Digital logistics platform construction method based on intelligent model

Abstract

The invention discloses a digital logistics platform construction method based on an intelligent model, and the method comprises the steps: assisting the construction of an intelligent optimization module through an improved FA module, completing the high-efficiency order batching, building a logistics department screening model based on LMDI, evaluating the contribution degree of a logistics department through the LMDI model, optimizing the logistics distribution through the weather, the route and the order batch, and improving the logistics distribution efficiency. According to the method, the influence of objective factor abnormity on the logistics efficiency is reduced, the transportation cost and the time delay are reduced, the production efficiency is improved by utilizing the aspects of credit, productivity, logistics regulation and control, the receiving and delivery efficiency of a transportation department, the credit investigation of customers and the like, the logistics efficiency and the logistics circulation robustness under the condition of multiple orders and multiple data are further improved, and the logistics efficiency and the logistics circulation robustness are improved. And an enterprise is assisted to optimize fund circulation and make a more accurate business decision.

Description

Technical Field

[0001] The present invention relates to the field of digital logistics, and particularly to a method for constructing a digital logistics platform based on an intelligent model. Background Art

[0002] Traditional inbound and outbound logistics platforms mainly rely on manual operations and basic computer systems to manage logistics activities. These platforms usually adopt several common warehouse management outbound strategies, such as FIFO, FEFO, and LIFO, etc. These strategies together constitute the basis for inbound and outbound management of traditional logistics platforms. With the progress of technology, digital intelligent logistics platforms can be more accurate and intelligent in aspects such as logistics link monitoring and commodity screening, and are an important pillar of the current logistics industry.

[0003] However, with the progress of technology, the development of society, and the continuous operation of various factory assembly lines, there are increasingly large orders and capital flows. Facing more and more complex data, the limitations of traditional logistics platforms that only focus on inbound and outbound types are also becoming higher and higher. It is difficult to select the optimal logistics plan through a large amount of logistics data, improve circulation efficiency, save transportation energy consumption, and then achieve the purpose of reducing enterprise costs or increasing profits. Therefore, there is an urgent need in the market for a method that further integrates digital models, extracts valuable information during the logistics process, optimizes inventory management, further reduces logistics transportation costs while improving the distribution speed and quality, so as to help enterprises make more accurate business decisions. Summary of the Invention

[0004] Object of the Invention: To solve the problems mentioned in the background art, the present invention discloses a method for constructing a digital logistics platform based on an intelligent model. By improving the FA algorithm to construct an intelligent optimization module for order batching, cooperating with the LMDI model to screen the best logistics cost plan and optimize logistics priority, it further improves the logistics efficiency and the robustness of the logistics cycle in the case of multiple orders and multiple data, assists enterprises in optimizing the capital cycle, and makes more accurate business decisions.

[0005] Technical Solution:

[0006] The present invention discloses a method for constructing a digital logistics platform based on an intelligent model, and the method includes the following steps:

[0007] S1 Build a credit platform with the raw material source, screen the raw material source, and establish a production workshop of the source factory.

[0008] S2 Construct a logistics platform, and based on step size optimization, improve the FA algorithm to construct an intelligent optimization module to batch the orders for the goods produced by the production workshop of the source factory.

[0009] S3 evaluates the contribution of each logistics department based on the intelligent model through the logistics operation efficiency, and screens the optimal logistics department for handover and transportation.

[0010] S4 introduces three parameters: weather conditions, optimized delivery routes, and the number of orders in different batches. The transportation fleet of the logistics department is quantitatively screened through the LMDI model, and the goods are transported to the freight station for review.

[0011] S5 Customers order the goods that have passed the review through the logistics platform. The logistics platform will send the commodity orders to different destinations according to the customer's needs. The platform establishes a personal credit index model to calculate the customer's credit, and feedbacks the credit information to the credit platform to improve the logistics efficiency by giving priority to processing the orders of customers with high credit.

[0012] Furthermore, the screening of the production workshops of the source factory described in S1 is based on the production capacity and production efficiency of the production workshops. The higher the production capacity and production efficiency, the higher the priority. The calculation formula for the production capacity of the production workshops of the source factory is as follows:

[0013]

[0014] Among them, M is the equipment production capacity, H e is the effective working time of the equipment in the planning period, and t i is the time quota for processing the product on this equipment, with the unit of hour / piece.

[0015] The calculation formula for the production efficiency of the production workshops of the source factory is as follows:

[0016]

[0017] Among them, η is the production efficiency, p a is the actual output, t s is the standard working hours, m a is the actual labor force, t w is the actual working time, t b is the blocked production hours, t o is the overtime hours.

[0018] Furthermore, the intelligent optimization module is constructed by optimizing and improving the FA algorithm based on the step size described in S2. The specific steps are as follows:

[0019] The FA algorithm uses real number coding to generate the initial population. Each firefly represents a potential solution, the position represents the variable value, and the brightness represents the fitness. When decoding, the solution space is converted into the actual problem solution space. Considering the maximum number of orders, the position is decoded through the objective function value.

[0020] The individual brightness of the FA algorithm

[0021] The luminous brightness of the FA algorithm is designed as the reciprocal of the total number of times of shelf handling, and the optimal solution to the order batching problem is found:

[0022]

[0023] Define the distance as the number of different digits at the same position. Assume firefly Z 1 =[1, 3, 1, 3, 4, 1, 2, 5], firefly Z 2 =[1, 4, 3, 3, 4, 1, 3, 5], then the distance is 3.

