A method and system for vehicle path optimization in a single-lane channel under dynamic demand

Abstract

The application discloses a method and system for vehicle path optimization in a single-vehicle channel under dynamic demand, acquires a path point request and adds the path point request to a request queue; a vehicle path optimization model is constructed, and according to a current position of the vehicle and the dynamic request queue, an optimal order of the request queue is dynamically calculated within a certain time limit after the request is sent, a next request is dynamically selected to minimize total travel time, and the next path point is determined; a priority queue is divided from the request queue according to a time when the request is sent, for the priority queue, a current target, a next target and a second next target are determined in sequence; when the vehicle reaches the next target, the next target is updated to the current target, and the second next target is updated to the next target, and the process is sequentially and circularly promoted to form a processing order until the requests in the priority queue are all responded.

Description

Technical Field

[0001] This invention belongs to the field of vehicle routing optimization, specifically relating to a method and system for optimizing vehicle routes in a single-vehicle lane under dynamic demand. Background Technology

[0002] With technological advancements, the roboticization of industry is rapidly progressing, not only improving production efficiency but also driving the intelligent upgrading of workshop facilities. Consequently, raw material transportation methods, limited by workshop settings, are constantly iterating and innovating. Workshop raw material transportation requires attention to path optimization. Current solutions to path optimization problems mainly include genetic algorithms, ant colony algorithms, and tabu search, most of which are solutions under static conditions. However, in actual factory applications, static models cannot solve the problem; therefore, a solution for dynamic demand scenarios is needed.

[0003] In real-world industrial settings, workshops typically feature straight, single-vehicle aisles, allowing only one vehicle to pass at a time. Therefore, the required model is a single-vehicle path optimization scheme. Furthermore, during production, demand arises dynamically and unpredictably based on production conditions. Static models, however, require all demand to be known and all information to remain constant during model creation, which is unsuitable for industrial production scenarios. In dynamic scenarios, demand arises at varying times, and multiple demands may arise simultaneously. These dynamically changing demands pose significant challenges to path optimization, and current technologies struggle to find a solution that addresses all these considerations. Summary of the Invention

[0004] To address the shortcomings of existing technologies and achieve path optimization for single-lane vehicles based on dynamic demand, this invention adopts the following technical solution:

[0005] A method for optimizing vehicle routing in a single-vehicle lane under dynamic demand includes the following steps:

[0006] Step S1: Obtain the path point request and add it to the request queue;

[0007] Step S2: Construct a vehicle route optimization model. Based on the vehicle's current location and the dynamic request queue, dynamically calculate the optimal order of the request queue within a certain time limit after the request is issued. Select the next request with the minimum total travel time to determine the next waypoint.

[0008] Step S3: Divide the request queue into a priority queue based on the time the request was sent. For the priority queue, calculate the time required to travel from the current vehicle's target to each pathpoint that sent the request, and compare it with the remaining time limit of the request. The pathpoint with the smallest difference is designated as the next target. Calculate the time required to travel from each pathpoint that sent the request to the next target, and compare it with the remaining time limit of the corresponding request to obtain the minimum difference and its corresponding pathpoint. Designate the pathpoint that sent the request closest to the vehicle's direction of travel as the provisional next-next target. If the minimum difference is greater than the time required to travel from the next target to the provisional next-next target and then to the pathpoint corresponding to the minimum difference, then the provisional next-next target is designated as the next-next target. When the vehicle reaches the next target, update the next target to the current target, update the next-next target to the next target, and so on, forming a processing order, until all requests in the priority queue are responded to.

[0009] Furthermore, in step S2, the objective function of the vehicle routing optimization model is:

[0010]

[0011] Where S represents the set of requests waiting to be processed. This represents the number of times the vehicle moves from the path point location of the i-th request to the path point location of the j-th request. This represents the total time consumed from the path point location of the i-th request to the path point location of the j-th request.

