Intelligent matching scheduling algorithm based on circulation and storage of different types of money boxes in financial industry

Abstract

The invention relates to an intelligent matching scheduling algorithm based on circulation and storage of different types of money boxes in the financial industry, and solves the technical problems of complexity response, cost control, operation efficiency, dynamic adaptability and the like. 2, constructing a cash box vehicle scheduling model based on hypotheses that the vehicle driving speed is constant, the cash box loading and unloading time is neglected, and each vehicle executes one scheduling task at a time; constructing a scheduling cost objective function and an operation efficiency objective function; step 3, carrying out first search; step 4, carrying out second search; 5, vehicle distribution is completed according to the priority; step 6, comparing a simulation test with an algorithm; according to the technical scheme, the problem is well solved, and the method can be used for dispatching different types of money boxes in the financial industry.

Description

Technical Field

[0001] The present invention relates to the field of, and specifically relates to an intelligent matching scheduling algorithm based on the transfer and storage of different types of cash boxes in the financial industry. Background Art

[0002] In the daily operation of the financial industry, the transfer and storage of different types of cash boxes are key links to ensure the normal development of business. Cash boxes contain various financial assets such as cash and important vouchers, and their safe and efficient scheduling is crucial. However, the current financial industry faces many challenges in cash box scheduling, which are specifically reflected in the following aspects: First, the scheduling complexity is high. The financial business involves many outlets and regions, and the transfer requirements of different types of cash boxes are complex and diverse. For example, the scheduling of cash boxes needs to strictly follow safety specifications and quantity management requirements, while important voucher cash boxes have special requirements for confidentiality and timeliness. The cash box requirements of each outlet vary significantly in terms of time, quantity, and type, which makes the cash box scheduling need to consider many factors comprehensively, increasing the difficulty and complexity of scheduling. Second, the problem of cost control. Traditional cash box scheduling methods often lack systematicness and scientificity, and do not fully consider various factors such as transportation costs and vehicle usage costs. On the one hand, unreasonable vehicle allocation leads to high empty load rates and circuitous transportation routes, increasing unnecessary transportation costs; on the other hand, vehicles are not reasonably configured according to the type of cash box and transportation needs, resulting in waste of vehicle resources and further increasing operating costs. While pursuing efficient scheduling, how to effectively control costs has become a major problem faced by financial enterprises. Third, the operating efficiency is low. With the continuous expansion of the financial business, the scale of cash box transfer is increasing day by day. The efficiency problem of the existing scheduling algorithms becomes more prominent when dealing with large-scale cash box scheduling. For example, the scheduling method based on simple priority rules is difficult to comprehensively consider the mutual influence of various factors when dealing with complex cash box scheduling tasks, resulting in an unoptimized scheduling plan, thus prolonging the transfer time of cash boxes and reducing the overall operating efficiency. This not only affects the normal development of financial business, but also may have a negative impact on customer service quality. Fourth, the lack of dynamic adaptability. The financial business scenario changes continuously with factors such as market environment and policies and regulations, and the cash box scheduling requirements also change dynamically accordingly. However, most current scheduling algorithms lack the dynamic adaptation ability to these changes and cannot adjust the scheduling strategy in a timely manner according to the actual situation. Once encountering sudden situations or temporary changes in business requirements, the existing scheduling algorithms may not be able to respond effectively quickly, resulting in scheduling chaos and affecting the stability and continuity of financial business.

[0003] In summary, the existing cash box scheduling methods in the financial industry have deficiencies in terms of complexity handling, cost control, operation efficiency, dynamic adaptability, vehicle congestion, etc., and it is difficult to meet the growing business needs of the financial industry. Therefore, there is an urgent need to develop an intelligent matching-based cash box scheduling algorithm to improve scheduling efficiency, reduce costs, enhance adaptability to complex and changing business scenarios, and reduce vehicle congestion. Summary of the Invention

[0004] The technical problem to be solved by the present invention is the technical problem of deficiencies in aspects such as complexity handling, cost control, operation efficiency, and dynamic adaptability existing in the prior art. A new intelligent matching scheduling algorithm based on the transfer and storage of different types of cash boxes in the financial industry is provided, and this intelligent matching scheduling algorithm based on the transfer and storage of different types of cash boxes in the financial industry has the characteristics of improving scheduling efficiency and reducing costs.

