Route planning methods, electronic devices, and storage media for dynamic vehicle delivery tasks
By predicting customer locations using K-MEANS clustering and Poisson distribution, and combining this with a tilted variable neighborhood descent search algorithm, the path planning for dynamic vehicle delivery tasks is optimized. This solves the problem of path planning depending on customer order and improves path quality and efficiency.
Patent Information
- Application Number
- CN202310032703.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-01-10
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-01-10
AI Technical Summary
In dynamic vehicle delivery tasks, existing technologies rely on the order in which customers appear, resulting in suboptimal path quality. In particular, in large-scale tasks, the convergence speed is slow and it is easy to get trapped in local optima.
By acquiring prior knowledge, K-MEANS clustering and Poisson distribution are used to predict customer locations, generating virtual warehouse demand. The skewed variable neighborhood descent search algorithm is then combined to optimize path planning.
It improves the adaptability and quality of route planning, enabling it to better cope with dynamic changes, reduce vehicle waiting time, and improve the efficiency and accuracy of route planning.
Smart Images

Figure CN115936568B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of vehicle route optimization, specifically a route planning method, electronic device, and storage medium for dynamic vehicle delivery tasks. Background Technology
[0002] Distribution and transportation are crucial links in the logistics system. Choosing a reasonable distribution route can reduce transportation distance and time, lower fuel consumption, and has significant economic and environmental benefits.
[0003] Path planning methods for dynamic delivery tasks can be mainly divided into two categories: exact algorithms and metaheuristic algorithms. Exact algorithms include branch and bound, integer linear programming, and dynamic programming. These exact methods are only suitable for small-scale dynamic delivery tasks because applying them to large-scale vehicle delivery tasks is very time-consuming, and they cannot find the optimal solution within a reasonable timeframe. Metaheuristic algorithms include ant colony optimization (ACO), particle swarm optimization (PSO), genetic algorithms, and local search. ACO is inspired by the food-searching behavior of ants in nature, using a positive feedback mechanism to continuously search the target space and gradually approach the optimal solution. However, ACO has a slow convergence speed, requiring a long search time to obtain the optimal solution, and is prone to getting trapped in local optima when pheromones are scarce in the population. PSO originates from the study of the predatory behavior of flocks of birds and schools of fish, utilizing the information-sharing function of individuals within the group to obtain the optimal solution. However, PSO is prone to premature convergence and getting trapped in local optima. Genetic algorithms are search heuristic algorithms that mimic the process of natural selection. They utilize natural evolutionary techniques such as heredity, mutation, selection, and crossover to generate path planning schemes for the task at hand. Genetic algorithms have good global optimization capabilities, but also suffer from drawbacks such as low search efficiency leading to slow convergence speeds. Local search, as a general metaheuristic algorithm for solving combinatorial optimization problems, features fast convergence speed and good convergence performance, and is therefore widely used in route planning for vehicle delivery tasks.
[0004] However, there is a default approach to solving dynamic vehicle routing problems: dividing the entire workday into consecutive time slices and generating static vehicle routing problems, which are then solved using a solver. This approach can lead to the quality of some routes being highly dependent on the order in which customers arrive. If some customers located further away arrive too early, it can mislead the solution process, resulting in an unsatisfactory solution. Summary of the Invention
[0005] The present invention addresses the shortcomings of the prior art by proposing a path planning method, electronic device, and storage medium for dynamic vehicle delivery tasks. The aim is to predict the location distribution of customers based on existing prior customer knowledge and use this location distribution to guide the solution and execution processes, thereby making the resulting path planning scheme more adaptable to future dynamic changes.
