A material collaborative distribution method for an electric fork truck assembly process

Abstract

The application provides a material collaborative distribution method for an electric forklift assembly process, relates to the technical field of intelligent logistics and production management, and is based on material demand information of an electric forklift assembly station, adopts a time period material consumption prediction model to obtain material demand prediction data of each station, combines material storage location information, adopts an improved ant colony algorithm to calculate collaborative distribution paths of multiple distribution vehicles, decomposes the collaborative distribution paths according to material distribution urgency, generates a material distribution task sequence, and adopts a bidirectional matching algorithm to adjust distribution tasks in a task interaction area. The application optimizes distribution path planning and task dynamic adjustment, fully utilizes task interaction area resources, and thus effectively solves the problems of insufficient prediction accuracy, poor path optimization effect and low distribution resource utilization in the prior art.

Description

Technical Field

[0001] This invention relates to the technical field of intelligent logistics and production management, and more specifically, to a method for collaborative material delivery during the assembly process of an electric forklift. Background Technology

[0002] With the rapid development of industrial automation and intelligence, the importance of logistics and distribution in the manufacturing process is becoming increasingly prominent, especially in the assembly of precision equipment such as electric forklifts. Material distribution efficiency directly affects the stability of the production cycle and the overall operational efficiency of the production line. Currently, traditional material distribution methods mainly rely on fixed-time, fixed-batch distribution strategies, which cannot respond to the dynamic material needs of each workstation on the production line in real time. This approach has revealed many shortcomings in the context of faster production pace and increased assembly complexity, such as material supply delays, waste of distribution resources, and excessive workstation storage pressure. To optimize distribution efficiency, in recent years, some scholars and enterprises have begun to research intelligent distribution systems based on material demand forecasting. By predicting the material needs of assembly workstations and combining this with path planning algorithms to optimize distribution routes, the timeliness and accuracy of distribution can be improved. However, existing research and practice generally suffer from problems such as single algorithm models, insufficient dynamic response capabilities, and a lack of collaborative scheduling mechanisms, making it difficult to adapt to complex assembly scenarios with multiple workstations and multiple materials.

[0003] Existing technologies still have several shortcomings in the field of collaborative material delivery. First, traditional material demand forecasting is mostly based on historical average consumption or static models, failing to fully consider the impact of cycle time at each workstation on the assembly line, material inventory status, and dynamic production load. This results in limited forecast accuracy and difficulty in adapting to real-time changes in production demands. Second, multi-delivery vehicle collaborative path planning typically employs classic ant colony or genetic algorithms. While these can optimize path length to some extent, they do not comprehensively consider multi-dimensional factors such as material urgency, task decomposition, and merging strategies, making them ineffective in handling complex operating conditions. Furthermore, existing technologies lack mechanisms for dynamically adjusting delivery tasks. When actual delivery demand deviates from the plan, delivery vehicle resources cannot be flexibly allocated, easily leading to material shortages at some workstations or wasted delivery resources. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention is proposed. This invention provides a material collaborative delivery method for the assembly process of electric forklifts, which can, to some extent, solve the problems of material delivery delays or over-delivery caused by the inability to accurately predict the dynamic material needs of each workstation, as well as the problem of low multi-vehicle collaborative efficiency caused by existing path planning algorithms not fully considering material urgency, delivery task decomposition and merging strategies.

[0005] According to one aspect of the present invention, a method for collaborative material delivery in the assembly process of an electric forklift is provided, comprising: Based on the material demand information of the electric forklift assembly station, a time-segmented material consumption prediction model is used to obtain the material demand prediction data for each station. Based on the material demand forecast data and combined with the material storage location information, an improved ant colony algorithm is used to calculate the collaborative delivery path of multiple delivery vehicles. In the collaborative delivery path, there are task interaction areas between the delivery vehicles. The collaborative delivery path is decomposed according to the urgency of material delivery to generate a material delivery task sequence; Based on the material delivery task sequence, a two-way matching algorithm is used to adjust the delivery tasks within the task interaction area.

[0006] Furthermore, the task interaction area is constructed based on material demand forecast data and warehouse location coordinate information, using an 8×8 grid-based collaborative delivery route planning model, determined by a dynamic labeling coefficient formula, expressed as:

[0007] in, Indicates delivery vehicle and The feasibility coefficient of task interaction between them Indicates the distance between the two workshops. The baseline distance for safe interaction between delivery vehicles. and These represent the current loads of the two vehicles, This refers to the maximum load capacity of the delivery vehicle.