[0005] To solve the above technical problems, the following technical solutions are adopted:

[0006] An intelligent matching scheduling algorithm based on the transfer and storage of different types of cash boxes in the financial industry, comprising the following steps:

[0007] Step 1, problem analysis and data preparation: Collect and organize cash box data, including cash box number B id (id ∈ {1, 2,..., n}, n is the total number of cash boxes), cash box type weight destination area vehicle data, including vehicle number v j (j ∈ {1, 2,..., m}, m is the total number of vehicles), vehicle load limit and vehicle location area and area data, including area number A k (k ∈ {1, 2,..., p}, p is the total number of areas), inter-area distance matrix d(A i , A j );

[0008] Step 2, construct a cash box vehicle scheduling model: Based on assumptions such as constant vehicle driving speed, negligible cash box loading and unloading time, and each vehicle executing one scheduling task at a time;

[0009] Define decision variable x ij , if vehicle V j is assigned to transport cash box B i , then x ij = 1, otherwise x ij = 0 (i ∈ {1, 2,..., n}, j ∈ {1, 2,..., m});

[0010] Construct a scheduling cost objective function

[0011] Among them, C 1 is the unit distance transportation cost coefficient, is the distance from the area where vehicle V j is located to the destination of the cash box, and y j is the variable indicating whether vehicle V j participates in the scheduling, and C 2 is the fixed cost for the vehicle to participate in the scheduling; The operating efficiency objective function Among them, t ij is the time required for vehicle V j to transport the cash box B i ;

[0012] Set the vehicle load constraint and the cash box allocation constraint

[0013] Step 3, perform the first search: For each unallocated cash box B i , calculate the probability that each vehicle V j is selected to transport this cash box

[0014] According to the probability P ij , construct a roulette wheel, and make a selection on the roulette wheel through a random number to determine the vehicle V i that transports the cash box B j , and repeat this process until all cash boxes are allocated.

[0015] Step 4, perform the second search: For each unallocated cash box B i , traverse all vehicles V j , calculate the transportation cost

[0016] Select the vehicle V j with the minimum cost to transport the cash box B i , and repeat this process until all cash boxes are allocated;

[0017] Step 5, complete vehicle allocation according to the priority: Define the priority rule, set the priority of the emergency cash box as the highest, and determine the priority of the non-emergency cash box according to the distance from the vehicle to the cash box destination; First, process the emergency cash boxes, allocate vehicles according to the distance priority principle, and then process the non-emergency cash boxes, and also allocate vehicles according to the distance priority principle until all cash boxes are allocated;

[0018] Step 6, simulation experiment and algorithm comparison: Generate multiple groups of simulation data and set the simulation experiment parameters;

[0019] Taking scheduling cost and operating efficiency as algorithm performance evaluation indicators, run the first search, the second search, and the priority allocation vehicle under the same simulation data and parameter settings, record the scheduling costs and operating efficiencies of the corresponding algorithm combinations, and conduct a comparative analysis;

[0020] Scheduling algorithm combination and weight adjustment: Determine the advantages and disadvantages of each algorithm in different scenarios according to the simulation test results, and select the algorithm or algorithm combination with the best performance for different types of container scheduling tasks;

[0021] Define weights w 1 、w 2 、w 3 respectively represent the weights of the first search, the second search, and the priority allocation vehicle in the comprehensive scheduling, and w 1 +w 2 +w 3 = 1, and dynamically adjust the weights according to the advantages and disadvantages in different scenarios.

[0022] Working principle of the present invention: This algorithm first collects container, vehicle, and area data to provide basic information for scheduling. Then, a container vehicle scheduling model is constructed. Assuming a constant vehicle driving speed, etc., by defining decision variables, the scheduling cost and operating efficiency are used as the objective functions, and vehicle load and container allocation constraints are set. Next, three heuristic search algorithms are used: The first search calculates the vehicle selection probability according to the inverse of the transportation cost, and allocates containers by the roulette method; the second search selects the vehicle with the minimum cost for the container; the algorithm based on the priority rule packages the scheduling instructions into data frames and sends them sorted by priority and length. Then, simulation tests are conducted to compare the algorithm performance, and finally, the algorithm combination and weights are adjusted according to the test results to optimize the scheduling.

[0023] In the above solution, for optimization, further, the container type in the container data includes but is not limited to cash containers and important voucher containers.

