electric refrigerated truck of the present invention;

[0055] Figure 2 This is a flow chart of the whale optimization algorithm in the present invention;

[0056] Figure 3 It is a flow chart of the population initialization method in the present invention;

[0057] Figure 4 It is a schematic diagram of the flow of the clustering algorithm in the present invention. DETAILED DESCRIPTION

[0058] In order to deepen the understanding of the present invention, the present invention will be further described in detail below with reference to embodiments and drawings.

[0059] Example 1

[0060] like Figure 1 As shown, a method for optimizing the multi-depot distribution path of a pure electric refrigerated truck in this embodiment includes the following steps:

[0061] S1, with the goal of minimizing the total distribution cost, constructs a multi-depot distribution route optimization function for pure electric refrigerated trucks;

[0062] S2, using clustering algorithm to allocate distribution points to different parking lots, transforming the multi-parking distribution path optimization function of pure electric refrigerated trucks into a single-parking distribution path optimization function of pure electric refrigerated trucks;

[0063] S3, using the whale optimization algorithm to solve the distribution path optimization function of a single parking lot of pure electric refrigerated trucks.

[0064] The whale optimization algorithm in S3 includes the following steps:

[0065] S301, setting the maximum number of iterations of the whale optimization algorithm and initializing its population;

[0066] S302, determine whether the maximum number of iterations has been reached, if so, the calculation ends, if not, jump to step S303;

[0067] S303, using the whale optimization algorithm to calculate the local optimal solution, and substitute it into the pure electric refrigerated vehicle single-park distribution path optimization function, after obtaining the target value, jump to S302,

[0068] The method for initializing the population in S301 includes the following steps:

[0069] S30101, select the customer point closest to the distribution center as the initial point;

[0070] S30102, declare an array dist to store the shortest distance from the initial point to each customer point and a set T of customer points of the shortest path;

[0071] S30103, the path weight of the initial point s is assigned to 0;

[0072] S30104, if there is a customer point m that can be directly reached from the initial point s, then set dist[m] = w, and at the same time set the path distances of the remaining customer points to infinity;

[0073] S30105, select the minimum data from the dist array, then the value is the shortest path from the initial point s to the other delivery point, and add the point to T. At this time, there are two delivery points in T;

[0074] S30106, check whether the newly added customer point can reach other customer points and check whether the path distance from the customer point to other customer points is shorter than the distance from the initial point directly. If so, replace the values ​​of these customer points in dist;

[0075] S30107, continue to find the minimum data from the dist array and repeat the above steps until all customer points are in T.

[0076] In the calculation of the local optimal solution by the whale optimization algorithm in S303, the update of the whale position can randomly select the spiral update strategy or the encirclement update strategy. If the spiral update strategy is selected, the update formula of the whale position is:

[0077]

[0078] D=|CX * (t)-X(t)| (2)

[0079]

[0080] C=2r2 (4)

[0081]

[0082] Where D is the distance between the whale individual and the optimal individual, t is the current iteration number, X * (t) represents the position of the optimal individual, X(t) represents the current position of the whale individual, X9t+1) represents the updated whale position, r2 is a random number between [0, 1], l represents a random number between [-1, 1], and T is the maximum number of iterations;

[0083] If the encirclement update strategy is selected, the update formula for the whale position is

[0084]

[0085] D=|CX * (t)-X(t)| (7)

[0086] D rand =|CX rand (t)-X(t)| (8)

[0087] A=2αr1-α (9)

[0088] C=2r2 (10)

[0089]

[0090] X rand (t) is the individual position of the random whale, r1 and r2 are random numbers between [0, 1], D is the distance between the individual whale and the optimal individual, t is the current iteration number, X * (t) represents the position of the optimal individual, X(t) represents the current position of the whale individual, X(t+1) represents the updated whale position, l represents a random number between [-1, 1], T is the maximum number of iterations, and 1 is the linear convergence factor.