[0008] Furthermore, the probability of delivery route selection calculated using the improved ant colony algorithm is expressed as follows:

[0009] in, Represents path ( , The concentration of pheromones on the surface. Represents path ( , Heuristic information values ​​on ) and These represent the pheromone importance factor and the heuristic information value importance factor, respectively. Indicates the first The set of allowed locations for an ant to choose from next. Indicates from workstation to candidate work station The actual pheromone concentration values ​​along the path between them. Indicates from workstation to candidate work station The actual heuristic information value of the path between them.

[0010] Furthermore, the calculation of the heuristic information value incorporates three dimensions: material urgency, delivery vehicle weight, and required time, expressed as follows: ,

[0011] in, Indicates position The material urgency coefficient, Indicates position arrive distance, This indicates the current load capacity of the delivery vehicle. Indicates the maximum load capacity of the delivery vehicle. Indicates position Material demand time, Indicates the time decay coefficient. For position Material requirements This represents the current inventory level. The safety stock threshold, This refers to the assembly cycle time for this workstation. This is the maximum assembly cycle time.

[0012] Furthermore, based on the improved ant colony algorithm, the optimal delivery route is output, including: the initial location of the delivery vehicle. Material storage locations and assembly station locations arranged in the order of access. ( =1,2,..., -1) and the path endpoint Each node has been marked with its coordinates in an 8×8 grid. , ), material loading and unloading information, arrival time requirements, estimated loading and unloading duration, and attributes of task interaction markers.

[0013] Furthermore, based on the optimal delivery route scheme, a complete delivery vehicle motion trajectory containing a smooth trajectory point sequence and time, speed, and acceleration information is generated by converting grid coordinates to actual coordinates, inserting trajectory sampling points, adding the time dimension, and optimizing with cubic spline interpolation.

[0014] Furthermore, material information of assembly stations is obtained, production time that can be supported by material inventory is calculated, and the material delivery urgency coefficient is obtained by comparing it with the remaining assembly time. The delivery priority is adjusted by comprehensively considering factors such as station distance and inventory level.

[0015] Furthermore, when the urgency coefficient of the workstation exceeds a preset threshold, it is determined to be an emergency delivery task, and its weight is calculated by combining the material value and the sharing coefficient. At the same time, the feasibility index of the task is calculated for workstations that do not reach the preset threshold. If the conditions are met, they are merged into one delivery batch.

[0016] Furthermore, the time window and comprehensive priority index of each delivery task are calculated, the task priority levels are divided, and the material delivery task sequence is generated by arranging them in descending order of priority index.

[0017] Furthermore, based on the material delivery task sequence and the collaborative delivery path of the delivery vehicles, a bipartite graph model is constructed in the task interaction area. The edge weights of the bipartite graph are dynamically adjusted and the Hungarian algorithm is used for optimal matching to generate a task list for each delivery vehicle.

[0018] Compared with existing technologies, this invention improves the accuracy of material demand forecasting at each workstation by employing a time-segmented material consumption prediction model, enabling real-time response to dynamically changing production needs. Combined with an improved ant colony algorithm, it optimizes the collaborative path planning of multiple delivery vehicles, fully considering material urgency, delivery task decomposition and merging strategies, achieving high efficiency and intelligence in path planning. The introduction of task interaction area design and a two-way matching algorithm dynamically adjusts delivery task allocation, improving multi-vehicle collaboration efficiency and reducing delivery resource waste. Overall, it significantly improves the accuracy and timeliness of material delivery, ensuring the smooth and efficient operation of the electric forklift assembly process. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart of a material collaborative delivery method for the assembly process of an electric forklift according to an embodiment of the present invention. Detailed Implementation

[0020] Hereinafter, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0021] As mentioned in the background section, existing technologies have three main problems: First, due to insufficient accuracy of material demand forecasting models, dynamic workstation demands, assembly cycle time, and inventory status are not fully considered, leading to material delivery delays or over-delivery. Second, existing path planning algorithms are simplistic and lack comprehensive consideration of material urgency, task decomposition, and merging strategies, making it difficult to achieve efficient collaborative delivery across multiple vehicles. Third, delivery task adjustment mechanisms are lacking, and the task interaction area is not fully utilized, making it impossible to flexibly allocate resources to cope with actual production changes.