[0012] Furthermore, in step S2, The expression form is:

[0013]

[0014] in, Indicates the first The path point location of each request. Indicates the first The path point location of each request This indicates the time taken for a vehicle to make a U-turn. express Is it in the direction of vehicle travel? Ahead, This represents the scaling factor used to measure the weight of distance.

[0015] Furthermore, in step S2, the constraints of the vehicle routing optimization model include:

[0016]

[0017]

[0018]

[0019] in, Let represent the time that the i-th request has been waiting for, and n represent the number of requests. Formula (1) means that the waiting time for each processing cannot exceed the time limit T. Formula (2) means that the number of times to arrive at and leave each waypoint is kept in balance. Formula (3) means that the vehicle starts from the initial parking point and eventually returns to the initial parking point.

[0020] Furthermore, in step S2, the decision variables of the vehicle routing optimization model for:

[0021] .

[0022] In this model, It is a constant, and the speed of the unmanned inspection vehicle is basically constant. The goal is to obtain the optimal order of the known request queue so that the total travel time is minimized and the requests are resolved within the time limit T when they are issued. Of course, due to the characteristics of the requests being generated at irregular times and in irregular quantities, there will be situations where they cannot be processed. This is a normal situation, belongs to the resource problem, is accepted by the factory, and is not a defect of this invention.

[0023] Furthermore, the priority queue division in step S3 involves adding requests that meet the following conditions to the priority queue:

[0024]

[0025] in, Indicates the length of a single-lane road. Indicates vehicle speed. This indicates the timeframe after the request is sent for it to be processed. This indicates the current waiting time for the i-th path point.

[0026] Furthermore, the response process of the priority queue in step S3 is as follows:

[0027] Calculate the time required to travel from the current vehicle's target location to each of the requesting path points. ;

[0028] calculate To obtain the minimum difference And use its corresponding path point as the next target. This indicates the timeframe after the request is sent for it to be processed. This indicates the current waiting time for the i-th request;

[0029] Calculate the time required to reach the next target from each of the requesting path points. ;

[0030] calculate Find the minimum difference and its corresponding path points, This indicates the current waiting time for the i-th request after the update;

[0031] Assuming the next target is the nearest pathpoint in the direction of travel that issued the request, if the minimum difference If the time required to reach the next target, the hypothetical next target, and then the path point corresponding to the minimum difference is greater than the time required for the next target, then the hypothetical next target will be determined as the next target.

[0032] Furthermore, in step S3, for the request queue, a cost function is used to select targets that need to be responded to first, thereby determining the response sequence. Each selection is only made for the request queue at the current moment. The expression of the cost function is as follows:

[0033]

[0034] in, This represents the total time consumed from the path point location of the i-th request to the path point location of the j-th request. This represents a scaling factor that adjusts the proportion of each consumption item. The base of the waiting time is used to measure its importance. This indicates the time already spent waiting, which increases as the waiting time grows, and covers the worker's operation time.

[0035] Furthermore, in step S3, The expression form is:

[0036]

[0037] in, Indicates the first The path point location of each request. Indicates the first The path point location of each request This indicates the time taken for a vehicle to make a U-turn. express Is it in the direction of vehicle travel? Ahead, This represents the scaling factor used to measure the weight of distance.

[0038] The complete expression of the cost function is as follows:

[0039] .

[0040] For hyperparameters in the cost function , Using regression analysis, an appropriate weighting is determined before scheduling begins. Specifically, the data generated in the production environment is cleaned, primarily handling missing and outlier values. Additional processing is applied to waiting times. During calculation, regression analysis is used... Instead, as a computational term, a linear regression model can be used for fitting, which may include mean squared error and coefficient of determination. These are used as model evaluation indicators to obtain specific numerical values.

[0041] A system for optimizing vehicle routes in a single-vehicle lane under dynamic demand includes a request queue generation module, a vehicle route optimization model, and a response module. According to the method for optimizing vehicle routes in a single-vehicle lane under dynamic demand, the system sequentially generates a request queue, constructs a vehicle route optimization model, divides a priority queue, and sorts and responds to requests in the priority queue.

