Intelligent distribution decision-making method and system for production materials

Abstract

The invention provides an intelligent distribution decision-making method and system for production materials, and the method comprises the steps: firstly obtaining the material demand information, including material types, demand priorities and demand time windows, of all production nodes in a preset period of a target factory, and then combining the real-time traffic and storage layout data of an area where the target factory is located; the method comprises the following steps: determining dynamic environment parameters including a road traffic state and a storage node capacity, then generating an initial distribution path set according to a demand priority and a time window, inputting the initial distribution path set and the dynamic environment parameters into a pre-trained path optimization model after each path is associated with a transportation tool and a material distribution scheme, and outputting a target delivery path and a dynamic scheduling instruction used for adjusting the loading capacity and the driving speed of the transportation tool, and finally controlling the transportation tool to deliver materials according to the target delivery path and the dynamic scheduling instruction, and re-planning the path on the way based on the real-time dynamic environment parameters to realize intelligent and efficient delivery.

Description

Technical Field

[0001] The present invention relates to the technical field of digital factories, and in particular, to an intelligent distribution decision-making method and system for production materials. Background Art

[0002] In modern manufacturing, the distribution link of production materials plays a key role in ensuring the efficient and stable production of factories. With the rapid development of manufacturing and the continuous improvement of the intelligent level, many limitations of traditional production material distribution decision-making methods have gradually emerged.

[0003] Most of the previous material distribution decision-making methods relied on experience and simple planning models. In terms of obtaining material demand information, only the basic types of materials were often considered, ignoring key factors such as demand priority and demand time window. This led to the inability to reasonably arrange the distribution sequence and time in the actual distribution process, and it was easy to have the situation that emergency materials were not supplied in time, while some non-emergency materials occupied too much distribution resources, seriously affecting the production progress.

[0004] Regarding the environmental factors in the distribution process, traditional methods rarely considered real-time traffic data and warehouse layout data, and basically operated according to fixed distribution routes and preset warehouse allocation methods. When encountering sudden road congestion, traffic accidents or changes in the capacity of warehouse nodes, etc., the distribution plan could not be adjusted in time, often resulting in distribution delays, cargo backlogs and other problems, greatly reducing the distribution efficiency and resource utilization rate.

[0005] In terms of distribution route planning, the routes generated by traditional methods were relatively single and fixed, without fully considering the priority and time window of material demand comprehensively. Moreover, once the route was determined, it was difficult to flexibly adjust according to the actual situation during the distribution process, lacking a dynamic optimization mechanism. This made it impossible to reasonably allocate and adjust the loading capacity and driving speed of transportation tools according to the real-time road conditions and material demands, resulting in resource waste and extended distribution time. Summary of the Invention

[0006] In view of the above-mentioned problems, in combination with the first aspect of the present invention, embodiments of the present invention provide an intelligent distribution decision-making method for production materials, and the method includes: Obtain the material demand information of each production node in the target factory within a preset period, where the material demand information includes material type, demand priority and demand time window; Based on the real-time traffic data and warehouse layout data in the area where the target factory is located, determine the dynamic environment parameters corresponding to the material demand information, where the dynamic environment parameters include road traffic status and warehouse node capacity; Generate an initial delivery route set according to the demand priority and demand time window in the material demand information, where each initial delivery route in the initial delivery route set is associated with at least one transportation vehicle and a corresponding material allocation plan; Input the initial delivery route set and the dynamic environment parameters corresponding to the material demand information into a pre-trained route optimization model, and output a target delivery route and dynamic scheduling instructions, where the dynamic scheduling instructions are used to adjust the loading capacity and driving speed of the transportation vehicle; Control the transportation vehicle to perform material delivery according to the target delivery route and dynamic scheduling instructions, and perform route replanning based on the dynamically updated dynamic environment parameters during the delivery process.

