Logistics distribution method based on path optimization

Abstract

The invention provides a logistics distribution method based on path optimization, and relates to the technical field of logistics intelligent scheduling. The logistics distribution method based on path optimization comprises the steps of dynamic information acquisition and preprocessing, multi-target collaborative optimization model construction, initial path optimization based on an improved genetic algorithm, local path adjustment triggered by a dynamic event, and path output and real-time updating. By constructing an integrated mechanism of dynamic information perception, multi-target collaborative optimization and real-time local adjustment, the problems of poor dynamic adaptability, insufficient multi-target collaboration and dynamic event response lag in the prior art are solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent logistics scheduling technology, specifically a logistics distribution method based on route optimization. Background Technology

[0002] With the rapid development of e-commerce and instant retail, the efficiency, cost, and customer satisfaction of logistics and distribution have become core competitive advantages in the industry. Route optimization, as a crucial link in logistics and distribution, currently suffers from the following shortcomings in existing technologies:

[0003] 1. Static optimization is the main approach, with poor dynamic adaptability: Existing route optimization methods are mostly based on historical data (such as fixed delivery points and preset traffic conditions) for one-time planning, which makes it difficult to respond to dynamic changes in real time, such as sudden traffic congestion, temporary addition or removal of delivery points (such as customers canceling orders temporarily or adding urgent delivery needs), and sudden weather changes (such as rain causing a decrease in road traffic efficiency). This results in a large deviation between the actual delivery route and the planned route, making it difficult to guarantee delivery time.

[0004] 2. Limitations of single-objective optimization and insufficient multi-objective collaboration: Existing technologies often focus on optimizing a single objective (such as the shortest distance or the shortest time) without fully integrating multi-dimensional objectives—such as transportation costs (including fuel consumption and vehicle wear and tear), customer satisfaction (such as whether the time window is met and the response speed to temporary demands), and resource utilization (such as vehicle loading rate). This results in optimization results that are "locally optimal" rather than "globally optimal."

[0005] 3. Low efficiency in responding to dynamic events: When dynamic events occur during the delivery process (such as the addition of new delivery points or the adjustment of the time window of the original delivery points), existing methods often adopt the strategy of "re-global optimization", which involves a large amount of computation and a slow response. This can easily lead to delivery delays, especially when there are a large number of delivery points. Summary of the Invention

[0006] (a) Technical problems to be solved

[0007] To address the shortcomings of existing technologies, this invention provides a logistics and distribution method based on path optimization. By constructing an integrated mechanism of dynamic information perception, multi-objective collaborative optimization, and real-time local adjustment, it solves the problems of poor dynamic adaptability, insufficient multi-objective collaboration, and delayed response to dynamic events in existing technologies.

[0008] (II) Technical Solution

[0009] To achieve the above objectives, the present invention provides the following technical solution: a logistics distribution method based on route optimization, comprising the following steps:

[0010] Step 1: Dynamic Information Collection and Preprocessing

[0011] A multi-source information acquisition module is constructed to acquire basic static data, real-time dynamic data and dynamic demand data related to delivery in real time. After removing outliers and quantifying parameters, the data is output as standardized data.

[0012] Step 2: Construct a multi-objective collaborative optimization model:

[0013] Based on the standardized data output from step one, a comprehensive objective function is established, which includes a cost objective function, a timeliness objective function, and a customer satisfaction objective function. The comprehensive objective function is adapted to different delivery scenarios through dynamic weights.

[0014] Step 3: Initial path optimization based on the improved genetic algorithm:

[0015] An improved genetic algorithm employing a hybrid "vehicle-delivery point" coding, adaptive crossover mutation, and elite retention strategy is used to solve the comprehensive objective function optimization model described in step two, and output the initial optimal delivery route that satisfies the constraints.

[0016] Step 4: Adjusting local paths triggered by dynamic events:

[0017] Real-time monitoring of the dynamic data in step one; when a dynamic event that meets the preset threshold is detected, local path adjustment is performed by insert-type local optimization or path segment rearrangement strategy according to the event type, and the adjusted path is verified by load and time window constraints until the path is feasible.