[0042] The advantages and beneficial effects of this invention are as follows:

[0043] The present invention provides a method and system for optimizing vehicle routes in a single-vehicle lane under dynamic demand. By dividing the request priority of dynamic demand and implementing different processing schemes for requests within different priorities, while taking into account the waiting time and potential consumption of the requests, the efficiency of request processing is improved and the probability of requests not being responded to within the deadline is reduced, thus solving the problem of low request processing efficiency under constraints. Attached Figure Description

[0044] Figure 1 This is a schematic diagram illustrating the application scenario of vehicle route optimization in an embodiment of the present invention.

[0045] Figure 2 This is a flowchart of a method according to an embodiment of the present invention.

[0046] Figure 3 This is a flowchart of the request classification method in an embodiment of the present invention.

[0047] Figure 4 This is a priority queue in the embodiments of the present invention. A flowchart of the request handling methods within the document. Detailed Implementation

[0048] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0049] like Figure 1As shown, a typical application scenario includes a straight, single-lane road with a parking area at the end. Initially, vehicles depart from and return to the parking area. Production machines or other devices that can generate requests are located on both sides of the road. These machines can be distributed symmetrically or asymmetrically, or only on one side may there be machines or devices capable of issuing requests; the specific location of each request simply needs to be clearly marked. This embodiment of the invention only addresses single-lane traffic scenarios and involves only one schedulable resource.

[0050] like Figure 2 As shown, a method for optimizing vehicle routes in a single-vehicle lane under dynamic demand includes the following steps:

[0051] Step S1: The scheduling center receives the processing request from the production machine, including the production machine number that issued the request, and adds it to the pending request queue Q. All requests are initially in queue Q.

[0052] Step S2: Based on the current location of the unmanned inspection vehicle and the pending request queue Q, the dispatch center selects the next request to be processed according to the vehicle path optimization model established by the dispatch center and sends it to the unmanned inspection vehicle.

[0053] The objective function of the vehicle routing optimization model is:

[0054]

[0055] in, Let S represent the total time consumed from position i to position j, and let S represent the set of requests waiting to be processed. Let be the decision variable, representing the number of times the vehicle moves from position i to position j. The expression form is:

[0056]

[0057] in, Indicates the location j of the production machine that issued the request. Indicates the location i of the production machine that issued the request. This indicates the time taken for a vehicle to make a U-turn. It is a binary function, representing Is it in the direction of vehicle travel? In the foreground, k1 is a scaling factor used to measure the weight of distance.

[0058] The constraints to be constructed include:

[0059]

[0060]

[0061]

[0062] in, The value represents the time that the request for production machine location i has been waiting for, with an initial value of 0. Formula (1) indicates that the waiting time for each process cannot exceed the time limit T. Formula (2) indicates that the number of times each production machine arrives and leaves is balanced. Formula (3) indicates that the unmanned inspection vehicle departs from the parking lot and eventually returns to the parking lot.

[0063] Decision variables for:

[0064]

[0065] In this model, It is a constant, and the speed of the unmanned inspection vehicle is basically constant. The goal is to obtain the optimal order of the known request queue so that the total travel time is minimized and the requests are resolved within the time limit T when they are issued. Of course, due to the characteristics of the requests being generated at irregular times and in irregular quantities, there will be situations where they cannot be processed. This is a normal situation, belongs to the resource problem, is accepted by the factory, and is not a defect of this invention.

[0066] Step S3: Set priorities based on the time the requests were sent, increase the priority of some requests, and add them to the high-priority queue. Different solutions are adopted for different queues, and the path is optimized locally.

[0067] like Figure 3 As shown, the priority division method is as follows: for request i that meets the following conditions:

[0068]

[0069] For requests that meet the criteria, add them to the priority queue. For example, the length of a single-lane road is x = 200 meters, and the vehicle speed is v. vehicle =2 km / h, then The value is 6 minutes. If the request needs to be processed within T=40 minutes after it was sent, and the current waiting time is... When the time reaches 35 minutes, the item needs to be added to the priority waiting queue. When the time limit is 10 minutes, the priority remains unchanged.