[0024] The FA algorithm updates the position by making the fireflies move towards the brightest firefly and updating the inconsistent order numbers. At the same time, the brightest firefly also makes random moves to avoid falling into a local optimum. The termination condition of the algorithm is to reach the maximum number of iterations or find a better objective function value. If the condition is met, the result is output; otherwise, it continues to execute;

[0025] Create an intelligent optimization module model with the order batch as the objective function. The calculation formula is as follows:

[0026]

[0027] Among them, Num is the total number of times of shelf handling required for order batching, N is the order batch, Y is the number of shelves, and e ny indicates whether the order batch N needs to handle shelf Y.

[0028] Furthermore, different constraint conditions need to be set for different attributes of the orders to enable the orders of various commodity types to operate normally. The constraint conditions of the intelligent optimization module model are as follows:

[0029]

[0030] Among them, A is the order set; B 1 is the order batch set; D is the shelf set; P is the storage capacity of the turnover shelf at the picking station; X is the commodity; Z is the order; d zn indicates whether the order Z is assigned to batch n; Count.y represents the number of times of handling shelf Y. Establish an order similarity matrix. Each element in the similarity matrix represents the similarity value between two orders. Find the order corresponding to the largest matrix value, construct an order batch set, consider whether the limit condition of the total storage capacity is met. If the condition is met, add a new order; otherwise, delete the order and establish a new order batch set. Repeat the above operations until the batch division of all orders is completed.

[0031] Furthermore, based on the step size optimization and improvement of the FA algorithm, an intelligent optimization module is constructed. The specific improvement steps are as follows:

[0032] (1) By changing the value of ξ, the performance and adaptability of the FA algorithm can be improved, which is used to evaluate and select similar order goods:

[0033]

[0034] Υ represents the distance between order goods, and Υ 0 represents the similarity of order goods, and β represents the distance between similar order goods;

[0035] By considering the similarity and difference between solutions, the FA algorithm can effectively explore and develop in the search space, and the calculation formula is as follows:

[0036] Assume that all fireflies are M best and M worst are the most similar and most different individuals in each iteration, and these fireflies have all been optimized. M 1 , M 2 , M 3 , M 4 are the types of goods randomly selected in the order, where M 1 ≠ M 2 ≠ M 3 ≠ M 4 , and the similarities of the 4 kinds of goods are compared pairwise. The value range of the random parameter α is [0,1]. Mnew1 and Mnew2 represent the most similar order types of goods after pairwise comparison. M 5 represents the difference degree between the most similar and most different orders, and the definition formula is as follows:

[0037] M 1 = {M new1 if k 1 ≤ k 2 M best if k 1 > k 2}

[0038] M 2 = {M new1 if k 3 ≤ k 2 M j if k 3 > k 2}

[0039] M 3 = {M new1，i if k 4 ≤ k 3 M j if k 4 > k 3}

[0040] M4 = M new1 if k 5 ≤ k 4 or k 5 > k 4

[0041] M 5 = δ × M worst + ψ × (M best - M worst )

[0042] where k 1 , k 2 , k 3 , k 4 , k 5 , ψ, and δ are all random parameters, taking values within [0, 1]. Assume that the function value of the i-th firefly is greater than that of the optimal firefly. The i-th firefly will replace the position of the optimal firefly. Assume that the function value of the i-th firefly is less than that of the optimal firefly, then no replacement will occur.

[0043] The random parameter ξ in the best objective iteration equation can control the random search of the algorithm. Modify the parameter ξ using the following formula and substitute it into the objective iteration equation:

[0044]

[0045] During the optimization process, the algorithm achieves a sufficient balance between local search and global search by changing the value of ξ, where Iter is the firefly iteration number, and G max is the maximum iteration number.

[0046] Change the value of Υ 0 value

[0047] In the original equation, Υ 0 is a constant. Now introduce the loop variable p to modify it:

[0048]

[0049] μ is a constant, MaxIter is the maximum iteration number. After modifying Υ 0 and substituting it into the mutual attraction equation, we get:

[0050]

[0051] Modify Υ 0 . After controlling the parameters and improving the algorithm, the objective iteration formula is:

[0052]

[0053] where Xi and Xj are the states of fireflies, is an exponential decay function used to simulate the phenomenon that the similarity or attraction between solutions rapidly decreases as the distance increases.

[0054] The search step size is optimized and improved, and the improvement formula is as follows:

[0055] R = φ × e -(σ×g) / Marge

[0056] where φ is the step size control factor, σ is the exponential regulation factor, g is the current iteration number, Marge is the maximum iteration number, and after the algorithm output optimization, the optimal order batch planning Order is obtained best .

[0057] Furthermore, the logistics operation efficiency described in S3 is the efficiency of the logistics platform during the process of goods transportation, storage, and distribution, including time efficiency, cost efficiency, and service level parameters. The formula is as follows:

[0058] η a = λ·η b + μ·η t + ρ·s

[0059] where λ, μ, and ρ are weight factors adjusted according to different business priorities; η b is the cost efficiency; η t is the time efficiency; s is the service level.