[0024] Further, in constructing the container vehicle scheduling model, t in the operating efficiency objective function ij is calculated according to the vehicle driving speed and distance and obtained.

[0025] Further, in the first search, the calculation of the probability P ij is such that the vehicle with a lower transportation cost has a higher probability of being selected to transport the container.

[0026] Further, in the second search, the calculation of the transportation cost Cost ij comprehensively considers the unit distance transportation cost coefficient C 1 , the distance from the vehicle to the container destination and the fixed cost C of the vehicle participating in the scheduling2 。

[0027] Further, in the heuristic algorithm based on priority rules, the principle of prioritizing emergency cash boxes ensures that emergency cash boxes are preferentially allocated vehicles during the scheduling process.

[0028] Further, in the step of combining scheduling algorithms and adjusting weights, the dynamic adjustment of the weights w 1 、w 2 、w 3 is carried out based on the scheduling costs and operating efficiency performances of each algorithm in different scenarios during the simulation experiment.

[0029] Further, the fifth step includes:

[0030] 1. Data frame generation and attribute definition

[0031] 1.1 Wrapping the scheduling instructions into data frames, and wrapping each cash box - vehicle combination and its related scheduling information into a data frame F k , where k ∈ {1, 2, …, q}, and q is the total number of data frames;

[0032] 1.2 Defining data frame attributes

[0033] Data frame length L(F k ): Calculation method: L(F k ) = n k , where n k is the number of vehicles included in the k - th data frame; Meaning of representation: Characterizes the number of vehicles in the same batch of cash box - vehicle combinations;

[0034] Data frame transmission rate R(F k ):

[0035] Calculation method: Assume that the vehicles in the data frame F k are expected to take time t k to complete the scheduling tasks of m k cash boxes, then Meaning of representation: Characterizes the efficiency of the same batch of cash - vehicle combinations to complete scheduling and transfer, with the unit of task / unit time;

[0036] Data frame priority P(F k ): Calculation method: For emergency cash boxes (such as cash boxes that need to be scheduled within a short time), P(F k ) = 0.8; For ordinary cash boxes (such as important voucher cash boxes replenished regularly), P k = 0.3; Meaning of representation: Represents the priority of the data frame, with the value range of [0, 1], and the larger the value, the higher the priority;

[0037] 2. Data frame sorting

[0038] 2.1 Priority sorting. First, sort all the generated data frames in descending order according to the priority P(F k ), ensuring that data frames with higher priorities are processed before those with lower priorities;

[0039] 2.2 Processing of the same priority. For data frames F i and F j (i.e., P(F i ) = P(F j ))), sort them in ascending order according to the data frame length L(F k ), that is, first send the data frames with shorter lengths and then those with longer lengths; the sorting formula is expressed as:

[0040] F k < F l if (P(F k ) > P(F l )) or ((P(F k ) = P(F l )) and (L(F k ) < L(F l )))

[0041] 3. Sending of scheduling instructions

[0042] 3.1 Complete the generation of data frames and calculation of attributes. Traverse all combinations of cases and vehicles, generate a data frame F k for each combination, and calculate its L(F k ), R(F k ) and P(F k );

[0043] 3.2 Complete the sorting process. Use the sorting formula to sort the set of data frames {F k} according to the priority and length;

[0044] 3.3 Complete the sending of scheduling instructions. Send the data frames in sequence according to the sorted order to execute the scheduling task of the cases.

[0045] The preferred solution mainly involves the combination of scheduling algorithms and weight adjustment. First, through simulation experiments, the performance of the first search algorithm, the second search algorithm, and the priority allocation algorithm based on the roulette wheel is evaluated, and their scheduling costs and operating efficiency performances in different scenarios are recorded. According to the results of these experiments, the advantages and disadvantages of each algorithm in different scenarios are clarified. Then, according to the type of cash box scheduling tasks, such as urgent or non-urgent, cost-sensitive or not, etc., the most suitable algorithm or algorithm combination is selected. At the same time, corresponding weights are assigned to each algorithm, and the sum of the weights is 1, and they will be dynamically adjusted according to the actual situation to ensure the optimal cash box scheduling effect in different business scenarios and achieve the best balance between scheduling cost and operating efficiency. Description of the Drawings

[0046] The present invention will be further described below in conjunction with the drawings and embodiments.