[0007] On the other hand, an embodiment of the present invention further provides an intelligent delivery decision-making system for production materials, including a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor. The machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0008] Based on the above aspects, the embodiments of the present application comprehensively consider the material type, demand priority, demand time window, as well as the road traffic status and warehouse node capacity. An initial delivery route set is generated based on the demand priority and time window, and then optimized by using a pre-trained route optimization model combined with dynamic environment parameters, which improves the rationality and timeliness of the delivery route. Through the dynamic scheduling instructions output by the route optimization model, the loading capacity and driving speed of the transportation vehicle can be adjusted in real time, realizing the precise allocation and efficient utilization of transportation resources. It not only improves the use efficiency of transportation vehicles, but also can flexibly adjust the delivery strategy according to the real-time road conditions and warehouse status, avoiding delivery delays caused by local congestion or warehouse capacity limitations. Route replanning is performed based on the dynamically updated dynamic environment parameters during the delivery process, further enhancing the flexibility and adaptability of the delivery, ensuring that material delivery can be efficiently and accurately completed under various complex conditions, and comprehensively improving the material delivery efficiency and management level of the target factory. Description of the Drawings

[0009] Figure 1 It is a schematic execution flow diagram of the intelligent delivery decision-making method for production materials provided by the embodiment of the present invention.

[0010] Figure 2 It is a schematic diagram of exemplary hardware and software components of the intelligent delivery decision-making system for production materials provided by the embodiment of the present invention. Detailed Embodiments

[0011] The present invention will be specifically described below in conjunction with the accompanying drawings of the specification. Figure 1It is a schematic flow chart of an intelligent distribution decision-making method for production materials provided by an embodiment of the present invention. The intelligent distribution decision-making method for production materials will be introduced in detail below.

[0012] Step S110: Obtain the material requirement information of each production node in the target factory within a preset period. The material requirement information includes material type, demand priority, and demand time window.

[0013] Taking a digital factory related to automotive power batteries as an example, the production process of automotive power batteries involves multiple production nodes. For example, at the electrode preparation node, materials such as positive electrode materials, negative electrode materials, and separators are required. Among them, the positive electrode material is crucial for the performance of the battery, and its demand priority is relatively high. Assuming that the preset period is one week, within this week, the demand time window for the positive electrode material at the electrode preparation node may be from Monday to Wednesday, because the subsequent battery assembly process depends on the timely supply of the positive electrode material to proceed normally.

[0014] For the negative electrode material, although it is also a key material, its demand priority is slightly lower than that of the positive electrode material, and its demand time window may be from Monday to Thursday. The demand priority of the separator material is relatively lower, and the demand time window is from Monday to Friday. This is because in the battery assembly process, the installation process of the separator can have a more flexible time arrangement to a certain extent.

[0015] In addition, as a special material, electrolyte has volatility and corrosiveness, and requires special storage and transportation conditions. Its demand priority is relatively high, and the demand time window is from Tuesday to Thursday to ensure that it is added to the battery assembly process at the right time and avoid the performance degradation or safety risks caused by the long-term exposure of the electrolyte to the environment. Different production nodes have clear type requirements, different demand priorities, and specific demand time windows for various materials according to the production plan and process flow.

[0016] Step S120: Based on the real-time traffic data and warehouse layout data in the area where the target factory is located, determine the dynamic environment parameters corresponding to the material requirement information. The dynamic environment parameters include road traffic status and warehouse node capacity.

[0017] Specifically, the target factory is located in an industrial park, surrounded by a complex road network connecting various warehousing nodes and production nodes. Through the sensor network deployed in the park, real-time traffic flow data can be collected. For example, on the road from a certain warehousing node to the electrode preparation production node, during the peak working hours from 8 to 10 am on weekdays, the traffic flow is large, the average vehicle speed is low, and the congestion level is high. By collecting traffic data on each road section in different time intervals (such as every half hour as an interval), the average vehicle speed and congestion level can be extracted. For example, in the time interval from 8 to 8:30 in the morning, the average vehicle speed on the road connecting the cathode material storage area and the electrode preparation workshop is 20 km / h, and the congestion level is level 3 (assuming the congestion level is divided into 1 - 5 levels, with level 5 being the most congested).

[0018] Meanwhile, obtain the warehousing layout data. In the cathode material storage area, its location coordinates are (x1, y1), and there are multiple shelves and storage areas. At the current moment, the remaining storage capacity is 1000 kg (assuming the cathode material is measured in kilograms), and the material handling equipment (such as forklifts, automated guided vehicles, etc.) is in good working condition and can normally perform material storage and retrieval operations. For the anode material storage area, the location coordinates are (x2, y2), the remaining storage capacity is 800 kg, but one of the forklifts is malfunctioning, which will affect the material storage and retrieval efficiency and thus the receivable material volume of this warehousing node.