[0006] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:
[0007] The method for route planning of dynamic vehicle delivery tasks according to the present invention is characterized by including the following steps:
[0008] Step 1: Obtain the total number of warehouse stations N, the total number of vehicles in the fleet K, the vehicle operating speed SP, and the maximum load capacity of the vehicles Q;
[0009] Get the location information of the warehouse site: Loc = {Loc1, Loc2, ..., Loc} i , ..., Loc N}, where Loc i This represents the location information of the i-th warehouse site;
[0010] Obtain the demand for warehouse sites D = {D1, D2, ..., D} i , ..., D N}, where D i This represents the demand for the i-th warehouse site;
[0011] Get the dwell time of the repository site DUT = {DUT1, DUT2, ..., DUT...} i ..., DUT N}, where DUT i This represents the demand for the i-th warehouse site;
[0012] Obtain the probability set P = {P1, P2, ..., P...} for the demand at each warehouse site. i , ..., P N}, where P i This represents the probability of demand occurring at the i-th warehouse site;
[0013] Obtain the set of time points where the demand occurs for each warehouse site, T = {T1, T2, ..., T...} i , ..., T N}, where T i D represents the demand for the i-th warehouse site. i The time point of occurrence; when T i >0 indicates that the demand D for the i-th warehouse site is 0. iAt time node T i When T appears i =0, indicating that the demand D of the i-th warehouse site is 0. i It already existed before the start of the workday, when T i <0 indicates that the demand D for the i-th warehouse site is 0. i It will not occur;
[0014] Define the information set of the warehouse site as DATA = {DATA1, DATA2, ..., DATA...} i , ..., DATA N}, where DATA i This represents the information of the i-th warehouse site, and DATA i ={i, LOC i T i D i DUT i};
[0015] The closing time of the delivery station is TW;
[0016] Define the number of time slices as NTS; define the time slice step size as... Define the sequence number of the current time slice as ts;
[0017] Define the set of warehouse requirements to be accessed in the ts-th time slice as LEF. ts ;
[0018] Define the set of warehouse requirements that have been executed at the end of the ts-th time slice as EXE. ts ;
[0019] Define AT as the set of vehicle running times after task allocation is performed in the ts-th time slice. ts ={AT ts 1, AT ts 2, ..., AT ts k , ..., AT ts K}, where AT ts k This represents the running time of the k-th vehicle in the ts-th time slice;
[0020] Step 2: Initialize ts = 1; Initialize the set of warehouse requirements to be accessed in the (ts-1)th time slice (LEF). ts-1 It is an empty set; initialize the set of warehouse requirements that have been executed at the end of the (ts-1)th time slice (EXE). ts-1 It is an empty set; initialize the set of vehicle running times AT after task allocation is performed in the (ts-1)th time slice. t s-1 All runtimes are TSL;
[0021] Step 3: Obtain the new warehouse requirements that appear in the ts-th time slice;
[0022] Step 3.1: Define the new warehouse demand that occurs in the ts-th time slice as NEW. ts and initialize NEW ts It is an empty set;
[0023] Step 3.2: Traverse the DATA:
[0024] If (ts-1)×TSL≤T i If ≤ ts × TSL, then the information DATA of the i-th warehouse site will be stored. i Add to set NEW ts Otherwise, continue iterating.
[0025] Step 4: Transfer LEF ts-1 Assign to LEF ts and NEW ts Elements in the set LEF are added to the set LEF. ts ;
[0026] Step 5: Based on the Poisson distribution, obtain the set of virtual warehouse demand locations SE for the ts-th time slice. ts :
[0027] Step 5.1: Define the set of warehouse sites where demand has already occurred at the end of the ts-th time slice as AP. ts and initialized as AP ts Empty set; define BP as the set of warehouse sites where no demand has occurred at the end of the ts-th time slice. ts and initialized to BP ts empty set;
[0028] Step 5.2: Traverse the DATA:
[0029] If T i If ≤ ts × TSL, then the information DATA of the i-th warehouse site will be stored. i Add to set AP ts Otherwise, the information DATA of the i-th warehouse site will be stored. i Add to set BP ts ;
[0030] Step 5.3: Perform probability correction on the probability set P to obtain the corrected probability set NP of the demand occurrence in the ts-th time slice. ts :
[0031] Step 5.3.1: Use equation (1) to estimate the demand quantity nl at the end of all time slices:
[0032] nl = floor(∑ i∈N P i (1)
[0033] In equation (1), floor represents the floor operation;
[0034] Step 5.3.2: Use equation (2) to obtain the corrected demand D for the i-th warehouse in the ts-th time slice. i The probability of occurrence NP i ts Thus, the probability set NP is obtained. ts ={NP1 ts NP2 ts , ..., NP i ts , ..., NP N ts};
[0035]
[0036] In equation (2), |AP ts | represents the set AP ts Number of Zhongcangshe sites, |BP ts | represents the set BP ts The number of Zhongcangshe stations;
[0037] Step 5.4: Set the number of clusters to NC, and use the K-MEANS clustering algorithm to cluster the warehouse site set BP for which no demand has been found. ts The warehouse sites in the data are clustered according to their location information, resulting in NC cluster sets (CLUs). ts ={CLU1 ts CLU2 ts , ..., CLU j ts ..., CLU NC ts}, where CLU j ts Let the j-th cluster be defined in time slice ts, and let the j-th cluster be defined by CLU. j ts The cluster center is CEN j ts ;
[0038] Step 5.5: Use equation (3) to obtain the j-th cluster CLU in the ts-th time slice. j ts Light-up weight PWj ts Thus, the lighting weight set PW under the ts-th time slice is obtained. ts ={PW1 ts PW2 ts ..., PW j ts ..., PW NC ts}:
[0039]
[0040] In equation (3), NP j,M ts This represents the corrected j-th cluster CLU. j ts The Mth warehouse site Cus j,M ts Demand D j,M The probability of its occurrence;
[0041] Step 5.6: Use equation (4) to obtain the j-th cluster CLU in the ts-th time slice. j ts Lighting probability PL j ts Thus, the lighting probability set PL under the ts-th time slice is obtained. ts ={PL1 ts PL2 ts , ..., PL j ts ,...PL NC ts}:
[0042]
[0043] Step 5.7: Utilize the lighting probability set PL ts Obtain the set of virtual warehouse required locations SE ts ;
[0044] Step 5.7.1: Initialize SE ts It is an empty set;
[0045] Step 5.7.2: Use a random number generator to generate a random number set R = {R1, R2, ..., R...} for the ts-th time slice. u , ..., R NC}; where R u Represents the u-th random number; initialize u = 1;
[0046] Step 5.7.3, if R u ≤PL uts Then the u-th cluster CLU will be... u ts Cluster center CENu t s added to SE ts Otherwise, assign u+1 to u and continue executing step 5.7.3 until j > NC, thus obtaining the final virtual warehouse demand location set SE. ts ; among which PL u ts PL represents the lighting probability of the u-th cluster in the ts-th time slice. j ts
[0047] Step Six: According to SE ts LEF ts Construct virtual warehouse requirement information;
[0048] Step 6.1, Calculate LEF ts The average demand of all warehouses in the region (BD) ts ;
[0049] Step 6.2: Define the information set (DSE) for the virtual repository site. ts ={DSE ts 1, DSE ts 2, ..., DSE ts j ..., DSE ts NC}, where DSE ts j This represents the information of the j-th virtual repository site, and LEF ts | represents the set LEF ts The number of elements in the table; TSE represents the appearance time of the virtual repository site and is set to 0; DUSE represents the service time of the virtual repository site and is set to 0;
[0050] Step 7: Apply the path solver RVNS based on tilted variable neighborhood descent search to the set LEF ts and DSE ts The warehouse site information is used to solve for the path planning scheme under the ts-th time slice. Among them, s r ts R represents the r-th path in the ts-th time slice. ts R represents the total number of paths in the ts-th time slice. ts <K; and LS r tsThis represents the r-th path s in the ts-th time slice. r ts The number of warehouses in c r,l ts Represents the r-th path s r ts The ID of the l-th visited warehouse;
[0051] Step 8: R in the convoy ts Each vehicle follows a path planning scheme s for the ts-th time slice. ts Delivery tasks are executed along each path within the system;
[0052] Step 8.1: Define a loop variable r and initialize r = 1;
[0053] Step 8.2, Define at r ts This represents the r-th path s in the ts-th time slice. r ts Find the travel time of the r-th vehicle and initialize at. r ts =AT ts-1 r ;AT ts-1 r This represents the running time of the r-th vehicle in the (ts-1)-th time slice;
[0054] Initialize l = 2;
[0055] Step 8.3, if c r,l ts ∈EXE ts-1 If the result is positive, proceed to step 8.6; otherwise, proceed to step 8.4.
[0056] Step 8.4, if c r,l ts >|LEF ts If the condition is met, proceed to step 8.7; otherwise, proceed to step 8.5.