[0018] Step 5: Path Output and Real-time Updates

[0019] The initial optimal delivery route from step three or the feasible partially adjustable route from step four is pushed to the delivery vehicle terminal in real time. The order of delivery point visits, estimated arrival time, and real-time navigation route are also pushed simultaneously. When the route is adjusted, a change notification and the reason are pushed.

[0020] Preferably, in step one, the basic static data includes the location of the distribution center, the coordinates of the initial delivery point, the weight / volume of the goods, the vehicle type, load capacity, and fuel consumption parameters of the available delivery vehicles; the real-time dynamic data includes real-time traffic conditions (including road speed and congestion level) obtained through the traffic API and real-time weather parameters obtained through the meteorological platform; the dynamic demand data includes delivery point addition / cancellation instructions, time window adjustment information, and urgency level classification results; during preprocessing, the real-time weather parameters are quantified into road traffic efficiency correction coefficients, and the urgency level is classified into three levels: "urgent," "normal," and "delayable."

[0021] Preferably, in step two, the expression for the cost objective function is:

[0022]

[0023] Where, x ij 0-1 variable (x) ij =1 indicates that the vehicle travels from delivery point i to j, otherwise it is 0), d ij f is the real-time distance from i to j. v ρ is the fuel consumption cost per unit distance for the vehicle, ρ is the road traffic efficiency correction coefficient, and t is the fuel consumption cost per unit distance for the vehicle. ij Let c be the real-time travel time from i to j. t This refers to the cost of vehicle wear and tear per unit of time.

[0024] Preferably, in step two, the objective function expression for the timeliness is:

[0025]

[0026] Where m is the total number of delivery points, T arrive,k T is the actual time it takes for the vehicle to arrive at delivery point k. start,k ,T end,k Let λ be the start and end time of the time window for delivery point k. k μ is the penalty coefficient for delayed delivery. k To ensure early delivery of the penalty coefficient.

[0027] Preferably, in step two, the objective function expression for customer satisfaction is:

[0028]

[0029] Where m is the total number of delivery points, Δt k r is the deviation rate between the actual delivery time and the expected delivery time at delivery point k. k For dynamic demand response latency, α k β k These are the weighting coefficients.

[0030] Preferably, in step two, the expression for the comprehensive objective function is:

[0031] F = min(w) C ·C+w T ·Tw S ·S)

[0032] Among them, w C w T w S The target weight is dynamically adjusted based on the delivery scenario.

[0033] Preferably, in step three, the adaptive crossover and mutation of the improved genetic algorithm specifically involves: the crossover operation using partial mapping crossover (PMX), with the crossover probability adaptively decreasing from 0.8 to 0.5 as the number of generations increases; the mutation operation randomly exchanges or reverses the order of distribution points in the chromosomes, with the mutation probability being related to the population fitness, increasing the mutation probability of individuals with low fitness to 0.3; and the elite retention strategy is to retain the best 5% of individuals in each generation to directly enter the next generation.

[0034] Preferably, in step four, the preset threshold includes:

[0035] The straight-line distance between the new delivery point and the current route is ≤5km; the time window adjustment of the original delivery point is ≥30 minutes; and road congestion increases the travel time of the road segment by ≥50%.

[0036] The insertion-based local optimization is used to handle newly added nearby delivery points, inserting the new points into the path position that minimizes the increment of the comprehensive objective function;

[0037] The route reordering is used to handle time window adjustments or road congestion events, and only performs local re-optimization on route segments consisting of 5-8 affected delivery points.

[0038] Preferably, in step four, the constraint verification specifically includes:

[0039] Check whether the vehicle's load capacity exceeds the rated load capacity after adjustment and whether the arrival time of each delivery point is within the corresponding time window; if the constraints are not met, expand the local adjustment range to 2-3 adjacent delivery points and repeat the adjustment operation.