[0019] Based on the congestion level and average vehicle speed in the real-time traffic flow data, construct a road traffic state matrix. The rows of the road traffic state matrix represent road section identifiers (such as R1 represents the road section connecting the cathode material storage area and the electrode preparation workshop), the columns represent time intervals (such as 8:00 - 8:30, 8:30 - 9:00, etc.), and the matrix elements are the traffic scores of the corresponding road sections in the time intervals. For example, the traffic score of road section R1 in the time interval from 8:00 to 8:30 is 2 points (assuming the traffic score is from 1 - 5 points, with 5 points being the best traffic condition). Based on the remaining storage capacity of the warehousing node and the working state of the material handling equipment, generate a warehousing capacity state vector. Each element of the vector represents the receivable material volume of the corresponding warehousing node at the current moment. For example, the element corresponding to the cathode material storage area is 1000 kg, and the element corresponding to the anode material storage area is the adjusted receivable volume due to the forklift failure (assuming it is adjusted to 500 kg according to the degree of failure impact). By associatively mapping the road traffic state matrix with the warehousing capacity state vector, the road traffic state and warehousing node capacity in the dynamic environment parameters can be generated.

[0020] Step S130, generate an initial delivery path set according to the demand priority and demand time window in the material demand information, and each initial delivery path in the initial delivery path set is associated with at least one transportation tool and the corresponding material allocation plan.

[0021] Specifically, each production node can be divided into an urgent demand node and a non-urgent demand node according to the demand priority in the material requirement information. For the electrode preparation node, due to the importance of the cathode material for production and the urgency of its demand time window (from Monday to Wednesday), the electrode preparation node belongs to the urgent demand node when it comes to the supply of the cathode material. Taking the demand for the cathode material at the electrode preparation node as an example, the starting time of its demand time window is extracted as 8:00 am on Monday. The current system time is 8:00 pm on Sunday. Calculate the delivery urgency. Since the time difference is 12 hours, the delivery urgency is relatively high (the delivery urgency is inversely proportional to the time difference).

[0022] After calculating the delivery urgency of each urgent demand node based on the difference between the starting time of the demand time window and the current system time, an initial candidate path is generated using the greedy algorithm according to the delivery urgency and the location of the storage nodes in the warehousing layout data. For example, starting from the main supplier warehouse of the cathode material, to reach the urgent demand node of the electrode preparation workshop, the greedy algorithm will preferentially select a path with a relatively short distance and good road traffic conditions (evaluated according to the previously constructed road traffic status matrix). Suppose the initial candidate path passes through several intermediate storage nodes, such as transfer storage point A and transfer storage point B.

[0023] Traverse each storage node in the initial candidate path and determine whether the remaining storage capacity of the storage node meets the material demand of the corresponding urgent demand node. For example, transfer storage point A was originally planned to provide some transfer storage of the cathode material for the electrode preparation workshop, but it is found that its remaining storage capacity is 500 kg, while the demand for the cathode material at the electrode preparation workshop during this period is 800 kg, which does not meet the demand. At this time, dynamically allocate 300 kg of the cathode material from adjacent storage nodes (such as transfer storage point B), and update the initial candidate path, adjusting the original material transfer plan at transfer storage point A to transfer storage point B.

[0024] Add the initial candidate paths that meet all urgent demand nodes to the initial delivery path set. For each initial candidate path, assign the type of transportation tool and the upper limit of the loading capacity. Suppose an electric forklift is assigned as the transportation tool for transporting the cathode material. Considering the load-bearing capacity and safety limitations of the electric forklift, the upper limit of its loading capacity is set to 1500 kg. At the same time, formulate a material distribution plan for this initial delivery path, specifying the loading and unloading amounts at each storage node. For example, load 1500 kg at the cathode material supplier warehouse, unload 300 kg at transfer storage point B, and finally unload 1200 kg at the electrode preparation workshop.

[0025] Step S140: Input the initial delivery path set and the dynamic environment parameters corresponding to the material requirement information into a pre-trained path optimization model, and output the target delivery path and dynamic scheduling instructions, where the dynamic scheduling instructions are used to adjust the loading capacity and driving speed of the transportation vehicle.