[0057] Step 8.5, if and Then c r,l ts Add to collection EXE ts-1 and will Assigning to at r ts If so, proceed to step 8.6; otherwise, jump to step 8.7; where, Indicates the Cth r,l ts The service time of each warehouse, c r,l-1 tsRepresents the r-th path s r ts The (l-1)th accessed repository number; Indicates the cth r,u ts The demand for each warehouse;
[0058] Step 8.6: Assign the value of l+1 to l, return to step 8.3, and continue until l > LS. r ts until;
[0059] Step 8.7, at r ts Assign a value to AT ts r Assign the value of r+1 to r, then return to step 8.2 and execute sequentially until r > R. ts Up to; where AT ts r This represents the running time of the r-th vehicle in the ts-th time slice;
[0060] Step 8.8: Execute the EXE file. ts-1 Assign to EXE ts and LEF ts With EXE ts Public elements in LEF ts Delete;
[0061] Step 9: Assign the value of ts+1 to ts, and return to Step 3; continue until ts > NTS, thus obtaining the final path planning scheme s for the NTS-th time slice. NTS .
[0062] The present invention provides an electronic device, including a memory and a processor, wherein the memory is used to store a program that supports the processor in executing the method path planning, and the processor is configured to execute the program stored in the memory.
[0063] The present invention discloses a computer-readable storage medium on which a computer program is stored, wherein the computer program is executed by a processor to perform the steps of the path planning method.
[0064] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0065] 1. This invention employs a strategy for predicting scenarios using prior knowledge. With prior knowledge available, this strategy integrates and predicts this knowledge to simulate a scenario with better guidance. This allows the solver to consider the customer distribution throughout the workday during computation, providing guidance from previously unseen information to the solution and execution processes, ultimately leading to a better final solution.
[0066] 2. This invention obtains a series of cluster centers by performing K-MEANS clustering on customers not appearing on the map. It then sums all prior probabilities within each cluster and uses this sum as a parameter of a Poisson distribution to simulate the demand occurrence within that cluster. Next, it calculates the probability of "demand existence" to obtain the probability of each cluster center being activated. This probability is then compared with random numbers to selectively add the locations of cluster centers to the prediction scenario. In this way, the resulting prediction scenario can effectively simulate the occurrence of demand during weekdays. Furthermore, it guides the solution during the solver's process, enabling the solution to consider this unseen information and thus obtain a higher-quality path solution. Attached Figure Description
[0067] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation
[0068] In this embodiment, a path planning method for dynamic vehicle delivery tasks involves: first, performing K-MEANS clustering on warehouse location information not appearing on the map to obtain a series of cluster centers; then, summing all prior probabilities within each cluster and using this sum as a parameter of a Poisson distribution to simulate the demand occurrence within that cluster; second, calculating the probability of "demand existence" to obtain the probability of each cluster center being "lit up," and comparing this probability with random numbers to selectively add the locations of cluster centers to the prediction scenario. This results in a prediction scenario that effectively simulates demand occurrence during weekdays and guides the solution during the solver's process, ensuring that the solution takes into account this unseen information. Specifically, as... Figure 1 As shown, the method includes the following steps:
[0069] Step 1: Obtain the total number of warehouse stations N, the total number of vehicles in the fleet K, the vehicle operating speed SP, and the maximum load capacity of the vehicles Q;
[0070] Get the location information of the warehouse site: Loc = {Loc1, Loc2, ..., Loc} i , ..., Loc N}, where Loc iThis represents the location information of the i-th warehouse site;
[0071] Obtain the demand for warehouse sites D = {D1, D2, ..., D} i , ..., D N}, where D i This represents the demand for the i-th warehouse site;