[0091] The multi-depot distribution path optimization function of pure electric refrigerated trucks in S1 is:

[0092]

[0093]

[0094]

[0095]

[0096]

[0097]

[0098]

[0099]

[0100] Among them, M is the parking lot set, N is the distribution point set, F is the charging station set, K is the vehicle set, is the time when vehicle k arrives at node i, is the travel time of vehicle k from node i to node j, is the service time at customer i, is the time when vehicle k departs from distribution center m, is the charging time required for vehicle k at the charging station, The time when all services are completed for vehicle k, SOC max is the maximum battery state of charge, SOC min The lowest safe state of charge. is the state of charge of vehicle k arriving at node i]1 is the unit value of fresh products,]2 is the unit refrigeration cost during the distribution process,] ^ is the unit cooling cost for service, ]4 is the unit charging cost for fast charging, ]5 is the unit charging cost for slow charging, d ij is the distance from node i to node j, c f is the fixed cost of vehicle travel, c a is the unit distance travel cost, c at is the penalty cost for late arrival per unit, TL2 is the heat load when the door is opened, h is the battery power consumption rate, Q is the maximum load of the vehicle, B is the maximum capacity of the vehicle battery, is the corruption coefficient of fresh products during distribution, is the corruption coefficient of fresh products during service, q i is the cargo demand of customer i, r is the charging rate of the vehicle battery, The quality of the remaining fresh products after the k-th vehicle has served customer point i, [ET i ,LT i ] is the time window of customer i, is a 0-1 variable, that is It means the kth vehicle passes through the road section ij, is a 0-1 variable, that is indicates that vehicle k serves customer i, is a 0-1 variable, that is indicates that vehicle k is charging at charging station i, r mk is a 0-1 variable, i.e. r mk =1 indicates that vehicle k is involved in fresh food delivery.

[0101] The S2 uses a clustering algorithm to allocate the delivery points to different parking lots. The clustering algorithm is a K-means algorithm, which includes the following steps:

[0102] S201, obtaining location information of X customer points;

[0103] S202, randomly selecting Y customers from the X customers as cluster centers, and determining the number of categories;

[0104] S203, calculating the distance between each customer and Y cluster centers, and then assigning the customer to the category to which the cluster center with the smallest distance belongs, repeating X times to form Y customer groups;

[0105] S204, recalculating the cluster center for the obtained customer group;

[0106] S205, return to S202, when the position of the cluster center no longer changes, the clustering ends and the corresponding result is output.

[0107] The logistics distribution network formed in this embodiment includes 2 parking lots with pure electric refrigerated trucks of the same brand and model, 10 public charging stations distributed outside the parking lots, and 45 customers waiting for delivery services. The time window and service time of each customer are different. The vehicle must be delivered within the customer's time window. If it fails to be delivered within the time window, a certain penalty cost will be incurred according to the standards of each customer.

[0108] The parking lot location, charging station location, and customer parameters are shown in Tables 1, 2, and 3, respectively.

[0109] Table 1 Parking lot parameters

[0110]

[0111] Table 2 Charging station parameters

[0112]

[0113] Table 3 Some customer parameters

[0114]

[0115]

[0116]

[0117] According to the k-means clustering algorithm, the 45 customer points are divided into two categories. The first category has 23 customers and the second category has 22 customers. The specific clustering results are shown in Table 4.5.

[0118] Table 4 Clustering results

[0119]

[0120] Then, the whale algorithm and the improved algorithm were used to solve the problem for the customer. The two algorithms were run 10 times respectively and the average value was taken. The total delivery costs calculated by the whale algorithm and the optimization method proposed by the present invention were 9129.301 yuan and 4830.518 yuan respectively. The calculation of the optimization method proposed by the present invention reduced the total delivery cost by 47.09% compared with the traditional whale algorithm. The effectiveness and superiority of the algorithm proposed by the present invention were proved.

[0121] The specific solutions of the whale algorithm and the improved two-stage hybrid algorithm proposed in this paper are shown in Table 5.

[0122] Table 5 Distribution cost table

[0123]

[0124] According to the algorithm proposed in this invention, the optimal delivery route is shown in Table 6.