[0047] Figure 1 , the schematic diagram of the algorithm flow in Embodiment 1. Detailed Embodiments

[0048] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention.

[0049] Embodiment 1

[0050] This embodiment provides an intelligent matching scheduling algorithm based on the transfer and storage of different types of cash boxes in the financial industry, such as Figure 1 , including the following steps:

[0051] Step 1, problem analysis and data preparation: Collect and sort out cash box data, including cash box number B id (id ∈ {1, 2,..., n}, n is the total number of cash boxes), cash box type weight destination area vehicle data, including vehicle number V j (j ∈ {1, 2,..., m}, m is the total number of vehicles), vehicle load limit and vehicle location area and area data, including area number A k (k ∈ {1, 2,..., p}, p is the total number of areas), distance matrix d(A i , A j ) between areas;

[0052] Step 2, construct a cash box vehicle scheduling model: Based on the assumptions that the vehicle driving speed is constant, the cash box loading and unloading time is ignored, and each vehicle executes one scheduling task at a time;

[0053] Define the decision variable x ij such that if vehicle V j is assigned to transport the money box B i , then x ij = 1; otherwise x ij = 0 (i ∈ {1, 2, …, n}, j ∈ {1, 2, …, m});

[0054] Construct the scheduling cost objective function

[0055] where C 1 is the unit - distance transportation cost coefficient, is the distance from the area where vehicle V j is located to the destination of the money box, y j is the variable indicating whether vehicle V j participates in the scheduling, and C 2 is the fixed cost for the vehicle to participate in the scheduling; Operating efficiency objective function where t ij is the time required for vehicle V j to transport the money box B i ;

[0056] Set the vehicle load constraint and the money box allocation constraint

[0057] Step 3: Conduct the first search: For each unassigned money box B i , calculate the probability that each vehicle V j is selected to transport this money box

[0058] According to the probability P ij , construct a roulette wheel and make a selection on the roulette wheel through a random number to determine the vehicle V i that transports the money box B j . Repeat this process until all money boxes are allocated.

[0059] Step 4: Conduct the second search: For each unassigned money box B i , traverse all vehicles V j , calculate the transportation cost

[0060] Select the vehicle V j with the minimum cost to transport the money box B i . Repeat this process until all money boxes are allocated;

[0061] Step 5: Complete vehicle allocation according to priorities: Define priority rules, set the highest priority for emergency cash boxes, and determine the priorities of non-emergency cash boxes based on the distance between the vehicle and the destination of the cash box. First, process the emergency cash boxes, allocate vehicles according to the principle of distance priority, then process the non-emergency cash boxes, and also allocate vehicles according to the principle of distance priority until all cash boxes are allocated.

[0062] Step 6: Simulation experiment and algorithm comparison: Generate multiple groups of simulation data and set simulation experiment parameters.

[0063] Take the scheduling cost and operation efficiency as the algorithm performance evaluation indicators. Run the first search, the second search, and the priority-based vehicle allocation under the same simulation data and parameter settings, record the scheduling costs and operation efficiencies of the corresponding algorithm combinations, and conduct comparative analysis.

[0064] Scheduling algorithm combination and weight adjustment: Determine the advantages and disadvantages of each algorithm in different scenarios according to the simulation experiment results, and select the algorithm or algorithm combination with the best performance for different types of cash box scheduling tasks.

[0065] Define weights w 1 、w 2 、w 3 to represent the weights of the first search, the second search, and the priority-based vehicle allocation in the comprehensive scheduling respectively, and w 1 +w 2 +w 3 = 1, and dynamically adjust the weights according to the advantages and disadvantages in different scenarios.

[0066] In this embodiment, first, collect cash box, vehicle, and regional data to provide basic information for scheduling. Then, construct a cash box vehicle scheduling model, assuming a constant vehicle driving speed, etc. By defining decision variables, use the scheduling cost and operation efficiency as the objective functions, and set vehicle load and cash box allocation constraints. Next, apply three heuristic search algorithms: In the first search, calculate the vehicle selection probability according to the inverse of the transportation cost, and allocate cash boxes by the roulette method; in the second search, select the vehicle with the minimum cost for the cash box; the algorithm based on priority rules packages the scheduling instructions into data frames and sends them sorted by priority and length. Then, conduct simulation experiments to compare the algorithm performances, and finally, adjust the algorithm combination and weights according to the experimental results to optimize the scheduling.