[0072] Get the dwell time of the repository site DUT = {DUT1, DUT2, ..., DUT...} i ..., DUT N}, where DUT i This represents the demand at the l-th warehouse site;
[0073] Obtain the probability set P = {P1, P2, ..., P...} for the demand at each warehouse site. i , ..., P N}, where P i This represents the probability of demand occurring at the i-th warehouse site;
[0074] Obtain the set of time points where the demand occurs for each warehouse site, T = {T1, T2, ..., T...} i , ..., T N}, where T i D represents the demand for the i-th warehouse site. i The time point of occurrence; when T i >0 indicates that the demand D for the i-th warehouse site is 0. i At time node T i When T appears i =0, indicating that the demand D of the i-th warehouse site is 0. i It already existed before the start of the workday, when T i <0 indicates that the demand D for the i-th warehouse site is 0. i It will not occur;
[0075] Define the information set of the warehouse site as DATA = {DATA1, DATA2, ..., DATA...} i , ..., DATA N}, where DATA i This represents the information of the i-th warehouse site, and DATA i ={i, LOC i T i D i DUT i};
[0076] The closing time of the delivery station is TW;
[0077] Define the number of time slices as NTS; define the time slice step size as... Define the sequence number of the current time slice as ts;
[0078] Define the set of warehouse requirements to be accessed in the ts-th time slice as LEF. ts ;
[0079] Define the set of warehouse requirements that have been executed at the end of the ts-th time slice as EXE. ts ;
[0080] Define AT as the set of vehicle running times after task allocation is performed in the ts-th time slice. ts ={AT ts 1, AT ts 2, ..., AT ts k , ..., AT ts K}, where AT ts k This represents the running time of the k-th vehicle in the ts-th time slice;
[0081] Step 2: Initialize ts = 1; Initialize the set of warehouse requirements to be accessed in the (ts-1)th time slice (LEF). ts-1 It is an empty set; initialize the set of warehouse requirements that have been executed at the end of the (ts-1)th time slice (EXE). ts-1 It is an empty set; initialize the set of vehicle running times AT after task allocation is performed in the (ts-1)th time slice. ts-1 All runtimes are TSL;
[0082] Step 3: Obtain the new warehouse requirements that appear in the ts-th time slice;
[0083] Step 3.1: Define the new warehouse demand that occurs in the ts-th time slice as NEW. ts and initialize NEW ts It is an empty set;
[0084] Step 3.2: Traverse the DATA:
[0085] If (ts-1)×TSL≤T i If ≤ ts × TSL, then the information DATA of the i-th warehouse site will be stored. i Add to set NEW ts Otherwise, continue iterating.
[0086] Step 4: Transfer LEF ts-1 Assign to LEF ts and NEW ts Elements in the set LEF are added to the set LEF.ts ;
[0087] Step 5: Based on the Poisson distribution, obtain the set of virtual warehouse demand locations SE for the ts-th time slice. ts :
[0088] Step 5.1: Define the set of warehouse sites where demand has already occurred at the end of the ts-th time slice as AP. ts and initialized as AP ts Empty set; define BP as the set of warehouse sites where no demand has occurred at the end of the ts-th time slice. ts and initialized to BP ts empty set;
[0089] Step 5.2: Traverse the DATA:
[0090] If T i If ≤ ts × TSL, then the information DATA of the i-th warehouse site will be stored. i Add to set AP ts Otherwise, the information DATA of the i-th warehouse site will be stored. i Add to set BP ts ;
[0091] Step 5.3: Perform probability correction on the probability set P to obtain the corrected probability set NP of the demand occurrence in the ts-th time slice. ts :
[0092] Step 5.3.1: Use equation (1) to estimate the demand quantity nl at the end of all time slices:
[0093] nl = floor(∑ i∈N P i (1)
[0094] In equation (1), floor represents the floor operation;
[0095] Step 5.3.2: Use equation (2) to obtain the corrected demand D for the i-th warehouse in the ts-th time slice. i The probability of occurrence NP i ts Thus, the probability set NP is obtained. ts ={NP1 ts NP2 ts , ..., NP i ts , ..., NP N ts};
[0096]
[0097] In equation (2), |AP ts | represents the set AP ts Number of Zhongcangshe sites, |BP ts | represents the set BP ts The number of Zhongcangshe stations;
[0098] Step 5.4: Set the number of clusters to NC, and use the K-MEANS clustering algorithm to cluster the warehouse site set BP for which no demand has been found. ts The warehouse sites in the data are clustered according to their location information, resulting in NC cluster sets (CLUs). ts ={CLU1 ts CLU2 ts , ..., CLU j ts ..., CLU NC ts}, where CLU j ts Let the j-th cluster be defined in time slice ts, and let the j-th cluster be defined by CLU. j ts The cluster center is CEN j ts ;