[0070] Requests that do not meet the criteria still remain in queue Q. Processing priority higher than Q for priority queues The request processing adopts the LOOK algorithm under certain constraints. For requests in the regular queue Q, a cost function is given, and the request with the smallest cost function value in the waiting queue is selected as the next request to be processed.

[0071] like Figure 4 As shown, for priority queues The processing steps for the request are as follows:

[0072] First, for each request, calculate the time required to travel from the current target location to the production machine that issued the request. Compare the remaining waiting time of the request with this value to obtain the difference between the remaining request processing time and the time required for the current target to travel to the production machine that issued the request. The request with the smallest difference is identified as the next processing target.

[0073] Based on the determined next processing target, calculate the time required for each request to reach the next processing target. Similarly, calculate the difference between the remaining time for the request and the time required for the next processing target to reach the corresponding production machine for each request. Record the request corresponding to the smallest difference. If the smallest difference is... If the LOOK algorithm is allowed, then the LOOK algorithm will be executed, as follows:

[0074] Assuming the next target is the nearest request in the direction the car is traveling, if the minimum difference is... If the time required to travel from the next target to the provisionally set next-next target and then to the production machine position corresponding to the minimum difference is greater than the time required to reach the next production machine position, then the assumed target is designated as the next production machine target. This allows us to obtain the current target, the next target, and the next production machine target. When the trolley reaches the next target, the current target is updated to the next target, and the next target is updated to the next production machine target. This process is repeated cyclically to form a processing order, until the priority queue is reached. All requests were responded to.

[0075] Completed After a request in Q is received, a response is made to the request in Q. The order of the responses is determined by the cost function.

[0076] Specifically, if there are no requests in the priority queue, the regular queue is processed. This invention uses a cost function to select priority targets for processing, thereby determining the processing sequence. Each selection only considers the request queue at the current moment. The cost function is expressed as follows:

[0077]

[0078] in, Same function This represents the scaling factor, which adjusts the proportion of each consumption metric. This represents the base of the waiting time and measures its importance. This increases as waiting time grows, encompassing the worker's operation time.

[0079] pass From the given expression, we obtain the complete expression of the cost function as:

[0080]

[0081] As for how to obtain the hyperparameters in the cost function , Using regression analysis, an appropriate weighting is determined before scheduling begins. Specifically, the data generated in the production environment is cleaned, primarily handling missing and outlier values. Additional processing is applied to waiting times. During calculation, regression analysis is used... Instead, as a computational term, a linear regression model can be used for fitting, which may include mean squared error and coefficient of determination. These are used as model evaluation indicators to obtain specific numerical values.

[0082] A system for optimizing vehicle routes in a single-vehicle lane under dynamic demand includes a request queue generation module, a vehicle route optimization model, and a response module. Based on the aforementioned method for optimizing vehicle routes in a single-vehicle lane under dynamic demand, the system sequentially generates a request queue, constructs a vehicle route optimization model, divides a priority queue, and sorts and responds to requests in the priority queue.