[0040] (III) Beneficial Effects

[0041] This invention provides a logistics distribution method based on route optimization, which has the following beneficial effects:

[0042] 1. By acquiring real-time data such as road conditions, weather, and dynamic demand through a multi-source information acquisition module, the limitations of existing "static planning" technologies are overcome, enabling route optimization results to be dynamically updated with the environment and reducing actual delivery deviations.

[0043] 2. Construct a comprehensive objective function that integrates cost, timeliness, and customer satisfaction, and adapt to different scenarios through dynamic weights to solve the problem of "single-objective local optimization" in existing technologies and achieve optimal overall benefits.

[0044] 3. For dynamic events, “local adjustment” is used instead of “global re-optimization”, which reduces the amount of calculation by 60-80% (e.g., in a scenario with 100 delivery points, local adjustment takes ≤10 seconds, while global re-optimization takes 60-90 seconds), greatly improving response speed and avoiding delivery delays. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the logistics and distribution method of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Example 1:

[0048] like Figure 1 As shown, this embodiment of the invention provides a logistics distribution method based on route optimization, including the following steps:

[0049] Step 1: Dynamic Information Collection and Preprocessing

[0050] Construct a multi-source information acquisition module to acquire and preprocess delivery-related dynamic data in real time, including:

[0051] 1.1 Basic static data: Location of distribution center, coordinates of initial delivery point and basic requirements (such as weight and volume of goods), information on available delivery vehicles (such as vehicle type, load capacity, and fuel consumption parameters);

[0052] 1.2 Real-time dynamic data: Obtain real-time road conditions (such as road speed and congestion level) within the delivery area through the traffic API interface, and obtain real-time weather parameters (such as rain / snow intensity, used to correct road traffic efficiency) through the meteorological platform;

[0053] 1.3 Dynamic Demand Data: Receive real-time dynamic demand changes at delivery points via client APP or logistics management system (such as adding / canceling delivery points, adjusting time windows (required delivery time periods), and marking urgent demands).

[0054] The collected data is preprocessed: outliers (such as erroneous coordinates) are removed, weather parameters are quantified into road traffic efficiency correction coefficients (e.g., the correction coefficient is set to 0.6 during heavy rain, meaning the actual traffic speed is 60% of the original speed), and dynamic demands are classified according to their urgency (e.g., "urgent", "normal", "delayable").

[0055] Step 2: Construct a multi-objective collaborative optimization model:

[0056] Based on the preprocessed data, a multi-objective optimization model is constructed with the goal of achieving "optimal overall benefits". The objective function includes:

[0057] 2.1 Cost Objective Function:

[0058]

[0059] Where, x ij 0-1 variable (x) ij =1 indicates that the vehicle travels from delivery point i to j, otherwise it is 0), d ij f is the real-time distance from i to j (based on real-time traffic correction). v ρ represents the fuel consumption cost per unit distance for the vehicle, ρ is the weather correction factor (taken from the preprocessing results of step one), and t is the fuel consumption cost per unit distance for the vehicle. ij Let c be the real-time travel time from i to j. t This refers to the cost of vehicle wear and tear per unit of time.

[0060] 2.2 Time-sensitive objective function:

[0061]

[0062] Where m is the total number of delivery points, t arrive，k T is the actual time it takes for the vehicle to arrive at delivery point k. start，k T end，k Let λ be the start and end time of the time window for delivery point k. k Penalty coefficient for delayed delivery (λ for urgent needs) k (3-5 times that of the usual), μ k To implement a penalty factor for early delivery (to avoid the customer being unable to receive the goods due to premature arrival).

[0063] 2.3 Customer satisfaction objective function:

[0064]

[0065] Where m is the total number of delivery points, Δt k Let r be the deviation rate between the actual delivery time and the expected delivery time at delivery point k. k α represents the dynamic requirement response latency rate (time from requirement change submission to system response / customer-acceptable latency). k β k Weighting coefficient (β of urgent needs) k Higher, emphasizing response speed.