[0099] Step 5.5: Use equation (3) to obtain the j-th cluster CLU in the ts-th time slice. j ts Light-up weight PW j ts Thus, the lighting weight set PW under the ts-th time slice is obtained. ts ={PW1 ts PW2 ts ..., PW j ts ..., PW NC ts}:
[0100]
[0101] In equation (3), NP j,M ts This represents the corrected j-th cluster CLU. j ts The Mth warehouse site Cus j,M ts Demand D j,M The probability of its occurrence;
[0102] Step 5.6: Use equation (4) to obtain the j-th cluster CLU in the ts-th time slice. j ts Lighting probability PLj ts Thus, the lighting probability set PL under the ts-th time slice is obtained. ts ={PL1 ts PL2 ts , ..., PL j ts ,...PL NC ts}:
[0103]
[0104] Step 5.7: Utilize the lighting probability set PL ts Obtain the set of virtual warehouse required locations SE ts ;
[0105] Step 5.7.1: Initialize SE ts It is an empty set;
[0106] Step 5.7.2: Use a random number generator to generate a random number set R = {R1, R2, ..., R...} for the ts-th time slice. u , ..., R NC}; where R u Represents the u-th random number; initialize u = 1;
[0107] Step 5.7.3, if R u ≤PL u ts Then the u-th cluster CLU will be... u ts Cluster center CENu t s added to SE ts Otherwise, assign u+1 to u and continue executing step 5.7.3 until j > NC, thus obtaining the final virtual warehouse demand location set SE. ts ; among which PL u ts PL represents the lighting probability of the u-th cluster in the ts-th time slice. j ts
[0108] Step Six: According to SE ts LEF ts Construct virtual warehouse requirement information;
[0109] Step 6.1, Calculate LEF ts The average demand of all warehouses in the region (BD) ts ;
[0110] Step 6.2: Define the information set (DSE) for the virtual repository site. ts ={DSEts 1, DSE ts 2, ..., DSE ts j ..., DSE ts NC}, where DSE ts j This represents the information of the j-th virtual repository site, and LEF ts | represents the set LEF ts The number of elements in the table; TSE represents the appearance time of the virtual repository site, set to 0; DUSE represents the service time of the virtual repository site, set to 0;
[0111] Step 7: Apply the path solver RVNS based on tilted variable neighborhood descent search to the set LEF ts and DSE ts The warehouse site information is used to solve for the path planning scheme under the ts-th time slice. Among them, s r ts R represents the r-th path in the ts-th time slice. ts R represents the total number of paths in the ts-th time slice. ts <K; and LS r ts This represents the r-th path s in the ts-th time slice. r ts The number of warehouses in c r,l ts Represents the r-th path s r ts The number of the l-th visited repository in the path s; for example, the path s r ts ={0, 1, 3, 5, 7, 0} indicates that in the ts-th time slice, the r-th path requires the corresponding delivery vehicle to visit the "delivery station", the 1st warehouse, the 3rd warehouse, the 5th warehouse, the 7th warehouse in sequence, and finally return to the "delivery station".
[0112] Step 8: R in the convoy ts Each vehicle follows a path planning scheme s for the ts-th time slice. ts Delivery tasks are executed along each path within the system;
[0113] Step 8.1: Define a loop variable r and initialize r = 1;
[0114] Step 8.2, Define at r ts This represents the r-th path s in the ts-th time slice.r ts Find the travel time of the r-th vehicle and initialize at. r ts =AT ts-1 r ;AT ts-1 r This represents the running time of the r-th vehicle in the (ts-1)-th time slice;
[0115] Initialize l = 2;
[0116] Step 8.3, if c r,l ts ∈EXE ts-1 If the warehouse's demand has been met, the algorithm should continue scanning and proceed to step 8.6; otherwise, if the warehouse's demand has not been met, the algorithm should consider whether to meet the warehouse's demand in this time slice and proceed to step 8.4.
[0117] Step 8.4, if c r,l ts >|LEF ts If the result is |, it means that the warehouse is a virtual warehouse in the predicted scenario. This indicates that new demand may arise in the vicinity when the next time slice arrives, so the vehicle should wait here until the next time slice. Therefore, the algorithm controls the vehicle to stop driving, and jumps to step 8.7; otherwise, it means that the warehouse is a customer in the real scenario, and we start to consider whether the warehouse meets the execution requirements, and execute step 8.5.