[0060] The steps to create the LMDI model are as follows:

[0061] The logistics operation efficiency η a , cost efficiency η b , time efficiency η t and service level s are decomposed through the LMDI model, and the change of each factor is expressed as:

[0062] Δη b = η b1 - η b0 = (η b1 / η b0 ) - 1

[0063] Δη t = η t1 - η t0 = (η t1 / η t0 ) - 1

[0064] Δs = s t1 - s t0 = (s t1 / s t0)-1

[0065] Among them, the subscripts t0 and t1 represent the initial and final states respectively;

[0066] The total change of the LMDI model can be expressed as:

[0067] Δη a = LMDI(η b ) + LMDI(η t ) + LMDI(s)

[0068] The LMDI model further decomposes each factor into:

[0069]

[0070] Among them, Wη b ,Wη t and Ws are the weights of cost efficiency, time efficiency and service level respectively, and these weights can be adjusted according to different business priorities.

[0071] The total change of the contribution degree of logistics operation efficiency can be expressed as:

[0072]

[0073] Select the logistics department with the highest total change of contribution degree as the cooperation partner, and let this logistics department screen the transportation fleet.

[0074] Furthermore, the specific steps of S4 are as follows:

[0075] The logistics distribution mechanism introduces three key parameters, and their calculation formulas are as follows:

[0076] Weather condition impact factor W:

[0077]

[0078] Among them, D w is the number of days of transportation delay caused by weather conditions; D max is the maximum number of delay days in historical data, and the exponential function is used to standardize the weather impact to a value between 0 and 1, where 1 means no delay and 0 means the maximum delay.

[0079] Route optimization factor R:

[0080]

[0081] Among them, C original is the transportation cost of the original route, C optimized is the cost of the optimized route, and the ratio measures the effect of route optimization, and the smaller the value, the better the optimization effect;

[0082] Order batch optimization factor B:

[0083]

[0084] Where Q i is the batch size of the i-th order; Q avg is the average batch size of all orders; n is the total number of orders;

[0085] The calculation formula of the LMDI model is as follows:

[0086] Where ΔV tot is the total change, and ΔV xi represents the change of each factor;

[0087] Quantify the impact factor of weather conditions:

[0088]

[0089] Where L(V T ,V 0 ) is the logarithmic mean Divisia index, X T and X 0 represent the transportation efficiency and the original transportation efficiency after considering the weather conditions respectively;

[0090] Quantify the order batch optimization factor:

[0091]

[0092] Where V iT and V i0 represent the distribution route efficiency before and after optimization respectively, and x k,iT and x k,i0 represent the distribution route parameters before and after optimization respectively.

[0093] Quantify the order batch optimization factor:

[0094]

[0095] Where V bT and V b0 represent the resource allocation efficiency before and after the optimization of the number of orders in different batches respectively, and x k,bT and x k,b0 represent the order number parameters before and after optimization respectively.

[0096] Combined with the LMDI model, the calculation formula of the change in contribution degree is as follows:

[0097] ΔV tot =ΔV w +ΔV r +ΔV b

[0098] Select the three with the highest contribution as the final transportation fleet.

[0099] Further, the personal credit index model described in S5 is as follows:

[0100] C s = w 1 ·C 1 + w 2 ·C 2 + w 3 ·C 3

[0101] C 1 = log(1 + C f )

[0102]

[0103] Among them, C s is the credit score, C 1 is the capital score, C 2 is the transaction volume score, C 3 is the return rate score, C f is the customer capital flow details, C t is the customer's cumulative transaction volume, C max the maximum transaction volume, C r is the customer's return rate, C' max is the maximum return rate, w 1 、w 2 、w 3 are the weight factors of capital, transaction volume, and return rate respectively, and the commodity orders are allocated from high to low according to the personal credit index.

[0104] Beneficial effects:

[0105] 1. By introducing an automated algorithm in combination with a digital platform, the present invention reduces the dependence on manual labor, improves the accuracy of logistics transportation while reducing logistics costs, and further reduces errors.

[0106] 2. The present invention uses the LMDI model to evaluate the contribution of the logistics department, optimizes logistics allocation through weather, routes, and order batches, reduces the impact of objective factor anomalies on logistics efficiency, and reduces transportation costs and time delays.

[0107] 3. The present invention uses an improved FA algorithm to construct an intelligent optimization module, batch orders in a refined manner, and records important customer information to incorporate it into the logistics system cycle, enhancing customer trust while accelerating the order processing efficiency and better maintaining the stability of the capital chain. Description of the Drawings

[0108] Figure 1 Schematic diagram of the structure of an embodiment of the present invention;

[0109] Figure 2 Flow chart of the operation of an embodiment of the present invention;

[0110] Figure 3 Flow chart of the FA algorithm of the present invention. Specific implementation manners

[0111] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0112] Further improve the logistics efficiency and the robustness of the logistics cycle in the case of multiple orders and multiple data, assist enterprises in optimizing the capital cycle, and make more accurate business decisions.

[0113] As Figures 1-3 shown, this embodiment discloses a method for constructing a digital logistics platform based on an intelligent model, which mainly includes the following steps:

[0114] S1: Build a credit platform with each raw material source, screen out suitable raw material sources and sign credit contracts with them (select the price, quantity, quality, etc. of raw materials by preference). The production workshop of the source factory can optimize the production process through the production workshop. By introducing automated and intelligent equipment, the dependence on manual labor can be reduced, the production efficiency and accuracy can be improved, and the production capacity and production efficiency of the production workshop can be increased. Using these devices can complete repetitive work, save time and labor costs, thereby improving product quality and stability.

[0115] The calculation formula for the production capacity of the production workshop is as follows:

[0116]

[0117] Among them, M is the production capacity of the equipment, and H e is the effective working time (hours) of the equipment in the planned period (year), and t i is the time quota (hours / piece) for processing the product on this equipment.