[0026] Specifically, for each initial delivery path in the initial delivery path set, normalization processing is first performed. For example, for an initial delivery path from the cathode material supplier's warehouse to the electrode preparation workshop, the corresponding transportation vehicle type is extracted as an electric forklift, the upper limit of the loading capacity is 1500 kg, the sequence of road section identifiers passed through (such as R1, R2, R3, etc. representing different road sections), and the sequence of associated warehouse node identifiers (such as S1 representing the cathode material supplier's warehouse, S2 representing the transfer warehouse point B, and S3 representing the electrode preparation workshop) are used as structured path description data.

[0027] The structured path description data is concatenated with the road traffic status matrix and the warehouse capacity status vector in the dynamic environment parameters to generate a multi-dimensional input feature vector for each initial delivery path. Assume that the multi-dimensional input feature vector includes the path length weight (calculated based on the actual length of the road section and the estimated driving time), the sequence of road traffic scores (such as the sequence of traffic scores of road section R1 in different time intervals), the sequence of warehouse node receiving capacities (such as the sequence of the current material receiving amounts of each warehouse node), and the transportation vehicle capacity ratio (such as the ratio of the current loading capacity to the upper limit of the loading capacity).

[0028] Extract the spatial correlation features in the multi-dimensional input feature vector through the convolutional layer in the pre-trained path optimization model, and use the temporal attention mechanism to capture the dynamic change patterns of the road passage score sequence over time intervals. For example, arrange the road passage score sequence in the order of time intervals as a three-dimensional tensor, where the first dimension represents the path identifier (the distribution path identifier of this cathode material), the second dimension represents the road segment identifier (R1, R2, etc.), and the third dimension represents the time interval index (e.g., 8:00 - 8:30 is index 1, 8:30 - 9:00 is index 2, etc.). Use a three-dimensional convolutional kernel to perform a sliding window operation on the three-dimensional tensor to generate local spatial feature maps of the road segment passage states at different time scales. Perform a max pooling operation on the local spatial feature maps to extract the key passage state features of each road segment within a preset time window (e.g., 8:00 - 9:00). Horizontally concatenate the key passage state features with the warehousing node receiving capacity sequence to form a spatio-temporal fusion feature vector. In the temporal attention mechanism, assign trainable attention weight coefficients to each time interval. For example, for the traffic peak period from 8 to 9 in the morning, assign a relatively high attention weight coefficient because the road passage conditions during this time period have a greater impact on the entire distribution process. Normalize the attention weight coefficients through the softmax function to characterize the influence degree of different time intervals on the path feasibility. Perform a weighted sum of the normalized attention weight coefficients and the spatio-temporal fusion feature vector to output a dynamic path feature vector with temporal weights.

[0029] Based on the spatial correlation features and dynamic change patterns, calculate the path feasibility score for each initial distribution path. The path feasibility score is positively correlated with the average value of the road passage score sequence and negatively correlated with the minimum value of the warehousing node receiving capacity sequence. For example, if the average value of the passage scores of the road segments passed by a path is relatively high and the minimum receivable capacity of the warehousing node can meet the material requirements, then its path feasibility score is relatively high.

[0030] Filter the initial distribution path set according to the path feasibility score, eliminate the initial distribution paths with scores lower than the preset threshold, and arrange the remaining paths in descending order of scores to generate a candidate optimized path queue. Assume that the preset threshold is 60 points (out of 100), and the initial distribution paths with scores lower than 60 are considered infeasible and eliminated.

[0031] Traverse each candidate optimized path in the candidate optimized path queue, and adjust the path feasibility score in combination with the demand priority coefficient in the material demand information to generate a comprehensive path evaluation value after priority weighting. For example, for a path involving the distribution of cathode materials, since the demand priority of cathode materials is high, its demand priority coefficient is relatively large, and the comprehensive path evaluation value will be significantly increased when adjusting the path feasibility score.

[0032] The candidate optimization path with the highest comprehensive path evaluation value is selected as the target delivery path, and the corresponding transportation tool type and load limit are analyzed. For example, the target delivery path is determined to be the path from the cathode material supplier warehouse, passing through the transit storage point B to the electrode preparation workshop, the transportation tool is an electric forklift, and the load limit is 1500 kg.