[0125] Table 6 Optimal delivery route

[0126]

[0127] As shown in Table 6, the algorithm proposed in the present invention needs 4 pure electric refrigerated trucks to solve the example, which means 4 routes appear and 4 charging times are required. The distribution cost required for distribution center 1 is 2332.523 yuan, and the distribution cost required for distribution center 2 is 2497.995 yuan.

[0128] Example 2

[0129] The specific method of Example 2 is the same as that of Example 1, except that this example includes 3 distribution centers and 15 charging stations. The coordinates of the distribution centers, the coordinates of the charging stations, and the customer order status are shown in Tables 7, 8, and 9, respectively.

[0130] Table 7 Distribution center coordinates

[0131]

[0132] Table 8 Charging station coordinates

[0133]

[0134]

[0135]

[0136]

[0137]

[0138]

[0139] First, the 100 customer points are divided into three categories using the k-means clustering algorithm. The first category has 21 customers, the second category has 44 customers, and the third category has 35 customers. The clustering is as follows: Figure 4 The specific clustering results are shown in Table 10.

[0140] Table 10 Distribution center corresponding customer number

[0141]

[0142]

[0143] Then, the whale algorithm, genetic algorithm and improved whale algorithm are used to solve each type of customer. The specific solutions are shown in Table 11.

[0144] Table 11 Distribution costs

[0145]

[0146] As shown in Table 11, the total delivery costs calculated by the whale algorithm, genetic algorithm and the algorithm proposed in this paper are 10916.27 yuan, 9981.099 yuan and 6759.825 yuan respectively. The delivery cost graphs are shown in 5.3, 5.4 and 5.5 respectively. Among them, the algorithm proposed in this paper reduces the total delivery cost by 38.08% and 31.91% compared with the traditional whale algorithm and genetic algorithm. The effectiveness and superiority of the algorithm proposed in this paper are proved.

[0147] According to the algorithm proposed in this paper, the optimal delivery route is shown in Table 12.

[0148] Table 12 Optimal delivery route

[0149]

[0150]

[0151] The above are only specific embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for optimizing the distribution path of a pure electric refrigerated truck in multiple parking lots, characterized in that: It includes the following steps: S1, with the goal of minimizing the total distribution cost, constructs a multi-depot distribution route optimization function for pure electric refrigerated trucks; S2, using clustering algorithm to allocate distribution points to different parking lots, transforming the multi-parking distribution path optimization function of pure electric refrigerated trucks into a single-parking distribution path optimization function of pure electric refrigerated trucks; S3, using the whale optimization algorithm to solve the distribution path optimization function of a single parking lot of pure electric refrigerated trucks.

2. The method for optimizing the multi-depot distribution path of a pure electric refrigerated truck as claimed in claim 1, characterized in that: The whale optimization algorithm in S3 includes the following steps: S301, setting the maximum number of iterations of the whale optimization algorithm and initializing its population; S302, determine whether the maximum number of iterations has been reached, if so, the calculation ends, if not, jump to step S303; S303, using the whale optimization algorithm to calculate the local optimal solution, and substitute it into the pure electric refrigerated vehicle single parking lot distribution path optimization function, after obtaining the target value, jump to S302.

3. The method for optimizing the multi-depot distribution path of a pure electric refrigerated truck as claimed in claim 2, characterized in that: The method for initializing the population in S301 includes the following steps: S30101, select the customer point closest to the distribution center as the initial point; S30102, declare an array dist to store the shortest distance from the initial point to each customer point and a set T of customer points of the shortest path; S30103, the path weight of the initial point s is assigned to 0; S30104, if there is a customer point m that can be directly reached from the initial point s, then set dist[m] = w, and at the same time set the path distances of the remaining customer points to infinity; S30105, select the minimum data from the dist array, then the value is the shortest path from the initial point s to the other delivery point, and add the point to T. At this time, there are two delivery points in T; S30106, check whether the newly added customer point can reach other customer points and check whether the path distance from the customer point to other customer points is shorter than the distance from the initial point directly. If so, replace the values ​​of these customer points in dist; S30107, continue to find the minimum data from the dist array and repeat the above steps until all customer points are in T.