[0022] Figure 1 This is a system block diagram of a material collaborative delivery method for the assembly process of an electric forklift according to an embodiment of the present invention. Figure 1 As shown, the material collaborative delivery method in the electric forklift assembly process includes: S1: Based on the material demand information of the electric forklift assembly station, a time-segmented material consumption prediction model is used to obtain the material demand prediction data for each station. The material demand prediction data includes the material code, the required quantity, and the required time.

[0023] Obtain the assembly cycle time and material assembly quota list for each workstation of the electric forklift assembly line. The material assembly quota list for each workstation includes four basic data items: material code, material name, material specifications and model, and quantity consumed by a single forklift. Based on historical assembly data of the assembly line, a complete work shift is divided into several time periods. The length of each time period is the same as the cycle time of the longest workstation on the assembly line. The actual material consumption data of each workstation within each time period is statistically analyzed. Based on the divided time periods, an input feature model for predicting material consumption by time period is constructed, which includes seven key data categories: actual material consumption quantity at each workstation, material inventory quantity, assembly cycle time, material delivery frequency, workstation capacity load rate, material storage location information, and material delivery path distance. Furthermore, when constructing the time-segmented prediction model for material consumption, a complete work shift of the electric forklift assembly line is divided into several time segments based on the longest assembly cycle time at each workstation. When each time segment is set to 30 minutes, 16 consecutive observation intervals are obtained. For each observation interval, the actual material consumption quantity, current material inventory, and assembly cycle time data for each workstation are obtained. In addition, the correlation data of the assembly progress and material supply status of the preceding and following processes at each workstation are collected, and the above data are used as the initial training samples. When an anomaly is found in the material consumption data of a certain time segment, the average material consumption of the three time segments before and after that time segment is used as the correction value. If the deviation between the corrected material consumption data and the material consumption of the adjacent time segments is within ±15%, the corrected data is included in the training sample set.

[0024] When constructing training samples using a sliding time window, six consecutive time periods are taken as the input time window each time, and the next two time periods are taken as the output time window. When the input window slides forward by one time period each time, the corresponding output window also slides synchronously, resulting in 128 training sample pairs. A prediction model is built using a Long Short-Term Memory (LSTM) network. The number of nodes in the input layer is set to 6, which is consistent with the length of the input time window. The hidden layer contains two LSTM units, with 64 units in the first layer and 32 units in the second layer. The dropout ratio between each layer is set to 0.2. The number of nodes in the output layer is set to 2, which corresponds to the length of the prediction time window. When it is detected that the predicted material consumption in a certain time period exceeds 80% of the current material inventory, the prediction weight coefficient of the adjacent time periods is increased to 1.2 times. The model is iteratively trained based on historical sample data. Training stops when the rate of change of the model loss function value is less than 0.1% after 50 consecutive iterations. If the average relative error between the prediction result and the actual material consumption exceeds 10%, the number of LSTM units in the first hidden layer is adjusted to 96 and the number in the second hidden layer is adjusted to 48. At the same time, the dropout ratio is adjusted to 0.25. The model is retrained until the average relative error between the prediction result and the actual material consumption is reduced to within 8%.

[0025] The output of the prediction model is combined with the material assembly quota list to generate material demand forecast data that includes material code, required quantity and required time. The time granularity of the material demand forecast data is consistent with the divided time period, and the required quantity is calculated by multiplying the predicted material consumption trend by the number of forklifts used.

[0026] S2: Based on the material demand forecast data and combined with the material storage location information, an improved ant colony algorithm is used to calculate the collaborative delivery path of multiple delivery vehicles. In the collaborative delivery path, there is a task interaction area between the delivery vehicles. Based on the acquired material demand forecast data and the warehouse location coordinates of each material, a collaborative delivery route planning model for delivery vehicles is constructed. First, the assembly workshop is divided into an 8×8 grid area, with each grid area having a side length of 10 meters. The locations of the material storage area, assembly workstations, and delivery vehicle parking positions are mapped onto the grid coordinate system. When the distance between any two delivery vehicles during delivery is less than 20 meters, this area is marked as a task interaction area. Within this area, delivery vehicles can exchange material delivery tasks. The dynamic marking coefficient for the task interaction area is calculated using the following formula:

[0027] in, Indicates delivery vehicle and The feasibility coefficient of task interaction between them Indicates the distance between the two workshops. The baseline distance for safe interaction between delivery vehicles. and These represent the current loads of the two vehicles, The maximum load capacity of the delivery vehicle; when When the value is ≥0.8, the area is marked as a valid task interaction area.