[0118] The calculation formula for the production efficiency of the production workshop is as follows:

[0119]

[0120] Among them, η is the production efficiency, p a is the actual output, and ts is the standard man-hour, m a is the actual workforce, t w is the actual working hours, t b is the downtime man-hour, t o is the overtime man-hour.

[0121] S2: Create an intelligent optimization module which uses the FA algorithm to refine the order batches of the source factories in the mass production workshop to meet the standard of transport batching.

[0122] The intelligent optimization module uses the FA algorithm to optimize the product order batching. The specific steps of the FA algorithm are as follows:

[0123] Definition 1: The brightness of a firefly relative to surrounding individuals

[0124]

[0125] I i is the brightness of firefly i, γ is the light absorption factor, γ ij is the Euclidean distance between fireflies i and j.

[0126] Definition 2: The relative attraction between firefly individuals

[0127]

[0128] β 0 is the maximum attraction of the firefly, located at the light source; γ, r ij have the same meaning as above.

[0129] Definition 3: Firefly j is attracted by firefly i, and the position update of firefly j

[0130]

[0131] t is the current iteration number of the algorithm, X j (t + 1) is the position of firefly j at the (t + 1)-th iteration, β ij is the attraction of firefly i to j, α is a constant, usually taking α ∈ [0, 1], and rand is a random factor on [0, 1].

[0132] Order batching belongs to the NP-hard problem. As the number of orders increases, the computational complexity also increases, and it is difficult to obtain accurate results. Heuristic algorithms can be used to solve it. The FA algorithm is used to optimize the order batching problem to obtain the optimal batching result. The design idea of the FA algorithm is as follows:

[0133] (1) Encoding of the FA algorithm

[0134] Generate an initial firefly population using real - number coding. Each firefly represents a potential solution. The position and brightness of a firefly represent the variable value and fitness value of the solution respectively.

[0135] (2) Decoding of the FA algorithm

[0136] The process of transforming the solution space of the FA algorithm into the solution space of the actual problem. Considering the maximum number of orders in a batch, decode the firefly position by calculating the objective function value.

[0137] (3) Individual brightness of the FA algorithm

[0138] The goal of the order batching problem is to solve a minimization problem. The luminous brightness of the FA algorithm is designed as the reciprocal of the total number of shelf handling times. The moving direction of a firefly individual is related to the magnitude of the brightness, and the moving distance is related to the magnitude of the attraction. Through algorithm iteration, update the brightness and attraction of the firefly individual to find the optimal solution to the order batching problem.

[0139]

[0140] It can be seen from the formula that the brightness of a firefly is related to the distance. To facilitate the calculation of brightness, define the distance as the number of different digits at the same position. Suppose firefly Z 1 =[1, 3, 1, 3, 4, 1, 2, 5], firefly Z 2 =[1, 4, 3, 3, 4, 1, 3, 5], then the distance is 3.

[0141] (4) Position update of the FA algorithm

[0142] After determining the brightest firefly k, firefly i moves in the direction of the brightest one. Search for the order numbers with inconsistent batches in the firefly solution and update the position with reference to the solution of the brighter firefly.

[0143] Random movement update of firefly individuals. For the brightest firefly k, perform random movement. To avoid falling into local optima, the random movement is to randomly replace any two orders in different batches.

[0144] (5) Termination condition of the FA algorithm

[0145] The termination condition of the algorithm is to judge whether the current iteration has reached the maximum number of times or there is a better objective function value. If the condition is met, output the order batches; otherwise, continue to execute the algorithm to solve the problem.

[0146] Create an intelligent optimization module model with the order batch as the objective function. The calculation formula is as follows:

[0147]

[0148] Among them, Num is the total number of times the shelves need to be moved for order batching, N is the order batch, Y is the number of shelves, and e ny indicates whether the order batch N needs to move the shelf Y.

[0149] Due to the different attributes of orders, different constraint conditions need to be set to enable orders of various commodity types to operate normally. The constraint condition formula is as follows:

[0150]

[0151]

[0152] Among them, A is the order set; B 1 is the order batch set; D is the shelf set; P is the storage capacity of the turnover shelves at the picking station; X is the commodity; Z is the order; d zn indicates whether the order Z is assigned to batch n; Count.y represents the number of times the shelf Y is moved.

[0153] Due to the different attributes of orders, different constraint conditions need to be set to enable orders of various commodity types to operate normally. The constraint condition formula is as follows:

[0154]

[0155] Among them, A is the order set; B 1 is the order batch set; D is the shelf set; P is the storage capacity of the turnover shelves at the picking station; X is the commodity; Z is the order; d zn indicates whether the order Z is assigned to batch n; Count.y represents the number of times the shelf Y is moved.