[0118] Step 8.5, if (The previous constraint states that if a vehicle's service to the current demand would exceed the current time slice, the vehicle will not serve the current warehouse. This consideration prevents a vehicle from skipping a time slice while serving a warehouse, thus missing information about some new warehouses.) Therefore, c... r,l ts Add to collection EXE ts-1 and will Assigning to at r ts If so, proceed to step 8.6; otherwise, jump to step 8.7; where, Indicates the cth r,l ts The service time of each warehouse, c r,l-1 ts Represents the r-th path s r ts The (l-1)th accessed repository number; Indicates the cth r,u ts The demand for each warehouse;
[0119] Step 8.6: Assign the value of l+1 to l, return to step 8.3, and continue until l > LS. r ts until;
[0120] Step 8.7, at r ts Assign a value to AT ts r Assign the value of r+1 to r, then return to step 8.2 and execute sequentially until r > R. ts Up to; where AT ts r This represents the running time of the r-th vehicle in the ts-th time slice;
[0121] Step 8.8: Execute the EXE file. ts-1 Assign to EXE ts and LEF ts With EXE ts Public elements in LEF ts Delete;
[0122] Step 9: Assign the value of ts+1 to ts, and return to Step 3; continue until ts > NTS, thus obtaining the final path planning scheme s for the NTS-th time slice. NTS .
[0123] In this embodiment, an electronic device includes a memory and a processor. The memory is used to store a program that supports the processor in executing the above-described method path planning, and the processor is configured to execute the program stored in the memory.
[0124] In this embodiment, a computer-readable storage medium stores a computer program, which is executed by a processor to perform the steps of the path planning method described above.
[0125] In summary, this invention uses the K-MEANS algorithm and Poisson distribution to predict scenarios, generating a predicted scenario based on the currently known scenario within each time slice. This predicted scenario increasingly approximates the actual scenario throughout the workday as time progresses. By performing path planning on the predicted scenario and implementing vehicle execution, the algorithm can control the vehicle to wait at appropriate locations, ensuring that the vehicle's execution process is always controlled by the overall scenario throughout the day, thereby guaranteeing the quality of the solution.
Claims
1. A route planning method for dynamic vehicle delivery tasks, characterized in that, Includes the following steps: Step 1: Obtain the total number of warehouse sites N Total number of vehicles in the fleet K、 vehicle speed SP Maximum load capacity of the vehicle Q ; Get warehouse site location information ,in, Indicates the first i Location information of each warehouse site; Demand for warehouse sites ,in, Indicates the first Demand for a warehouse site; Get the dwell time of the warehouse site ,in, Indicates the first The dwell time at each warehouse site; Obtain the probability set of demand occurrences for each warehouse site. ,in, Indicates the first The probability of demand occurring at a warehouse site; Obtain the set of time points when demand occurs at each warehouse site. ,in, Indicates the first i Demand for a warehouse site The time point of occurrence; when , indicating the first Demand for a warehouse site At the time point When , indicating the first Demand for a warehouse site ,when , indicating the first Demand for a warehouse site It will not occur; Define the information set of the warehouse site ,in, Indicates the first i Information about each warehouse site, and ; Get the closing time of the delivery station TW ; Define the number of time slices as NTS Define the time slice step size as Define the current time slice number as... ts ; Definition of the first ts The set of warehouse requests to be accessed in each time slice is: ; Definition of the first ts The set of warehouse requirements that have been executed by the end of each time slice is: ; Defined in the ts The set of vehicle running times after task allocation in each time slice is as follows: ={ ,in, Indicates the first k The car in ts The runtime corresponding to each time slice; Step 2: Initialization ts =1; initialize the first ts- The set of warehouse requests to be accessed in a time slice ; Initialize the first ts- The set of warehouse requirements that have been executed at the end of a time slice It is an empty set; initialization is performed at the 1st stage. ts- The set of vehicle running times after task allocation in one time slice All runtimes are ; Step 3: Obtain the information at the... ts New warehouse needs emerge during a specific time period; Step 3.1: Define in the first... ts The new warehouse demand that appears in each time slice is and initialize It is an empty set; Step 3.2: For DATA Perform the traversal: like Then the information of the i-th warehouse site will be... Add to collection Otherwise, continue iterating. Step 4: Assign to and will Adding elements from the set ; Step 5: Based on the Poisson distribution law, obtain the... ts Virtual warehouse location requirements for each time slice : Step 5.1: Define the first ts The collection of warehouse sites that have generated demand by the end of each time slice is as follows: and initialized to Empty set; Define the first ts The set of warehouse sites where no demand was observed