[0028] The delivery path was calculated based on an improved ant colony algorithm, with an initial pheromone concentration of 1.0 and a ant population of 50. When the... Only ants in the location Select the next access location When the path selection probability is calculated, it is expressed as:

[0029] in, The pheromone concentration along the path, For heuristic information values, and These are the pheromone importance factor and the heuristic information value importance factor, respectively. For the first The set of allowed locations for the next step of an ant to choose from.

[0030] Furthermore, the calculation of heuristic information values ​​incorporates three dimensions: material urgency, delivery vehicle weight, and demand time, expressed as: ,

[0031] in, Indicates position The material urgency coefficient, Indicates position arrive distance, Indicates the current load capacity of the delivery vehicle. Indicates the maximum load capacity of the delivery vehicle. Indicates position Material demand time, Indicates the time decay coefficient. For position Material requirements This represents the current inventory level. As the safety stock threshold, This refers to the assembly cycle time at this workstation. This is the maximum assembly cycle time.

[0032] It is important to note that at this time The initial material urgency coefficient, considering only the internal factors of the workstation, is used for heuristic information value calculation in the ant colony algorithm to guide initial path planning.

[0033] In each iteration, the path pheromone is dynamically updated, as expressed by the formula: ,

[0034] in, The pheromone evaporation coefficient, For pheromone increment, For the first Only ants on the path ( , The increase in pheromones left on the surface.

[0035] The optimal delivery route is output when the algorithm has iterated 500 times or the rate of change of the optimal route after 50 consecutive iterations is less than 0.5%, including the initial location of the delivery vehicle. Material storage locations and assembly station locations arranged in the order of access. ( =1,2,..., -1) and the path endpoint Each node has been marked with its coordinates in an 8×8 grid. , ), material loading and unloading information, arrival time requirements, estimated loading and unloading duration, and attributes of task interaction markers.

[0036] If the load factor of delivery vehicles on a certain delivery route exceeds 90%, then calculate the task redistribution factor:

[0037] in, For delivery vehicles The current load capacity, , This is the load warning threshold.

[0038] when When the value is ≥0.85, delivery tasks on the path that are close to the task interaction area are reassigned to delivery vehicles with a load rate of less than 60%.

[0039] The calculated delivery route is converted into a sequence of coordinates of the delivery vehicle's trajectory, with a sampling interval of 5 meters for each coordinate point.

[0040] Specifically, based on the calculated optimal delivery route, the grid coordinate positions ( , Multiplying by the grid side length of 10 meters gives the actual coordinate value. , ); Transition trajectory points are generated between adjacent locations. If two locations are on the same horizontal or vertical line, a trajectory sampling point is inserted every 5 meters starting from the starting point. If two locations are on different horizontal and vertical lines, the horizontal and vertical distances are compared to determine the turning point. If the horizontal distance is less than the vertical distance, the turning point is set at the intersection of the same horizontal line as the starting point and the same vertical line as the ending point. Otherwise, the turning point is set at the intersection of the same vertical line as the starting point and the same horizontal line as the ending point. After determining the turning point, trajectory sampling points are inserted at 5-meter intervals on the paths from the starting point to the turning point and from the turning point to the ending point. Further time dimension information is added to all trajectory points. The maximum operating speed of the delivery vehicle is initially set to 2 m / s and the acceleration to 0.5 m / s². When the delivery vehicle starts moving from a standstill or needs to stop, it adopts an acceleration and deceleration motion mode within 20 meters of the start and stop point. The timestamp of each point is calculated by the time increment calculation formula of the trajectory point in the acceleration segment and the kinematic formula of the deceleration segment. The uniform speed motion mode is adopted in other sections. When it reaches the material storage location point or the assembly station location point, the dwell time is determined according to the amount of material loaded and unloaded. The baseline time is initially set to 60 seconds for each 100 kg of material processed. Furthermore, the trajectory is smoothed and optimized by detecting turning points. A smooth turning trajectory is generated within a range of two trajectory points before and after each turning point using cubic spline interpolation. Simultaneously, the curvature change between adjacent trajectory points is monitored. When the rate of curvature change exceeds 0.2 / meter, the density of trajectory sampling points is increased to make the motion smoother. All trajectory points are smoothed using a five-point cubic spline fitting method to ensure the continuity of velocity and acceleration. After completing the coordinate calculation, time planning, and smoothing of all trajectory points, the trajectory data is organized in chronological order, outputting a complete trajectory coordinate sequence including trajectory point number, X-coordinate, Y-coordinate, timestamp, velocity, and acceleration.