[0156] By calculating the similarity method, the similarity result value between any two orders is obtained, and an order similarity matrix is established. Each element in the similarity matrix represents the similarity value between two orders, and the order corresponding to the largest matrix value is found. A set of order batches is constructed, considering whether the limit condition of the total storage capacity is satisfied. If the condition is met, a new order is added; otherwise, the order is deleted, and a new set of order batches is established. Repeat the above operations until the batch division of all orders is completed. The similarity matrix model is as follows:

[0157]

[0158] Because the FA algorithm may concentrate in certain areas prematurely due to the attraction mechanism of the algorithm during the search process, resulting in the algorithm falling into a local optimal solution and being difficult to jump out, thus unable to refine the distinction of order batches, the FA algorithm needs to be optimized. The optimization steps are as follows:

[0159] (1) By changing the value of ξ, the performance and adaptability of the FA algorithm can be improved, which is used to evaluate and select similar order goods:

[0160]

[0161] Υ represents the distance between order goods, and Υ 0 represents the similarity of order goods, and β represents the distance between similar order goods;

[0162] By considering the similarity and difference between solutions, the FA algorithm can effectively explore and develop in the search space, and the calculation formula is as follows:

[0163] Mnew1 = M 1 +α×(M 2 -M 3 )+α×(M 3 -M 4 )

[0164] Mnew2 = Mnew1+α×(A best -B worst )

[0165] Mnew2 = Mnew1+α×(A best -B worst )

[0166] Assume that all fireflies have M best and M worst as the most similar and most different individuals in each iteration, and these fireflies are all optimized. M 1 , M 2 , M 3 , M 4 are the types of goods randomly selected in the order, where M 1 ≠M 2 ≠M 3 ≠M 4 , and the similarities of these 4 types of goods are compared pairwise. The value range of the random parameter α is [0,1]. Mnew1 and Mnew2 represent the types of goods orders that are the most similar after pairwise comparison, and M 5 represents the difference degree between the most similar and the most different orders, and the definition formula is as follows:

[0167] M 1 ={M new1 if k 1 ≤k 2 M best if k 1 >k 2}

[0168] M 2 ={Mnew1 if k 3 ≤k 2 M j if k 3 >k 2}

[0169] M 3 ={M new1，i if k 4 ≤k 3 M j if k 4 >k 3}

[0170] M 4 =M new1 if k 5 ≤k 4 or k 5 >k 4

[0171] M 5 =δ×M worst +ψ×(M best -M worst )

[0172] Among them, k 1 , k 2 , k 3 , k 4 , k 5 , ψ, and δ are all random parameters, taking values in [0, 1]. Assume that the function value of the i-th firefly is greater than that of the optimal firefly. The i-th firefly will replace the position of the optimal firefly. Assume that the function value of the i-th firefly is less than that of the optimal firefly, then no replacement will occur.

[0173] The random parameter ξ in the optimal objective iteration equation can control the random search of the algorithm. Usually, the parameter ξ controls the range of values of each firefly performing random motion in [0, 1]. By improving the random parameter ξ, the global search and local search capabilities of the algorithm can be optimized, thereby realizing the adaptive control process of ξ. Modify the parameter ξ using the following formula and substitute it into the objective iteration equation:

[0174]

[0175] During the optimization process, the algorithm achieves a sufficient balance between local search and global search by changing the value of ξ. Among them, Iter is the firefly iteration number, and G max is the maximum iteration number.

[0176] (2) Change Υ 0 value

[0177] In the original equation, Υ 0 is a constant, and its value will not change throughout the algorithm. Now, a loop variable p is introduced to modify it:

[0178]

[0179] μ is a constant, and MaxIter is the maximum number of iterations. Substituting the modified Υ 0 into the mutual attraction equation gives:

[0180]

[0181] Modifying Υ 0 is to control the search method of the algorithm. In the initial stage, the algorithm conducts a global search to find the optimal value. Changing the value of Υ 0 can control the algorithm to focus on local search in the later stage of the search, thereby improving the convergence speed of the algorithm, which greatly improves the convergence speed of the algorithm. After controlling the parameters and improving the algorithm, the target iteration formula is:

[0182]

[0183] where Xi and Xj are the states of fireflies, is an exponential decay function, which is used to simulate the phenomenon that the similarity or attraction between solutions rapidly decreases as the distance increases.

[0184] The search step size is optimized and improved, and the improvement formula is as follows:

[0185] R = φ × e -(σ×g) / Marge

[0186] where φ is the step size control factor, σ is the exponential regulation factor, g is the current iteration number, and Marge is the maximum iteration number.

[0187] After optimizing the algorithm output, the optimal order batch planning Order best .

[0188] S3: The screening of the logistics department is carried out according to the operation efficiency of logistics. An LMDI model is created to evaluate the contribution degree created by each logistics department, and the optimal logistics department is selected according to the level of contribution degree. Optionally, the transportation cost is first advanced by the capital transfer station. After the handover is completed, the source factory repays the money to the capital transfer station. If a department has tight capital turnover, it can also apply for a loan from the capital transfer station, and the source factory needs to inject start-up funds into the capital transfer station to ensure the normal operation of the logistics platform.

[0189] Logistics operation efficiency (η a<1), Logistics operation efficiency (η a <1) generally refers to the efficiency of the logistics platform during the processes of goods transportation, storage, distribution, etc., including parameters such as time efficiency, cost efficiency, and service level. The formula is as follows:

[0190] η a = λ·η b + μ·η t + ρ·s

[0191] Where λ, μ, and ρ are weight factors, adjusted according to different business priorities; η b is the cost efficiency; η t is the time efficiency; s is the service level.

[0192] The steps to create the LMDI model are as follows:

[0193] (1) Decompose the logistics operation efficiency (η a ) which is the cost efficiency (η b ), time efficiency (η t ), and service level (s) through the LMDI model. The change of each factor is expressed as:

[0194] Δη b = η b1 - η b0 =(η b1 / η b0 ) - 1

[0195] Δη t = η t1 - η t0 =(η t1 / η t0 ) - 1

[0196] Δs = s t1 - s t0 =(s t1 / s t0 ) - 1

[0197] Where the subscripts t0 and t1 represent the initial and final states respectively.