[0067] Preferably, the cash box types in the cash box data include but are not limited to cash boxes and important voucher boxes.

[0068] Preferably, in constructing the cash box vehicle scheduling model, t ij in the operation efficiency objective function is calculated based on the vehicle driving speed and distance.

[0069] Preferably, in the first search, the probability Pij The calculation is such that the higher the probability that a vehicle with lower transportation cost is selected to transport the money boxes.

[0070] Preferably, in the second search, the transportation cost Cost ij The calculation comprehensively considers the transportation cost coefficient C per unit distance 1 , the distance from the vehicle to the destination of the money box and the fixed cost C of the vehicle participating in the scheduling 2 .

[0071] Preferably, in the heuristic algorithm based on the priority rule, the principle of giving priority to emergency money boxes ensures that emergency money boxes are preferentially allocated vehicles during the scheduling process.

[0072] Preferably, in the step of combining scheduling algorithms and adjusting weights, the weights w 1 , w 2 , w 3 The dynamic adjustment is based on the scheduling costs and operation efficiency performances of each algorithm in different scenarios in the simulation experiment.

[0073] Preferably, the step five includes:

[0074] 1. Data frame generation and attribute definition

[0075] 1.1 Pack the scheduling instructions into data frames, and pack each combination of money box and vehicle and its related scheduling information into a data frame F k , where k ∈ {1, 2, …, q}, and q is the total number of data frames;

[0076] 1.2 Define the data frame attributes

[0077] Data frame length L(F k ): Calculation method: L(F k ) = n k , where n k is the number of vehicles included in the k-th data frame; Meaning: Characterize the number of vehicles in the same batch of money box and vehicle combinations;

[0078] Data frame transmission rate R(F k ):

[0079] Calculation method: Assume that the vehicles in the data frame F k are expected to complete the scheduling tasks of m k money boxes and require time t k , then Meaning: Characterize the efficiency of the same batch of money and vehicle combinations to complete scheduling and transfer, in units of tasks / unit time;

[0080] Data frame priority P(F k): Calculation method: For emergency cash boxes (such as cash boxes that need to be scheduled within a short period of time), P(F k ) = 0.8; For ordinary cash boxes (such as important voucher cash boxes that are replenished regularly), P k = 0.3; Meaning: Represents the priority of the data frame, with a value range of [0, 1]. The larger the value, the higher the priority;

[0081] 2. Data frame sorting

[0082] 2.1 Priority sorting. First, sort all generated data frames in descending order according to the priority P(F k ) to ensure that data frames with higher priorities are processed before those with lower priorities;

[0083] 2.2 Processing of the same priority. For data frames F i and F j (i.e., P(F i ) = P(F j ))), sort them in ascending order according to the data frame length L(F k ), that is, send the data frames with shorter lengths first and then those with longer lengths; The sorting formula is expressed as:

[0084] F k < F l if (P(F k ) > P(F l )) or ((P(F k ) = P(F l )) and (L(F k ) < L(F l )))

[0085] 3. Sending of scheduling instructions

[0086] 3.1 Complete data frame generation and attribute calculation. Traverse all combinations of cash boxes and vehicles, generate a data frame F k for each combination, and calculate its L(F k ), R(F k ) and P(F k );

[0087] 3.2 Complete sorting processing. Use the sorting formula to sort the set of data frames {F k} according to priority and length;

[0088] 3.3 Complete sending of scheduling instructions. Send the data frames in sequence according to the sorted order to execute the scheduling task of the cash boxes.

[0089] The preferred solution mainly involves the combination of scheduling algorithms and weight adjustment. First, through simulation experiments, the performance of the first search algorithm, the second search algorithm, and the priority allocation algorithm based on roulette is evaluated, and their scheduling costs and operation efficiency performances in different scenarios are recorded. According to the results of these experiments, the advantages and disadvantages of each algorithm in different scenarios are clarified. Then, according to the type of cash box scheduling tasks, such as urgent or non-urgent, cost-sensitive or not, the most suitable algorithm or algorithm combination is selected. At the same time, corresponding weights are assigned to each algorithm, and the sum of the weights is 1, and they will be dynamically adjusted according to the actual situation to ensure the optimal cash box scheduling effect in different business scenarios and achieve the best balance between scheduling cost and operation efficiency.