[0041] S3: Decompose the collaborative delivery path according to the urgency of material delivery to generate a material delivery task sequence. The urgency of material delivery is calculated by the ratio of the workstation material inventory to the assembly cycle time.

[0042] Obtain real-time material information for each assembly station, including current material inventory levels. Material consumption rate Assembly cycle time Remaining quantity of product batches currently being processed For assembly stations Calculate the production time that its material inventory can support, expressed as:

[0043] in, The factor representing the impact of assembly cycle time at the workstation. This represents the maximum assembly cycle time across all workstations.

[0044] Compare this time with the remaining assembly time at the workstation Compare and calculate the urgency coefficient of material delivery:

[0045] in, For workstations The remaining quantity of products in the batch currently being processed. This is the distance impact coefficient, used to adjust for the influence of delivery distance on urgency. For workstation Distance to the nearest material storage point This represents the maximum delivery distance within the delivery system. This is the inventory depletion coefficient. This represents the maximum material storage capacity of the workstation.

[0046] It should be noted that the material delivery urgency coefficient at this time, compared to the initial material delivery urgency coefficient, takes into account the delivery distance from the workstation to the material storage point and the remaining assembly time at the workstation, and is used for priority ranking of delivery tasks and batch merging decisions.

[0047] When the urgency coefficient of the workstation Greater than the threshold When an emergency delivery task is identified, its weighting coefficient is calculated.

[0048] in, This is the material value impact coefficient. The unit value of the material. The highest unit value of materials; For material sharing coefficient, This refers to the quantity of this material that is needed by other workstations. This represents the total material requirements.

[0049] For workstations with an urgency coefficient between 0.8 and 1.2, calculate the feasibility index for merging tasks between workstations:

[0050] in, For workstation and The distance between them The maximum allowable distance to consider when merging tasks. For workstation and Common material requirements This indicates the urgency coefficient of material delivery between the two workstations. This is the threshold for the difference in urgency coefficients between two workstations. When the difference in urgency is too great, it indicates that merging is not suitable. For the maximum batch material quantity, This is the batch effect coefficient.

[0051] when In this case, the delivery needs of the two workstations can be combined into one batch.

[0052] Further calculate the time window for each delivery task, and for urgent delivery tasks, determine the latest start time. and latest completion time for: ,

[0053] in, Indicates the current moment. This is the maximum allowable delay time from task generation to the start of delivery. The standard execution time for completing an emergency delivery task, Indicates the urgency threshold. This represents the maximum urgency coefficient allowed in the system. Indicates the quantity of materials delivered. Indicates delivery distance. This represents the influence coefficient of material quantity. This represents the distance influence coefficient.

[0054] For the merged batch delivery task, the time window calculation needs to consider the overall status of each workstation within the batch, as shown below: ,

[0055] in, This represents the number of workstations in the batch. This is the batch size coefficient. Indicates the first in the batch The urgency coefficient of each workstation. Indicates the first in the batch Material requirements for each workstation. This serves as the baseline response time for batch tasks. This is the baseline execution time for batch delivery.

[0056] Then, calculate the overall priority index for each task:

[0057] in, The resource sharing impact coefficient, This refers to the number of shared material types involved in the task. This represents the total number of material types.

[0058] After calculating the priority index for all delivery tasks, they are categorized into four priority levels based on task type: Urgent Delivery Tasks (single tasks with an urgency coefficient greater than 1.2), Batch Delivery Tasks (formed by merging workstations with urgency coefficients between 0.8 and 1.2), Second-Urgent Delivery Tasks (single tasks with an urgency coefficient between 0.8 and 1.2), and Regular Delivery Tasks (single tasks with an urgency coefficient less than 0.8). Within each level of a single task, the overall priority index of the task is used again. Sort the values ​​in descending order to obtain the material delivery task sequence.