[0033] According to the real-time traffic score of the road traffic state matrix in the target delivery path and the remaining capacity data of the storage capacity state vector, the maximum speed allowed for the transport tool on each road section and the optimal stay time at the storage node are dynamically calculated. The real-time traffic score of the road section through which the target delivery path passes in the current time interval is extracted from the road traffic state matrix. For example, the real-time traffic score of road section R1 from 9:00 to 9:30 is 3 points. According to the preset score-speed mapping table (such as a traffic score of 1 point corresponds to a speed of 10 kilometers per hour, 2 points correspond to 20 kilometers per hour, 3 points correspond to 30 kilometers per hour, etc.), the real-time traffic score is converted into a road section speed limit value. Obtain the upper limit of the benchmark speed corresponding to the type of transport tool (electric forklift) (assuming it is 40 kilometers per hour), and take the minimum value of the road section speed limit value and the upper limit of the benchmark speed as the maximum allowable speed, which is 30 kilometers per hour here. When it is detected that the real-time traffic score of the road segment ahead (such as R2) is updated in the next time interval (9:30-10:00), the score-speed conversion operation is re-executed and the maximum speed allowed for the subsequent road segment is updated. The maximum speeds allowed for consecutive road segments are arranged in chronological order to generate a speed adjustment instruction sequence, which is embedded in the dynamic scheduling instruction for real-time reading by the transportation vehicle controller.

[0034] Based on the maximum allowed speed and the optimal dwell time, dynamic scheduling instructions are generated. The dynamic scheduling instructions include a speed adjustment instruction sequence and a load redistribution instruction. The load redistribution instruction dynamically allocates the unloading ratio of each node according to the remaining capacity data of the storage nodes passing through. At each storage node of the target distribution path, the remaining capacity data of the node at the current moment and the capacity overflow status of the adjacent nodes are obtained. For example, at the transit storage point B, the current remaining capacity is 600 kg, and the adjacent transit storage point C is full (capacity overflow status). The difference between the theoretical unloading amount (assuming 500 kg) pre-allocated to the node (transit storage point B) in the current load of the transport tool and the remaining capacity data is -100 kg (500-600). If the difference is less than zero, no reallocation is required. If the difference is greater than zero, the excess part will be reallocated according to the capacity overflow status ratio of the adjacent nodes, and the theoretical unloading amount of the subsequent nodes will be updated. The actual unloading instructions of each node are generated according to the reallocated unloading amount, and the instruction execution timestamp is synchronized with the working status of the material storage and retrieval equipment of the storage node.

[0035] Step S150: According to the target delivery path and the dynamic scheduling instruction, control the transport vehicle to perform material delivery, and perform path replanning based on the dynamically updated dynamic environment parameters during the delivery process.

[0036] Specifically, according to the target delivery path and the dynamic scheduling instruction, the transport vehicle (electric forklift) starts to perform the delivery task of the cathode material. After the transport vehicle starts, it continuously receives the real-time traffic flow data uploaded by the sensor network and the capacity update information of the warehousing node.

[0037] When the electric forklift is traveling on road segment R1, assume that it is detected that the congestion level of road segment R1 in the current path exceeds the preset threshold (for example, the congestion level reaches level 4, and the preset threshold is level 3), or when the electric forklift is about to reach the transfer warehouse point B, it is detected that the remaining capacity of this warehousing node is lower than the critical value (for example, the remaining capacity is 100 kg, and the critical value is 200 kg). At this time, a path replanning event is triggered.

[0038] Pause the travel of the electric forklift, and recalculate the set of alternative paths for the target delivery path according to the latest dynamic environment parameters. For example, reconsider other possible combinations of road segments and the utilization of warehousing nodes, and generate several alternative paths.

[0039] Use a multi-objective evaluation strategy to score each alternative path in the set of alternative paths. The multi-objective evaluation strategy includes delivery delay risk, transportation cost increment, and warehousing node load balance degree. For an alternative path, it may bypass the congested road segment R1, but increase the transportation distance, resulting in an increment in transportation cost. At the same time, it is necessary to evaluate the impact of this path on the warehousing node load balance degree, such as whether it will overly rely on a certain warehousing node, and the impact on the delivery delay risk, such as whether it will miss the demand time window of the electrode preparation workshop. Score the alternative paths according to these evaluation results, select the optimal alternative path as the new target delivery path, and then the electric forklift continues to perform the material delivery task according to the new target delivery path and the corresponding dynamic scheduling instruction to ensure the timely supply of materials in the production process of automotive power batteries.