[0198] (2) The total change of the LMDI model can be expressed as:

[0199] Δη a = LMDI(η b ) + LMDI(η t ) + LMDI(s)

[0200] (3) The LMDI model further decomposes each factor into:

[0201]

[0202] Here, Wη b , Wη t and Ws are the weights of cost efficiency, time efficiency, and service level respectively, and these weights can be adjusted according to different business priorities.

[0203] (4) The total change in the contribution degree of logistics operation efficiency can be expressed as:

[0204]

[0205] Select the logistics department with the highest total change in contribution degree as the cooperation partner, and let this logistics department screen the transport fleet.

[0206] S4: The screening of the logistics department's fleet is to optimize the logistics distribution mechanism through weather conditions, optimized delivery routes, and the number of orders in different batches, and then introduce these three parameters into the LMDI model for quantification to screen out the three transport fleets with the best contribution degree, and record in detail data such as the transport time, cost, and quantity of goods transported by the fleet.

[0207] Three key parameters are introduced into the logistics distribution mechanism, and their calculation formulas are as follows:

[0208] (1) Weather condition impact factor (W):

[0209]

[0210] Among them, D w is the number of days of transport delay caused by weather conditions; D max is the maximum number of delay days in historical data. The exponential function is used to standardize the weather impact to a value between 0 and 1, where 1 represents no delay and 0 represents the maximum delay.

[0211] (2) Route optimization factor (R):

[0212]

[0213] Among them, C original is the transport cost of the original route, and C optimized is the cost of the optimized route. This ratio measures the effect of route optimization, and the smaller the value, the better the optimization effect.

[0214] (3) Order batch optimization factor (B):

[0215]

[0216] Among them, Q i is the batch size of the i-th order; Q avg is the average batch size of all orders; n is the total number of orders.

[0217] (4) The calculation formula of the LMDI model is as follows:

[0218] ΔV tot = ∑ i ΔV xi

[0219] Among them, ΔV tot is the total change amount, and ΔV xi represents the change amounts of each factor.

[0220] (5) Quantify the impact factor of weather conditions:

[0221]

[0222] Among them, L(V T , V 0 ) is the logarithmic mean Divisia index, X T and X 0 represent the transportation efficiency and the original transportation efficiency after considering the weather conditions, respectively.

[0223] (6) Quantify the order batch optimization factor:

[0224]

[0225] Among them, V iT and V i0 represent the distribution route efficiencies before and after optimization, respectively, and x k,iT and x k,i0 represent the distribution route parameters before and after optimization, respectively.

[0226] (7) Quantify the order batch optimization factor:

[0227]

[0228] Among them, V bT and V b0 represent the resource allocation efficiencies before and after the optimization of the order numbers in different batches, respectively, and x k,bT and x k,b0 represent the order number parameters before and after optimization, respectively.

[0229] (8) Combining the LMDI model, the calculation formula for quantifying the change in contribution degree is as follows:

[0230] ΔV tot = ΔV w + ΔV r + ΔV b

[0231] Select the three with the highest contribution degrees as the final transportation fleet.

[0232] S5: Transport to the freight station for a preliminary review of mass-produced goods. Optionally, the goods that fail the review can be sent back to the original factory for rework. The goods that pass the review can be sent to the transfer review platform for a second review. After completion, the review list is sent to the shipping department, and the un-reviewed goods are sent back to the freight station through the logistics platform, and then the freight station performs the rework operation.

[0233] S6: Order the goods that have completed the review through the logistics platform. The logistics platform will send the goods orders to different destinations according to the customers' needs. The platform will record important information of these customers (such as customers' funds, transaction amounts, return rates, etc.) and incorporate it into personal credit records. When handing over to the factory, priority is given to customers with a higher personal credit index to facilitate the factory's ordering of raw materials to process orders and enhance the trust between the factory and the customers, thereby improving the logistics efficiency.

[0234] The formula for the personal credit index is as follows:

[0235] C s = w 1 ·C 1 + w 2 ·C 2 + w 3 ·C 3

[0236] Where C s is the credit score, C 1 is the funds score, C 2 is the transaction amount score, C 3 is the return rate score, w 1 、w 2 、w 3 are the weight factors of funds, transaction amount, and return rate respectively. These weight factors can be determined by methods such as expert scoring and historical data analysis to ensure that the relative importance of each factor in the credit score is reasonably reflected.

[0237] C 1 = log(1 + C f )

[0238] Where C f is the customer's available funds balance or historical cumulative funds inflow on the platform.

[0239]

[0240] Where C t is the customer's historical cumulative transaction amount on the platform, and C max is the maximum transaction amount among all customers on the platform.

[0241]

[0242] Among them, C r is the return rate of customers on the platform, and C' max is the maximum return rate among all customers on the platform.

[0243] Commodity orders are allocated in descending order of the credit investigation index.

[0244] The above description of the embodiments enables those skilled in the art to implement or use the present invention. Various modifications to the embodiments will be obvious to those skilled in the art. The general principles of the present invention can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention should not be limited to the embodiments shown herein, but should cover the widest scope that conforms to the principles and novel features disclosed in the present invention.