[0090] Specifically:

[0091] Suppose there is a large financial institution whose business covers multiple urban areas and has many bank branches. Every day, a large number of different types of cash boxes need to be transferred between different branches, including cash boxes, important voucher boxes, etc. The current scheduling method is inefficient and costly, so it is decided to use the intelligent matching scheduling algorithm of the present invention for optimization.

[0092] II. Data Preparation

[0093] Cash box data:

[0094] Cash box: Serial numbers B 1 , B 2 , B 3 , with weights of 50 kg, 60 kg, and 40 kg respectively, and the destinations are branch A, branch B, and branch C respectively.

[0095] Important voucher box: Serial numbers B 4 , B 5 , with weights of 20 kg and 30 kg respectively, and the destinations are branch D and branch E respectively. At the same time, mark the cash box as an urgent cash box and the important voucher box as an ordinary cash box.

[0096] Vehicle data:

[0097] Vehicle V 1 , with a load limit of 200 kg, located in area X.

[0098] Vehicle V 2 , with a load limit of 150 kg, located in area Y.

[0099] Area data:

[0100] It is known that branches A, B, C, D, and E are located in different areas A 1 , A 2 , A 3 , A 4 , A 5, the inter-region distance matrix is as follows:

[0101]

[0102] III. Construction of the vehicle scheduling model for cash boxes

[0103] Assumptions: Assume that the vehicle travels at a constant speed during the journey, without considering unexpected situations such as traffic congestion, and the loading and unloading time of the cash box can be ignored. Each vehicle can only execute one scheduling task at a time and returns to the initial area after completing the task.

[0104] Decision variable: Let x ij be the decision variable. If vehicle V j is assigned to transport cash box B i , then x ij = 1; otherwise x ij = 0.

[0105] Objective function:

[0106] Scheduling cost objective function: Assume that the unit distance transportation cost coefficient C 1 = 1, and the fixed cost C 2 = 100 for the vehicle to participate in the scheduling.

[0107] Operating efficiency objective function: Calculate based on the estimated driving speed and transportation time of each vehicle according to experience.

[0108] Constraints:

[0109] Vehicle load constraint: Ensure that the total weight of the cash boxes loaded on each vehicle does not exceed its load limit.

[0110] Cash box allocation constraint: Ensure that each cash box has and only has one vehicle responsible for transportation.

[0111] IV. Application of the heuristic search algorithm

[0112] Based on the first search algorithm: For cash box B 1 , calculate the selection probabilities of vehicles V 1 and V 2 . The distance from vehicle V 1 to network point A is 50, and the transportation cost is 50×1 + 100 = 150; the distance from vehicle V 2 to network point A is 90, and the transportation cost is 90×1 + 100 = 190.

[0113] Calculate the selection probability of V 1 The selection probability of V V 2 The selection probability of

[0114] Through roulette selection, assume that V is selected1 Transport B 1 Repeat this process to complete the allocation of all types of boxes.

[0115] Second search algorithm: For box B 2 , calculate the transportation cost of vehicle V 1 The transportation cost is 30×1 + 100 = 130, and for vehicle V 2 The transportation cost is 110×1 + 100 = 210.

[0116] Select the V with the minimum cost 1 Transport B 2 And so on to complete the allocation of all types of boxes.

[0117] Scheduling algorithm based on priority rules: Package the scheduling instructions for vehicle use as data frames. Set the data frame of the scheduling instructions corresponding to the cash box as high priority, and that corresponding to the important voucher box as low priority.

[0118] For data frames with the same priority, sort them according to the data frame length (i.e., the number of vehicles). For example, if the cash boxes B 1 and B 3 have the same priority, assuming the data frame length corresponding to B 1 is 1 (only one vehicle is needed for transportation), and the data frame length corresponding to B 3 is 2 (two vehicles are needed for transportation), then send the scheduling instruction data frame corresponding to B 1 first.

[0119] V. Simulation Experiments and Algorithm Comparison

[0120] Simulation experiment design: Use a computer simulation system to generate a large number of combinations of box, vehicle, and area data similar to the above, and set different experimental scenarios, such as different numbers of boxes, vehicles, and area distributions.

[0121] Algorithm performance evaluation metrics:

[0122] Scheduling cost: Calculate the scheduling cost of each algorithm under different simulation data. For example, when using the first search algorithm, count the total transportation cost of all boxes.