[0040] Based on the above steps, in the embodiment of the present application, by comprehensively considering the material type, demand priority, demand time window, as well as the road traffic status and the capacity of the warehousing nodes, an initial set of distribution routes is generated based on the demand priority and time window, and then a pre-trained route optimization model is used to optimize in combination with dynamic environment parameters, improving the rationality and timeliness of the distribution routes. Through the dynamic scheduling instructions output by the route optimization model, the loading capacity and driving speed of the transportation tools can be adjusted in real time, realizing the precise allocation and efficient utilization of transportation resources. This not only improves the utilization efficiency of the transportation tools, but also can flexibly adjust the distribution strategy according to the real-time road conditions and warehousing status, avoiding distribution delays caused by local congestion or warehousing capacity limitations. During the distribution process, route replanning is carried out based on the dynamically updated dynamic environment parameters, further enhancing the flexibility and adaptability of the distribution, ensuring that the material distribution can be efficiently and accurately completed under various complex conditions, and comprehensively improving the material distribution efficiency and management level of the target factory.

[0041] In a possible implementation manner, step S120 includes: Step S121, collecting real-time traffic flow data through a sensor network deployed in the target factory area, and extracting the average vehicle speed and congestion level of each road segment within a preset time interval.

[0042] In this embodiment, the sensor network fully covers each road segment. Taking the road segment connecting the cathode material storage area and the electrode preparation workshop as an example, the sensor network continuously monitors the vehicle flow on this road. Within the preset time interval from 7 am to 9 am on a weekday, through a large amount of data collection and analysis, the average vehicle speed of this road segment is obtained as 25 kilometers per hour. At the same time, according to the density of vehicles and the fluctuation of driving speed, it is determined that the congestion level of this road segment within this preset time interval is level 2 (the congestion level is set from 1 to 5 levels in total, level 1 indicates smooth traffic, and level 5 indicates severe congestion). Similarly, for the road segment connecting the anode material storage area and the battery assembly workshop, within the preset time interval from 5 pm to 7 pm, the average vehicle speed is 20 kilometers per hour, and the congestion level is level 3.

[0043] Step S122, obtaining the position coordinates, remaining storage capacity of each warehousing node in the warehousing layout data, and the working status of the material access equipment.

[0044] For example, in the cathode material storage area, its location coordinates are (X1, Y1), and this coordinate information precisely locates its geographical position within the factory area. At the current moment, after inventory checking and querying the system records, it is found that its remaining storage capacity is 800 kilograms. For the material handling and storage equipment in this storage area, such as the automatic guided vehicle, after inspection by the equipment status monitoring system, its working status shows normal, and it can efficiently carry out material handling and storage operations. For the separator material storage area, the location coordinates are (X2, Y2), the remaining storage capacity is 600 kilograms, but one of the forklifts in its material handling and storage equipment has malfunctioned, affecting the material handling and storage efficiency, requiring manual assistance, and the overall working status is not good.

[0045] Step S123: Based on the congestion level and average vehicle speed in the real-time traffic flow data, construct a road traffic status matrix. The rows of the road traffic status matrix represent road segment identifiers, the columns represent time intervals, and the matrix elements are the traffic scores of the corresponding road segments in the time intervals.

[0046] For example, the road segment connecting the cathode material storage area and the electrode preparation workshop is identified as R1, the road segment connecting the anode material storage area and the battery assembly workshop is identified as R2, etc. The columns represent time intervals, with each half-hour as a time interval, such as 7:00 - 7:30, 7:30 - 8:00, etc. The matrix elements are the traffic scores of the corresponding road segments in the time intervals. For road segment R1, in the time interval of 7:00 - 7:30, since the average vehicle speed is 25 kilometers per hour and the congestion level is 2, according to the preset conversion rules, a traffic score of 3 is given (assuming the traffic score range is 1 - 5, and 5 represents the best traffic status). For road segment R2, in the time interval of 17:00 - 17:30, because the average vehicle speed is 20 kilometers per hour and the congestion level is 3, a traffic score of 2 is given.

[0047] Step S124: Based on the remaining storage capacity of the storage node and the working status of the material handling and storage equipment, generate a storage capacity status vector, where each element of the storage capacity status vector represents the amount of material that the corresponding storage node can receive at the current moment.