[0066] 2.4 Comprehensive Objective Function:

[0067] F = min(w) C ·C+w T ·Tw S ·S)

[0068] Among them, w C w T w SThe target weight is dynamically adjusted based on the delivery scenario (such as during major e-commerce promotions). T Improvement, emphasizing timeliness; in cost-sensitive scenarios... C improve).

[0069] Step 3: Initial path optimization based on the improved genetic algorithm:

[0070] An improved genetic algorithm is used to solve the multi-objective optimization model in step two, generating an initial optimal delivery route, specifically including:

[0071] 3.1 Coding method: A hybrid "vehicle-delivery point" coding system is adopted, with a chromosome structure of [vehicle number, delivery point 1, delivery point 2, ..., delivery point n], where the order of delivery points represents the access order, realizing multi-vehicle collaborative path planning.

[0072] 3.2 Adaptive Crossover Mutation:

[0073] Crossover operation: Partial mapping crossover (PMX) is used, and the crossover probability is adaptively adjusted with the number of generations (initially 0.8, and later reduced to 0.5) to avoid premature convergence;

[0074] Mutation operation: Randomly exchange or reverse the order of delivery points in chromosomes. The mutation probability is related to the fitness of the population (the mutation probability of individuals with low fitness increases to 0.3, promoting population diversity).

[0075] 3.3 Selection Operation: An elite retention strategy is adopted, with the best 5% of individuals from each generation directly entering the next generation, and the rest selected through roulette wheel selection to ensure the transmission of superior genes.

[0076] 3.4 Termination condition: When the optimal solution has not improved for 20 consecutive generations, or when the number of iterations reaches a preset value (such as 100 generations), output the initial optimal path (including the order of delivery point visits for each vehicle and the estimated arrival time).

[0077] Step 4: Adjusting Local Paths Triggered by Dynamic Events

[0078] During delivery execution, the dynamic data from step one is monitored in real time. When a preset dynamic event occurs (such as the addition of a delivery point, a change in the time window of the original delivery point, or sudden road congestion), a local route adjustment mechanism is triggered, rather than a global re-optimization. Specifically, this includes:

[0079] 4.1 Dynamic Event Judgment: Set trigger thresholds, such as: the straight-line distance between the new delivery point and the current route is ≤5km (judged as "near-distance dynamic event"); the time window adjustment of the original delivery point is ≥30 minutes; road congestion causes the travel time of a certain road segment to increase by ≥50%.

[0080] 4.2 Local Adjustment Strategy:

[0081] If it is a "near-distance dynamic event" (such as adding a nearby delivery point): adopt "insertion-based local optimization" - calculate the distance and time between the new point and the adjacent delivery points in the current path, and insert it into the path at the position that minimizes the increment of the comprehensive objective function F (such as inserting it between two delivery points to minimize the sum of the new driving cost and the time penalty).

[0082] If it is a "time window / congestion event": adopt "path segment rearrangement" - extract the affected path segments (such as 3 points around the delivery point adjusted by the time window, or the path part involved in the congested road segment), and perform local re-optimization on the access order of the delivery points in the segment (based on the simplified algorithm of step 3, only optimize 5-8 points locally) to reduce the amount of computation.

[0083] 4.3 Feasibility verification: After adjustment, check whether the vehicle load capacity and time window still meet the constraints (e.g., whether the load capacity exceeds the limit after inserting a new point). If not, expand the local adjustment range (e.g., add 2 adjacent delivery points) until it is feasible.

[0084] Step 5: Path Output and Real-time Updates

[0085] The initial route generated in step three or the adjusted route in step four is pushed to the delivery vehicle terminal (such as the driver's APP) in real time through the logistics dispatch platform. This includes: the order of access to each delivery point, the estimated arrival time, and the real-time navigation route (combined with real-time traffic conditions). At the same time, when the adjustment is triggered in step four, a route change notification and the reason are pushed to the terminal (such as "Due to the addition of delivery point A, the route is adjusted to: B→A→C") to ensure that delivery execution is synchronized with planning.