[0123] Operating efficiency: Record the total time required for each algorithm to complete the scheduling of all boxes to evaluate the operating efficiency.

[0124] Algorithm comparison and result analysis: Run the first, second, and priority rule-based scheduling algorithms respectively under the same simulation data and parameter settings. It is found that the first search algorithm is relatively balanced between cost and efficiency, the second search algorithm has a lower cost in some scenarios but is slightly less efficient, and the priority rule-based heuristic algorithm performs well in dealing with emergency boxes and can quickly respond to scheduling requirements.

[0125] VI. Scheduling Algorithm Combination and Weight Adjustment

[0126] Algorithm combination based on superiority and inferiority: According to the results of simulation tests, for the scheduling of emergency cash boxes, a heuristic algorithm based on priority rules is preferentially used to ensure efficiency; for non-emergency and cost-sensitive scheduling tasks, a first search algorithm is used. For example, in this instance, a heuristic algorithm based on priority rules is used for the scheduling of cash boxes, and a first search algorithm is used for the scheduling of important voucher boxes.

[0127] Weight adjustment: Define the weight w1 、 w 2 、w 3 represent the weights of the first search algorithm, the second search algorithm, and the scheduling algorithm based on priority rules in the comprehensive scheduling respectively, and w 1 +w 2 +w 3 = 1. In subsequent actual scheduling, according to the changes in different business scenarios, such as an increase in emergency cash boxes during business peaks, appropriately increase the weight w 3 of the heuristic algorithm based on priority rules to optimize the overall scheduling effect.

[0128] Although the above describes the illustrative specific embodiments of the present invention so that those skilled in the art of this technology can understand the present invention, the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art of this technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, all inventions and creations using the concept of the present invention are within the scope of protection.

Claims

1. An intelligent matching scheduling algorithm based on the circulation and storage of different types of money boxes in the financial industry, characterized by: The steps include: Step 1: Problem analysis and data preparation: Collect and organize the cash box data, including the cash box number B id (id∈{1, 2, ..., n}, n is the total number of boxes), box type weight Destination area Vehicle data, including vehicle number V j (j∈{1, 2, ..., m}, m is the total number of vehicles), vehicle load Limit, vehicle location area and regional data, including region number A k (k∈{1, 2, ..., p}, p is the total number of regions), the distance matrix between regions d(A i , A j ); Step 2: Construct a cash and box vehicle scheduling model: based on the assumptions that the vehicle speed is constant, the cash and box loading and unloading time is negligible, and each vehicle performs one scheduling task at a time; Define the decision variable x ij , if the vehicle V j Assigned shipping box B i , then x ij =1, otherwise x ij =0(i∈{1,2,…,n}, j∈{1,2,…,m}); Constructing the Scheduling Cost Objective Function Where C1 is the unit distance transportation cost coefficient, For vehicle V j The distance from the area to the destination of the cash box, y j For vehicle V j The variable of whether to participate in the dispatch, C2 is the fixed cost of the vehicle participating in the dispatch; the operating efficiency objective function where t ij For vehicle V j Transport Box B i Time required; Setting vehicle load constraints and cash box allocation constraints Step 3: Perform the first search: For each unallocated box B i , calculate each vehicle V j The probability of being selected to transport this box According to the probability P ij Construct a roulette wheel and select a random number on the roulette wheel to determine the transport box B i Vehicle V j , repeat this process until all the boxes are allocated. Step 4: Perform a second search: For each unallocated box B i , traverse all vehicles V j , calculate the shipping cost Select the vehicle V with the minimum cost j Transport Box B i , repeat this process until all the money boxes are allocated; Step 5: Complete vehicle allocation based on priority: Define priority rules, set the priority of emergency boxes to the highest, and prioritize non-emergency boxes based on the distance between the vehicle and the box destination; process emergency boxes first, and allocate vehicles based on the distance priority principle, then process non-emergency boxes, and allocate vehicles based on the distance priority principle until all boxes are allocated; Step 6: Comparison between simulation test and algorithm: Generate multiple sets of simulation data and set simulation test parameters; Taking dispatch cost and operation efficiency as algorithm performance evaluation indicators, the first search, second search and priority allocation of vehicles are run under the same simulation data and parameter settings, and the dispatch cost and operation efficiency of the corresponding algorithm combination are recorded and compared and analyzed; Step 7: Scheduling algorithm combination and weight adjustment: Determine the pros and cons of each algorithm in different scenarios based on the simulation test results, and select the algorithm or algorithm combination with the best performance for different types of box scheduling tasks; The weights w1, w2, and w3 are defined to represent the weights of the first search, second search, and priority allocation vehicles in the comprehensive dispatch, respectively, and w1+w2+w3=1. The weights are dynamically adjusted according to the merits in different scenarios.