4. The method for optimizing the multi-depot distribution path of a pure electric refrigerated truck as claimed in claim 2, characterized in that: In the calculation of the local optimal solution by the whale optimization algorithm in S303, the update of the whale position can randomly select the spiral update strategy or the encirclement update strategy. If the spiral update strategy is selected, the update formula of the whale position is: D=|CX * (t)-X(t)| (2) C＝2r2 (4) Where D is the distance between the whale individual and the optimal individual, t is the current iteration number, X * (t) represents the position of the optimal individual, X(t) represents the current position of the whale individual, X9t+1) represents the updated whale position, r2 is a random number between [0, 1], l represents a random number between [-1, 1], and T is the maximum number of iterations; If the encirclement update strategy is selected, the update formula for the whale position is D=|CX * (t)-X(t)| (7) D rand =|CX rand (t)-X(t)| (8) A＝2αr1-α (9) C＝2r2 (10) X rand (t) is the individual position of the random whale, r1 and r2 are random numbers between [0, 1], D is the distance between the whale individual and the optimal individual, t is the current iteration number, X * (t) represents the position of the optimal individual, X(t) represents the current position of the whale individual, X(t+1) represents the updated whale position, l represents a random number between [-1, 1], T is the maximum number of iterations, and α is the linear convergence factor.

5. The method for optimizing the multi-depot distribution route of a pure electric refrigerated truck as claimed in claim 1, characterized in that: The multi-depot distribution path optimization function of pure electric refrigerated trucks in S1 is: Among them, M is the parking lot set, N is the distribution point set, F is the charging station set, K is the vehicle set, is the time when vehicle k arrives at node i, is the travel time of vehicle k from node i to node j, is the service time at customer i, is the time when vehicle k departs from distribution center m, is the charging time required for vehicle k at the charging station, The time when all services are completed for vehicle k, SOC max is the maximum battery state of charge, SOC min The lowest safe state of charge. is the state of charge of vehicle k arriving at node i, P1 is the unit value of fresh products, P2 is the unit refrigeration cost during the distribution process, P ^ is the unit cooling cost during service, P4 is the unit charging cost during fast charging, P5 is the unit charging cost during slow charging, d ij is the distance from node i to node j, c f is the fixed cost of vehicle travel, c d is the unit distance travel cost, c dt is the penalty cost for late arrival per unit, TL2 is the heat load when the door is opened, h is the battery power consumption rate, Q is the maximum load of the vehicle, B is the maximum capacity of the vehicle battery, is the corruption coefficient of fresh products during distribution, is the corruption coefficient of fresh products during service, q i is the cargo demand of customer i, r is the charging rate of the vehicle battery, The quality of the remaining fresh products after the k-th vehicle has served customer point i, [ET i ,LT i ] is the time window of customer i, is a 0-1 variable, that is It means the kth vehicle passes through the road section ij, is a 0-1 variable, that is indicates that vehicle k serves customer i, is a 0-1 variable, that is m k Indicates that vehicle k is charging at charging station i, and r is mk 0-1 variable, i.e. r = 1 means vehicle k is involved in fresh food delivery.

6. The method for optimizing the distribution path of a pure electric refrigerated truck in multiple parking lots as claimed in claim 1, wherein S2, using a clustering algorithm to allocate distribution points to different parking lots, is a K-means algorithm, and comprises the following steps: S201, obtaining location information of X customer points; S202, randomly selecting Y customers from the X customers as cluster centers, and determining the number of categories; S203, calculating the distance between each customer and Y cluster centers, and then assigning the customer to the category to which the cluster center with the smallest distance belongs, repeating X times to form Y customer groups; S204, recalculating the cluster center for the obtained customer group; S205, return to S202, when the position of the cluster center no longer changes, the clustering ends and the corresponding result is output.

Images