[0083] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing vehicle routes in a single-vehicle lane under dynamic demand, characterized in that... Includes the following steps: Step S1: Obtain the path point request and add it to the request queue; Step S2: Construct a vehicle route optimization model. Based on the vehicle's current location and the dynamic request queue, dynamically calculate the optimal order of the request queue within a certain time limit after the request is issued. Select the next request with the minimum total travel time to determine the next waypoint. Step S3: Divide the request queue into a priority queue based on the time the request was sent. For the priority queue, calculate the time required from the current vehicle's target to each path point that sent the request, and compare it with the remaining time limit of the request. Select the path point with the smallest difference as the next target. Calculate the time required from each path point that sent the request to the next target, and compare it with the remaining time limit of the corresponding request to obtain the smallest difference and its corresponding path point. The nearest path point in the vehicle's direction of travel that issued the request is taken as the provisional next target. If the minimum difference is greater than the time required from the next target to the provisional next target and then to the path point corresponding to the minimum difference, then the provisional next target is determined as the next target. When a vehicle reaches the next target, the next target is updated to the current target, and the next target is updated to the next target. This process is repeated in a loop to form a processing order until all requests in the priority queue are responded to. The response process of the priority queue is as follows: Calculate the time required to travel from the current vehicle's target location to each of the requesting path points. ; calculate To obtain the minimum difference And use its corresponding path point as the next target. This indicates the timeframe after the request is sent for it to be processed. This indicates the current waiting time for the i-th request; Calculate the time required to reach the next target from each of the requesting path points. ; calculate Find the minimum difference and its corresponding path points, This indicates the current waiting time for the i-th request after the update; Assuming the next target is the nearest pathpoint in the direction of travel that issued the request, if the minimum difference If the time required to reach the next target, the hypothetical next target, and then the path point corresponding to the minimum difference is greater than the time required to reach the next target, then the hypothetical next target is determined as the next target. For the request queue, a cost function is used to select targets that need to be responded to first, thereby determining the response sequence. Each selection is made only for the request queue at the current moment. The expression of the cost function is as follows: in, This represents the total time consumed from the path point location of the i-th request to the path point location of the j-th request. This represents a scaling factor that adjusts the proportion of each consumption item. The base of the waiting time is used to measure its importance. This indicates the time already spent waiting, and the time increases as the waiting period grows.

2. The method for optimizing vehicle routes in a single-vehicle lane under dynamic demand as described in claim 1, characterized in that: In step S2, the objective function of the vehicle routing optimization model is: Where S represents the set of requests waiting to be processed. This represents the number of times the vehicle moves from the path point location of the i-th request to the path point location of the j-th request. This represents the total time consumed from the path point location of the i-th request to the path point location of the j-th request.

3. The method for optimizing vehicle routes in a single-vehicle lane under dynamic demand as described in claim 2, characterized in that: In step S2 The expression form is: in, Indicates the first The path point location of each request. Indicates the first The path point location of each request This indicates the time taken for a vehicle to make a U-turn. express Is it in the direction of vehicle travel? Ahead, This represents the scaling factor used to measure the weight of distance.

4. The method for optimizing vehicle routes in a single-vehicle lane under dynamic demand as described in claim 2, characterized in that: In step S2, the constraints of the vehicle routing optimization model include: in, Let represent the time that the i-th request has been waiting for, and n represent the number of requests. Formula (1) means that the waiting time for each processing cannot exceed the time limit T. Formula (2) means that the number of times to arrive at and leave each waypoint is kept in balance. Formula (3) means that the vehicle starts from the initial parking point and eventually returns to the initial parking point.

5. The method for optimizing vehicle routes in a single-vehicle lane under dynamic demand as described in claim 2, characterized in that: In step S2, the decision variables of the vehicle routing optimization model for: 。 6. The method for optimizing vehicle routes in a single-vehicle lane under dynamic demand as described in claim 1, characterized in that: The priority queue division in step S3 involves placing requests that meet the following conditions into the priority queue: in, Indicates the length of a single-lane road. Indicates vehicle speed. This indicates the timeframe after the request is sent for it to be processed. This indicates the current waiting time for the i-th request.

7. The method for optimizing vehicle routes in a single-vehicle lane under dynamic demand as described in claim 1, characterized in that: In step S3 The expression form is: in, Indicates the first The path point location of each request. Indicates the first The path point location of each request This indicates the time taken for a vehicle to make a U-turn. express Is it in the direction of vehicle travel? Ahead, This represents the scaling factor used to measure the weight of distance; The complete expression of the cost function is as follows: 。 8. A system for optimizing vehicle routes in a single-vehicle lane under dynamic demand, comprising a request queue generation module, a vehicle route optimization model, and a response module, characterized in that: According to claim 1, a method for optimizing vehicle routes in a single-vehicle lane under dynamic demand involves sequentially generating a request queue, constructing a vehicle route optimization model, dividing a priority queue, and sorting and responding to requests in the priority queue.

Images