at the end of each time slice is: and initialized to empty set; Step 5.2: For DATA Perform the traversal: like Then the first i Information on each warehouse site Add to collection Otherwise, the first i Warehouse site information Add to collection ; Step 5.3: For the probability set Perform probability correction to obtain the corrected first probability. ts Set of probabilities of demand occurring in a given time slice : Step 5.3.1: Use equation (1) to estimate the demand at the end of all time slices. nl : (1) In equation (1), This indicates a round-down operation; Step 5.3.2: Use equation (2) to obtain the corrected first... ts The first time slice Demand for each warehouse probability of occurrence Thus, the set of occurrence probabilities is obtained. ; (2) In equation (2), Represents a set The number of warehouse sites in China Represents a set The number of warehouse sites in China; Step 5.4: Set the number of clusters to [number]. NC The K-MEANS clustering algorithm was used to analyze the set of warehouse sites for which no demand was observed. The warehouse sites in the data are clustered according to their location information to obtain... NC Clusters ,in, Indicates the first ts The first time slice j Let the nth cluster be the i-th cluster. j Clusters The cluster centers are ; Step 5.5: Use equation (3) to obtain the first... ts The first time slice j Clusters Lighting weight Thus, the first ts The set of illumination weights for each time slice : (3) In equation (3), Indicates the corrected number j Clusters The Middle warehouse sites demand The probability of its occurrence; Step 5.6: Use equation (4) to obtain the first... ts The first time slice j Clusters Lighting probability Thus, the first ts Set of probabilities of lighting up in a given time slice : (4) Step 5.7: Utilize the lighting probability set Obtain the set of virtual warehouse locations ; Step 5.7.1, Initialization It is an empty set; Step 5.7.2: Use a random number generator to generate the first... ts A set of random numbers in a time slice ;in, Indicates the first u 1 random number; initialize u =1; Step 5.7.3, if Then the first u Clusters Cluster center Add to Otherwise, Assign to Then, continue with step 5.7.3 until... By reaching NC, the final set of virtual warehouse location requirements is obtained. ;in Indicates the first ts The first time slice u Lighting probability of each cluster ; Step Six: According to , Construct virtual warehouse requirement information; Step 6.1, Calculation Average demand of all warehouses ; Step 6.2: Define the information set of the virtual warehouse site. ={ },in, Indicates the first j Information about each virtual warehouse site, and ; Represents a set The number of elements in; Indicates the time when the virtual repository site appeared, and is set to 0; This indicates the service time of the virtual repository site and is set to 0. Step 7: Use the RVNS path solver based on tilted variable neighborhood descent search to solve the set. and The warehouse site information in the solution is used to obtain the first... ts Path planning scheme under a time slice ,in, Indicates the first ts The r-th path under time slice . Indicates the first ts The total number of paths per time slice. ;and , Indicates the first ts The first time slice r Path The number of warehouses in the country Indicates the first r Path The Middle l The warehouse number visited; Step 8: In the convoy The vehicles are respectively in accordance with the first ts Path planning scheme under a time slice Delivery tasks are executed along each path within the system; Step 8.1: Define a loop variable r and initialize r = 1; Step 8.2, Definition Indicates the first ts The r-th path in time slice r Find the travel time of the r-th vehicle and initialize it. ; Indicates the first r The car in ts- The runtime corresponding to one time slice; initialization ; Step 8.3, if If the result is positive, proceed to step 8.6; otherwise, proceed to step 8.
4. Step 8.4, if If so, proceed to step 8.7; otherwise, proceed to step 8.
5. Step 8.5, if and Then and will Assign to If so, proceed to step 8.6; otherwise, jump to step 8.7; where, Indicates the first Service hours for each warehouse Indicates the first r Path The Middle The accessed warehouse number; Indicates the first The demand for each warehouse; Step 8.6, The value assigned to Return to step 8.3 until... until; Step 8.7, Assign to Assign the value of r+1 to r, then return to step 8.2 and execute sequentially until... Up to; among which, Indicates the first r The car in ts The runtime corresponding to each time slice; Step 8.8, Assign to and will and Public elements in Delete; Step Nine: The value assigned to Return to step three; until Until then, thus obtaining the final number. Path planning scheme under time slice .
2. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports the processor in executing the path planning method of claim 1, and the processor is configured to execute the program stored in the memory.
3. A computer-readable storage medium storing a computer program, characterized in that, The computer program is executed by the processor to perform the steps of the path planning method of claim 1.
Citation Information
Patent Citations
Clustering method in wireless sensor network and recordable media thereof
KR1020080044468A
System and method for dynamic fleet management
US20180341918A1