Claims

1. A method for constructing a digital logistics platform based on an intelligent model, characterized in that: The method comprises the following steps: S1 builds a credit platform with raw material sources, selects raw material sources and establishes production workshops in source factories; S2 builds a logistics platform, builds an intelligent optimization module based on the step-size optimization and improved FA algorithm, and batches the orders of the goods produced in the production workshop of the source factory; S3 evaluates the contribution of each logistics department through logistics operation efficiency based on the LMDI model, and selects the best logistics department for transportation; S4 introduces three parameters, namely weather conditions, optimized delivery routes, and the number of orders in different batches, and uses the LMDI model to quantitatively screen the logistics department's transport fleet and deliver the goods to the freight station for review; S5 customers order approved goods through the logistics platform, and the logistics platform will deliver the goods orders to different destinations based on customer needs. The platform establishes a personal credit index model to calculate customer credit and feeds the credit information back to the credit platform, thereby improving logistics efficiency by giving priority to orders from customers with high credit ratings.

2. The method for constructing a digital logistics platform based on an intelligent model according to claim 1 is characterized in that: The selection of the production workshop of the source factory described in S1 is based on the production capacity and production efficiency of the production workshop. The higher the production capacity and production efficiency, the higher the priority. The production capacity calculation formula of the production workshop of the source factory is as follows: Among them, M is the equipment production capacity, H e is the effective working time of the equipment during the planning period, t i The time quota for processing the product on the equipment, in hours / piece; The formula for calculating the production efficiency of the production workshop of the source factory is as follows: Among them, η is the production efficiency, p a is the actual output, t s is the standard working time, m a is the actual manpower, t w is the actual working time, t b is the production blocking time, t o For overtime working hours.

3. The method for constructing a digital logistics platform based on an intelligent model according to claim 1 is characterized in that: S2 describes the construction of an intelligent optimization module based on the improved FA algorithm based on step size optimization. The specific steps are as follows: The FA algorithm uses real number coding to generate the initial population. Each firefly represents a potential solution. The position represents the variable value and the brightness represents the fitness. During decoding, the solution space is converted into the actual problem solution space. The maximum number of orders is considered and the position decoding is performed through the objective function value. Individual brightness of FA algorithm The luminous brightness of the FA algorithm is designed to be the inverse of the total number of shelf handling times, and the optimal solution to the order batching problem is found: Define the distance as the number of different numbers in the same position. Assume that firefly Z1 = [1,3,1,3,4,1,2,5] and firefly Z2 = [1,4,3,3,4,1,3,5], then the distance is 3; The FA algorithm updates the position of fireflies by moving them toward the brightest firefly and updating inconsistent order numbers. At the same time, the brightest firefly will also move randomly to avoid falling into the local optimum. The algorithm terminates when the maximum number of iterations is reached or a better objective function value is found. If the conditions are met, the result is output, otherwise the algorithm continues to execute. Create an intelligent optimization module model, with order batch as the objective function, and the calculation formula is as follows: Among them, Num is the total number of times the shelves need to be moved in batches, N is the order batch, Y is the number of shelves, e ny Indicates whether order batch N needs to move shelf Y.

4. The method for constructing a digital logistics platform based on an intelligent model according to claim 3 is characterized in that: Different constraints need to be set for different attributes of orders so that orders of various types of goods can run normally. The constraints of the intelligent optimization module model are as follows: Among them, A is the order set; B1 is the order batch set; D is the shelf set; P is the number of storage positions of the turnover shelf at the picking station; X is the product; Z is the order; d zn Indicates whether order Z is divided into batch n; Count.y indicates the number of times shelf Y is moved. An order similarity matrix is ​​established. Each element in the similarity matrix represents the similarity value between two orders. The order corresponding to the largest matrix value is found, and an order batch set is constructed. Consider whether the constraint of the total number of storage locations is met. If the constraint is met, a new order is added; otherwise, the order is deleted and a new order batch set is established. Repeat the above operation until all orders are divided into batches.

5. The method for constructing a digital logistics platform based on an intelligent model according to claim 1 or 3, characterized in that: Based on the step size optimization to improve the FA algorithm to build an intelligent optimization module, the specific improvement steps are as follows: Changing the value of ξ can improve the performance and adaptability of the FA algorithm for evaluating and selecting similar order goods: Υ represents the distance between order goods, Υ0 represents the similarity of order goods, and β represents the distance between similar order goods; By considering the similarities and differences between solutions, the FA algorithm can effectively explore and develop in the search space. The calculation formula is as follows: Mnew1=M1+α×(M2-M3)+α×(M3-M4) Mnew2=Mnew1+α×(A best -B worst ) Assume that all fireflies are M in each iteration best and M worst are the most similar and most different individuals. These fireflies have been optimized. M1, M2, M3, and M4 are the types of goods randomly selected in the order, where M1≠M2≠M3≠M4, and the similarities of the four types of goods are compared two by two. The range of the random parameter α is [0,1]. Mnew1 and Mnew2 represent the most similar types of goods orders after pairwise comparison. M5 represents the difference between the most similar and most different orders. The definition formula is as follows: M1={M new1 if k1≤k2 M best if k1>k2} M2={M new1 if k3≤k2 M j if k3>k2} M3={M new1，i if k4≤k3 M j if k4>k3} M4=M new1 if k5≤k4 or k5>k4 M5=δ×M worst +ψ×(M best -M worst ) Among them, k1, k2, k3, k4, k5, ψ, δ are all random parameters with values ​​in [0,1]. If the function value of the i-th firefly is greater than the function value of the optimal firefly, the i-th firefly will replace the position of the optimal firefly. If the function value of the i-th firefly is less than the function value of the optimal firefly, no replacement will occur. The random parameter ξ in the optimal target iteration equation can control the random search of the algorithm. Use the modified parameter ξ of the following formula and substitute it into the target iteration equation: During the optimization process, the algorithm achieves an adequate balance between local search and global search by changing the value of ξ, where Iter is the number of firefly iterations and G max is the maximum number of iterations; Change the Y0 value In the original equation, Υ0 is a constant, and now the loop variable p is introduced to modify it: μ is a constant, MaxIter is the maximum number of iterations, and the modified Υ0 is inserted into the mutual attraction equation as follows: After modifying Υ0, adjusting the control parameters and improving the algorithm, the target iteration formula is: Among them, Xi, Xj are the states of fireflies, is an exponential decay function, which is used to simulate the phenomenon that the similarity or attraction between solutions decreases rapidly as the distance increases; The search step length is optimized and improved, and the improved formula is as follows: R=φ×e -(σ×g) / Marge Among them, φ is the step size control factor, σ is the exponential control factor, g is the current number of iterations, Marge is the maximum number of iterations, and after the algorithm output is optimized, the optimal order batch planning Order is obtained. best .