[0059] S4: Based on the material delivery task sequence, a two-way matching algorithm is used to adjust the delivery tasks within the task interaction area.

[0060] A dynamically updated bipartite graph model is constructed in the task interaction area. When a delivery vehicle enters the task interaction area, the status information of each delivery vehicle is obtained, including the grid coordinates of the current position, the actual remaining load capacity, the details of the currently loaded materials, the remaining power percentage, the number of workstations that have completed delivery, and the list of workstations that have not completed delivery. At the same time, it obtains detailed demand information for all workstations awaiting delivery within the interactive area, including workstation number, grid location coordinates, required quantity of each material, material priority delivery level, required delivery time window, and current material inventory of the workstation.

[0061] If the total weight of a single material type required at a certain workstation exceeds the remaining load capacity of any delivery vehicle, the material requirement will be split according to material type and priority: If it is the same type of material, it is divided into multiple sub-tasks that do not exceed the maximum remaining load capacity of the delivery vehicle. When dividing, priority is given to the quantity of materials with higher urgency. If multiple materials are involved, the material with the highest urgency coefficient will be divided into a separate sub-task, and the remaining materials will be combined and split according to their weight and type correlation.

[0062] The remaining load capacity of all delivery vehicles is used as the set of left vertices in a bipartite graph, and the material requirements of all workstations (including the sub-tasks after splitting) are used as the set of right vertices.

[0063] The edge weights between two vertex sets are dynamically adjusted: if the shortest path distance from the delivery vehicle's current location to the workstation is less than 100 meters, the edge weights are increased by 30%. If the distance is between 100 meters and 200 meters, the weighting will be increased proportionally based on the actual distance, with an increase ranging from 10% to 20%. If the distance exceeds 200 meters, the weight remains unchanged based on the base score; Furthermore, if the deviation between the time window for material demand at the workstation and the estimated arrival time of the delivery vehicle is less than 10 minutes, the edge weight will be increased by 25%. If the time deviation is between 10 and 20 minutes, the weight will be increased proportionally based on the actual deviation value, with an increase ranging from 15% to 20%. If the time deviation exceeds 20 minutes, the weight remains unchanged from the previous step; Furthermore, if the type of materials required at the workstation is the same as the type of materials already loaded in the delivery vehicle, the edge weight increases by another 20%. If two lists contain some of the same material types, the weighting will be increased proportionally to the proportion of the number of the same material types to the total number of required material types, with an increase ranging from 10% to 15%. If the types of materials are completely different, the weights will remain unchanged from the previous step.

[0064] The Hungarian algorithm is used to calculate the optimal matching for the weighted bipartite graph, continuously searching for augmenting paths in each iteration: First, assign temporary markers to the unmatched delivery vehicle nodes in the left vertex set, and expand along the unsaturated edges to the right vertex set until an unmatched workstation node is found on the right or expansion can no longer continue. If an augmenting path is found, update the current matching scheme and recalculate the edge weights, then continue to the next iteration; If no augmenting path can be found, then the current matching scheme is the optimal solution.

[0065] The matching results are then redistributed: a new delivery task list is generated for each delivery vehicle, which includes the assigned workstation number, details of the materials to be delivered, suggested delivery routes, estimated arrival time, and loading / unloading time windows.

[0066] In summary, the material collaborative delivery method for the electric forklift assembly process based on the embodiments of the present invention has been clarified. By employing a time-segmented material consumption prediction model, it improves the accuracy of material demand prediction for each workstation and can respond in real time to dynamically changing production needs. Combined with an improved ant colony algorithm, it optimizes the collaborative path planning of multiple delivery vehicles, fully considering material urgency, delivery task decomposition and merging strategies, achieving high efficiency and intelligence in path planning. The introduction of task interaction area design and a two-way matching algorithm dynamically adjusts the allocation of delivery tasks, improving multi-vehicle collaborative efficiency and reducing waste of delivery resources. Overall, it significantly improves the accuracy and timeliness of material delivery, ensuring the stable and efficient operation of the electric forklift assembly process.