[0086] Example 2:

[0087] To further illustrate the feasibility of Embodiment 1 of the present invention, the following examples are provided in specific scenarios:

[0088] Scenario: A city distribution center needs to deliver goods to 15 initial delivery points using 2 delivery vehicles (each with a load capacity of 5 tons). 3 of these delivery points are marked as "urgent need" (with a time window of 1 hour), and the initial time is 9:00 AM.

[0089] Step 1: Data collection includes: coordinates of the distribution center (116.4°E, 39.9°N), coordinates of 15 distribution points and cargo weight (all ≤ 1 ton), vehicle fuel consumption parameters (0.1 yuan / ton·km); real-time traffic conditions obtained through traffic API (the speed on a certain road section has been reduced to 40% of the original speed due to morning rush hour congestion); the weather is sunny (correction factor ρ = 1).

[0090] Step 2: Set the weight w C =0.3, w T=0.5 (due to emergency needs), w S =0.2; Delay penalty coefficient λ for urgent needs k =5 (normally 1), response latency weight β k =0.6 (the usual value is 0.3).

[0091] Step 3: Optimize using an improved genetic algorithm, with an initial population size of 50. After 80 iterations, convergence occurs, and the initial path is output:

[0092] Vehicle 1: Distribution Center → Delivery Point 1 → 3 → 5 → 7 → 9 → 11 → Distribution Center (estimated completion time: 10:30, no delay);

[0093] Vehicle 2: Distribution Center → Delivery Point 2 → 4 → 6 → 8 → 10 → 12 → 13 → 14 → 15 → Distribution Center (Expected to be completed at 11:00, with urgent needs at points 2, 6, and 10 all within the time window).

[0094] Step 4: At 9:30 AM, the system receives "New Delivery Point 16" (coordinates 3km away from delivery point 8 in the current route of vehicle 2, cargo weight 0.8 tons), which is determined to be a "near-range dynamic event"; triggering a local adjustment: calculate the minimum comprehensive target increment of the insertion position (delivery point 8→16→10) (new cost 2.3 yuan, time delay 0 minutes), and the load capacity of vehicle 2 becomes 4.8 tons (≤5 tons, feasible), and the adjusted route is pushed to the terminal of vehicle 2.

[0095] Step S5: Vehicle 2 followed the adjusted route and completed all deliveries at 11:05. The newly added delivery point 16 was delivered on time, with no urgent delays. The customer satisfaction score was 96 points (the traditional method would take about 70 seconds if the entire system were re-optimized, which could cause an emergency point 10 to be delayed by 10 minutes).

[0096] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A logistics distribution method based on route optimization, characterized in that: Includes the following steps: Step 1: Dynamic Information Collection and Preprocessing A multi-source information acquisition module is constructed to acquire basic static data, real-time dynamic data and dynamic demand data related to delivery in real time. After removing outliers and quantifying parameters, the data is output as standardized data. Step 2: Construct a multi-objective collaborative optimization model: Based on the standardized data output from step one, a comprehensive objective function is established, which includes a cost objective function, a timeliness objective function, and a customer satisfaction objective function. The comprehensive objective function is adapted to different delivery scenarios through dynamic weights. Step 3: Initial path optimization based on the improved genetic algorithm: An improved genetic algorithm employing a hybrid "vehicle-delivery point" coding, adaptive crossover mutation, and elite retention strategy is used to solve the comprehensive objective function optimization model described in step two, and output the initial optimal delivery route that satisfies the constraints. Step 4: Adjusting local paths triggered by dynamic events: Real-time monitoring of the dynamic data in step one; when a dynamic event that meets the preset threshold is detected, local path adjustment is performed by insert-type local optimization or path segment rearrangement strategy according to the event type, and the adjusted path is verified by load and time window constraints until the path is feasible. Step 5: Path Output and Real-time Updates The initial optimal delivery route from step three or the feasible partially adjustable route from step four is pushed to the delivery vehicle terminal in real time. The order of delivery point visits, estimated arrival time, and real-time navigation route are also pushed simultaneously. When the route is adjusted, a change notification and the reason are pushed.