6. The method for constructing a digital logistics platform based on an intelligent model according to claim 1, characterized in that: The logistics operation efficiency described in S3 is the efficiency of the logistics platform in the process of cargo transportation, storage and distribution, including time efficiency, cost efficiency and service level parameters. The formula is as follows: or a =l·h b +m·h t +ρ·s Among them, λ, μ, ρ are weight factors, which are adjusted according to different business priorities; η b is cost efficiency; t is time efficiency; s is service level; The steps to create the LMDI model are as follows: The logistics operation efficiency η a , cost efficiency η b , time efficiency η t The service level s is decomposed through the LMDI model, and the change of each factor is expressed as: See you later. b =the b1 -or b0 =(the b1 / or b0 )-1 See you later. t =the t1 -or t0 =(the t1 / or t0 )-1 Δs=s t1 -s t0 =(s t1 / s t0 )-1 Here, the subscripts t0 and t1 represent the initial and final states, respectively; The total change of the LMDI model can be expressed as: <h2 style=";text-align:left;direction:ltr">Δη<h2 style=";text-align:left;direction:ltr"> a <h2 style=";text-align:left;direction:ltr"> =LMDI(η<h2 style=";text-align:left;direction:ltr"> b <h2 style=";text-align:left;direction:ltr"> )+LMDI(η<h2 style=";text-align:left;direction:ltr"> t <h2 style=";text-align:left;direction:ltr"> )+LMDI(s) The LMDI model further decomposes each factor into: Among them, Wη b , Wη t and Ws are the weights of cost efficiency, time efficiency, and service level, respectively, which can be adjusted according to different business priorities; The total change in contribution to logistics operation efficiency can be expressed as: The logistics department with the highest total contribution change is selected as the cooperation partner, and the logistics department is responsible for selecting the transport fleet.

7. The method for constructing a digital logistics platform based on an intelligent model according to claim 1, characterized in that: The specific steps of S4 are as follows: The logistics allocation mechanism introduces three key parameters, and the calculation formula is as follows: Weather condition influencing factor W: Among them, D w The number of days of transportation delay caused by weather conditions; D max is the maximum delay day in the historical data, and an exponential function is used to normalize the weather impact to a value between 0 and 1, where 1 indicates no delay and 0 indicates the maximum delay; Route optimization factor R: Among them, C original is the transportation cost of the original route, C optimized is the cost of the optimized route. The ratio measures the effect of route optimization. The smaller the value, the better the optimization effect. Order batch optimization factor B: Among them, Q i is the batch size of the i-th order; Q avg is the average batch size of all orders; n is the total number of orders; The calculation formula of LMDI model is as follows: Where, ΔV tot is the total change, ΔV xi Indicates the amount of change of each factor; Quantify the factors affecting weather conditions: Among them, L(V T , V0) is the logarithmic mean Dirichlet index, X T and X0 represent the transportation efficiency after considering weather conditions and the original transportation efficiency respectively; Quantitative order batch optimization factors: Among them, V iT and V i0 Respectively represent the delivery route efficiency before and after optimization, x k,iT and x k,i0 Respectively represent the delivery route parameters before and after optimization; Quantitative order batch optimization factors: Among them, V bT and V b0 Respectively represent the resource allocation efficiency before and after optimization of different batches of orders, x k,bT and x k,b0 Respectively represent the order number parameters before and after optimization; Based on the LMDI model, the formula for calculating the change in quantitative contribution is as follows: ΔV tot =ΔV w +ΔV r +ΔV b The three with the highest contribution are selected as the final transport fleet.

8. The method for constructing a digital logistics platform based on an intelligent model according to claim 1, characterized in that: The personal credit index model described in S5 is as follows: C s =w1·C1+w2·C2+w3·C3 C1=log(1+C f ) Among them, C s is the credit score, C1 is the capital score, C2 is the transaction volume score, C3 is the return rate score, and C f It is the customer's fund flow details, C t is the customer's cumulative transaction amount, C max Maximum transaction amount, C r is the customer return rate, C' max is the maximum return rate, w1, w2, and w3 are the weight factors of funds, transaction amount, and return rate respectively. Product orders are allocated from high to low according to the personal credit index.

Images