2. According to claim 1, the intelligent matching scheduling algorithm for the circulation and storage of different types of money boxes in the financial industry is characterized in that: The cash box type in the cash box data Including but not limited to cash boxes and important voucher boxes.

3. According to claim 1, the intelligent matching and scheduling algorithm for the circulation and storage of different types of money boxes in the financial industry is characterized in that: In constructing the cash box vehicle dispatching model, the operating efficiency objective function is ij Based on vehicle speed and distance Calculated.

4. According to claim 1, the intelligent matching and scheduling algorithm for the circulation and storage of different types of money boxes in the financial industry is characterized in that: In the first search, the probability P ij The calculation makes it possible for vehicles with lower transportation costs to be selected to transport boxes.

5. According to claim 1, the intelligent matching scheduling algorithm for the circulation and storage of different types of money boxes in the financial industry is characterized in that: In the second search, the shipping cost is ij The calculation takes into account the unit distance transportation cost coefficient C1, the distance from the vehicle to the destination of the box, And the fixed cost C2 of vehicles participating in the dispatch.

6. According to claim 1, the intelligent matching scheduling algorithm for the circulation and storage of different types of money boxes in the financial industry is characterized in that: In the heuristic algorithm based on priority rules, the emergency money box priority principle ensures that the emergency money box has priority in vehicle allocation during the dispatch process.

7. According to claim 1, the intelligent matching scheduling algorithm for the circulation and storage of different types of money boxes in the financial industry is characterized in that: In the scheduling algorithm combination and weight adjustment step, the dynamic adjustment of the weights w1, w2, and w3 is performed based on the scheduling cost and operating efficiency performance of each algorithm in different scenarios in the simulation test.

8. According to claim 1, the intelligent matching scheduling algorithm for the circulation and storage of different types of money boxes in the financial industry is characterized in that: The step five comprises:

1. Data frame generation and attribute definition 1.1 The dispatch instruction is packaged into a data frame, and each box-car combination and its related dispatch information are packaged into a data frame F k , where k∈{1, 2, …, q}, q is the total number of data frames; 1.2 Define data frame attributes Data frame length L(F k ): Calculation method: L(F k )=n k , where n k is the number of vehicles contained in the kth data frame; meaning: represents the number of vehicles in the same batch of box truck combinations; Data frame transmission rate R(F k ): Calculation method: Assuming the data frame F k The vehicles in are expected to complete m k The scheduling task of a box takes time t k ,but Meaning: Indicates the efficiency of dispatching and circulating the same batch of vehicle combinations, in units of tasks / unit time; Data frame priority P(F k ): Calculation method: For emergency cash boxes (such as cash boxes that need to be dispatched in a short time), P(F k )=0.8; for ordinary cash boxes (such as important voucher cash boxes that are replenished regularly), P k =0.3; Meaning: indicates the priority of the data frame, the value range is [0, 1], the larger the value, the higher the priority; 2. Data frame sorting 2.1 Priority sorting, first, all generated data frames are sorted according to the priority P(F k ) are sorted in descending order to ensure that high-priority data frames are processed before low-priority data frames; 2.2 Same priority processing, for data frames with the same priority F i and F j (i.e. P(F i )=P(F j )), sort in ascending order according to the data frame length L(Fx), that is, send the shorter data frames first, and then send the longer data frames; the sorting formula is expressed as: F k <F l if(P(F k )>P(F l ))or((P(F k )=P(F l ))and(L(F k )<L(F l ))) 3. Sending dispatch instructions 3.1 Complete data frame generation and attribute calculation, traverse all combinations of box and vehicle, and generate a data frame F for each combination k , and calculate its L(F k )、R(F k ) and P(F k ); 3.2 Complete the sorting process and use the sorting formula to sort the data frame set {F k }Sort by priority and length; 3.3 Complete the sending of scheduling instructions and send the data frames in sequence according to the sorted order to execute the scheduling task of the cash box.

Images