2. The logistics distribution method based on route optimization according to claim 1, characterized in that: In step one, the basic static data includes the location of the distribution center, the coordinates of the initial delivery point, the weight / volume of the goods, the vehicle type, load capacity, and fuel consumption parameters of the available delivery vehicles; the real-time dynamic data includes real-time traffic conditions (including road speed and congestion level) obtained through the traffic API and real-time weather parameters obtained through the meteorological platform; the dynamic demand data includes delivery point addition / cancellation instructions, time window adjustment information, and urgency level classification results; during preprocessing, the real-time weather parameters are quantified into road traffic efficiency correction coefficients, and the urgency level is classified into three levels: "urgent," "normal," and "delayable." 3. The logistics distribution method based on route optimization according to claim 1, characterized in that: In step two, the expression for the cost objective function is: Where, x ij 0-1 variable (x) ij =1 indicates that the vehicle travels from delivery point i to j, otherwise it is 0), d ij f is the real-time distance from i to j. v ρ is the fuel consumption cost per unit distance for the vehicle, ρ is the road traffic efficiency correction coefficient, and t is the fuel consumption cost per unit distance for the vehicle. ij Let c be the real-time travel time from i to j. t This refers to the cost of vehicle wear and tear per unit of time.

4. The logistics distribution method based on route optimization according to claim 3, characterized in that: In step two, the objective function expression for the timeliness is: Where m is the total number of delivery points, t arrive，k T is the actual time it takes for the vehicle to arrive at delivery point k. start，k T end，k Let λ be the start and end time of the time window for delivery point k. k μ is the penalty coefficient for delayed delivery. k To ensure early delivery of the penalty coefficient.

5. The logistics distribution method based on route optimization according to claim 4, characterized in that: In step two, the objective function expression for customer satisfaction is: Where m is the total number of delivery points, Δt k r is the deviation rate between the actual delivery time and the expected delivery time at delivery point k. k For dynamic demand response latency, α k β k These are the weighting coefficients.

6. The logistics distribution method based on route optimization according to claim 5, characterized in that: In step two, the expression for the comprehensive objective function is: F=min(w C ·C+w T ·T-w S ·S) Among them, w C w T w S The target weight is dynamically adjusted based on the delivery scenario.

7. The logistics distribution method based on route optimization according to claim 1, characterized in that: In step three, the adaptive crossover and mutation of the improved genetic algorithm are as follows: the crossover operation adopts partial mapping crossover, and the crossover probability adaptively decreases from 0.8 to 0.5 with the number of generations; the mutation operation randomly exchanges or reverses the order of distribution points in the chromosome, and the mutation probability is related to the fitness of the population, with the mutation probability of individuals with low fitness increasing to 0.3; the elite retention strategy is to retain the best 5% of individuals in each generation to directly enter the next generation.

8. The logistics distribution method based on route optimization according to claim 1, characterized in that: In step four, the preset threshold includes: The straight-line distance between the new delivery point and the current route is ≤5km; the time window adjustment of the original delivery point is ≥30 minutes; and road congestion increases the travel time of the road segment by ≥50%. The insertion-based local optimization is used to handle newly added nearby delivery points, inserting the new points into the path position that minimizes the increment of the comprehensive objective function; The route reordering is used to handle time window adjustments or road congestion events, and only performs local re-optimization on route segments consisting of 5-8 affected delivery points.

9. A logistics distribution method based on route optimization according to claim 8, characterized in that: In step four, the constraint verification specifically includes: Check whether the vehicle's load capacity exceeds the rated load capacity after adjustment and whether the arrival time of each delivery point is within the corresponding time window; if the constraints are not met, expand the local adjustment range to 2-3 adjacent delivery points and repeat the adjustment operation.

Images