Claims

1. A method for collaborative material delivery during the assembly process of an electric forklift, characterized in that, include: Based on the material demand information of the electric forklift assembly station, a time-segmented material consumption prediction model is used to obtain the material demand prediction data for each station. Based on the material demand forecast data and combined with the material storage location information, an improved ant colony algorithm is used to calculate the collaborative delivery path of multiple delivery vehicles. In the collaborative delivery path, there are task interaction areas between the delivery vehicles. The collaborative delivery path is decomposed according to the urgency of material delivery to generate a material delivery task sequence; Based on the material delivery task sequence, a two-way matching algorithm is used to adjust the delivery tasks within the task interaction area; The task interaction area is constructed based on material demand forecast data and warehouse location coordinates, using an 8×8 grid-based collaborative delivery route planning model. This model is determined by a dynamic labeling coefficient formula, expressed as follows: in, Indicates delivery vehicle and The feasibility coefficient of task interaction between them Indicates the distance between the two workshops. The baseline distance for safe interaction between delivery vehicles. and These represent the current loads of the two vehicles, This refers to the maximum load capacity of the delivery vehicle. Based on the material delivery task sequence and the collaborative delivery path of the delivery vehicles, a bipartite graph model is constructed in the task interaction area. The edge weights of the bipartite graph are dynamically adjusted and the Hungarian algorithm is used for optimal matching to generate a task list for each delivery vehicle.

2. The material collaborative delivery method for the electric forklift assembly process according to claim 1, characterized in that, The probability of delivery route selection calculated using the improved ant colony algorithm is expressed as follows: in, Represents path ( , The concentration of pheromones on the surface. Represents path ( , Heuristic information values ​​on ) and These represent the pheromone importance factor and the heuristic information value importance factor, respectively. Indicates the first The set of allowed locations for an ant to choose from next. Indicates from workstation to candidate work station The actual pheromone concentration values ​​along the path between them. Indicates from workstation to candidate work station The actual heuristic information value of the path between them.

3. The material collaborative delivery method for the electric forklift assembly process according to claim 2, characterized in that, The calculation of the heuristic information value incorporates three dimensions: material urgency, delivery vehicle weight, and required time, and is expressed as follows: in, Indicates position The material urgency coefficient, Indicates position arrive distance, Indicates the current load capacity of the delivery vehicle. Indicates the maximum load capacity of the delivery vehicle. Indicates position Material demand time, Indicates the time decay coefficient. For position Material requirements This represents the current inventory level. As the safety stock threshold, This refers to the assembly cycle time at this workstation. This is the maximum assembly cycle time.

4. The material collaborative delivery method for the electric forklift assembly process according to claim 3, characterized in that, The optimal delivery route is output based on the improved ant colony algorithm, including: the initial location of the delivery vehicle. Material storage locations and assembly station locations arranged in the order of access. , ,2,..., -1 and the path endpoint Each node has been marked with its coordinates in the 8×8 grid. The attributes include material loading and unloading information, arrival time requirements, estimated loading and unloading duration, and task interaction markers.

5. The material collaborative delivery method for the electric forklift assembly process according to claim 4, characterized in that, Based on the optimal delivery route scheme, a complete delivery vehicle motion trajectory containing a smooth trajectory point sequence and time, speed, and acceleration information is generated by converting grid coordinates to actual coordinates, inserting trajectory sampling points, adding the time dimension, and optimizing with cubic spline interpolation.

6. The material collaborative delivery method for the electric forklift assembly process according to claim 5, characterized in that, Obtain material information from assembly stations, calculate the production time that material inventory can support, compare it with the remaining assembly time to obtain the material delivery urgency coefficient, and adjust the delivery priority by comprehensively considering factors such as station distance and inventory level.

7. The material collaborative delivery method for the electric forklift assembly process according to claim 6, characterized in that, When the urgency coefficient exceeds the preset threshold, it is determined to be an emergency delivery task. The weight of the task is calculated by combining the material value and the sharing coefficient. At the same time, the feasibility index of the task is calculated for workstations that do not reach the preset threshold. If the conditions are met, the task is merged into a delivery batch.

8. The material collaborative delivery method for the electric forklift assembly process according to claim 7, characterized in that, Calculate the time window and comprehensive priority index for each delivery task, divide the task priority levels, and generate a material delivery task sequence by arranging them in descending order of priority index.

Images