[0048] For example, since the remaining storage capacity of the cathode material storage area is 800 kilograms and the material handling and storage equipment is working normally, the corresponding element of the amount of material that can be received is 800 kilograms. For the separator material storage area, due to the malfunction of the forklift affecting the material handling and storage efficiency, although the remaining storage capacity is 600 kilograms, considering possible material turnover delays and other factors, after comprehensive evaluation, it is determined that the element of the amount of material that can be received is 300 kilograms.

[0049] Step S125: Perform an associated mapping on the road traffic state matrix and the storage capacity state vector to obtain the road traffic state and the storage node capacity in the dynamic environment parameters.

[0050] For example, when planning the material distribution path from the cathode material storage area to the electrode preparation workshop, the traffic scores of road segment R1 in the road traffic state matrix in each time interval will affect the timeliness of distribution, and the receivable material quantity (800 kg) in the cathode material storage area limits the quantity of materials that can be stored in this storage area. This associated mapping can comprehensively reflect the dynamic environment parameters such as the current road traffic and the accommodation capacity of storage nodes, providing an accurate basis for subsequent material distribution planning. Whether considering the driving route selection of transportation tools or determining the allocation and storage of materials at storage nodes, these dynamic environment parameters play a crucial role. In the production process of automotive power batteries, the material requirements of each production node strictly depend on these accurate dynamic environment parameters to achieve efficient and stable supply.

[0051] In a possible implementation manner, step S130 includes: Step S131: Divide each production node into an urgent demand node and a non-urgent demand node according to the demand priority in the material demand information, and extract the start time of the demand time window of the urgent demand node.

[0052] For example, in the production process of automotive power batteries, the electrode preparation link is crucial for the entire production process, and the supply of cathode materials directly affects the start and continuous progress of electrode preparation. Therefore, in the electrode preparation production node involving the supply of cathode materials, due to its key impact on the production process, it is determined as an urgent demand node. The demand time window for this node for cathode materials is from 8:00 am on Monday to 5:00 pm on Wednesday, so the start time is 8:00 am on Monday. The production node of the battery shell, relatively speaking, has less impact on the entire production process and has certain flexibility in production arrangements, and is classified as a non-urgent demand node.

[0053] Step S132: Calculate the distribution urgency of each urgent demand node based on the difference between the start time of the demand time window and the current system time, and the distribution urgency is inversely proportional to the difference.

[0054] For example, assume that the current system time is 10 PM on Sunday. Then, from 10 PM on Sunday to 8 AM on Monday, there is a total time difference of 10 hours. To calculate the delivery urgency, using 24 hours as a reference complete cycle, divide 24 by this time difference to obtain the delivery urgency. That is, the delivery urgency is 24 divided by 10, and the result is 2.4. The calculation principle here is that with 24 hours as a complete time cycle, the shorter the time to the start time of the demand time window, the more urgent the delivery, and the higher the delivery urgency. Through such a calculation method, the delivery urgency of each emergency demand node can be quantified to prioritize nodes with high urgency in subsequent path planning.

[0055] Step S133, according to the delivery urgency and the positions of the warehousing nodes in the warehousing layout data, use the greedy algorithm to generate an initial candidate path, where the initial candidate path covers all emergency demand nodes and has the shortest path length.

[0056] For example, in the warehousing layout of an automotive power battery factory, there are multiple warehousing nodes such as the positive electrode material warehousing area, the negative electrode material warehousing area, and the separator warehousing area. Taking the distribution of the positive electrode material as an example, starting from the main supply source of the positive electrode material, considering the urgent demand for the positive electrode material at the electrode preparation production node (the distribution urgency is 2.4), the greedy algorithm first searches for a warehousing node that is relatively close to the supply source and close to the electrode preparation production node. Suppose there are warehousing nodes A, B, and C. First, calculate the distance from warehousing node A to the supply source. Assume the coordinates of the supply source are (0, 0) and the coordinates of warehousing node A are (3, 4). According to the Pythagorean theorem, calculate the distance. First, calculate the square of the difference in the abscissa, which is 3 squared equal to 9, and the square of the difference in the ordinate, which is 4 squared equal to 16. The sum of the two is 25, and then take the square root of 25 to get 5 kilometers, that is, the distance from warehousing node A to the supply source is 5 kilometers. Then, calculate the distance from warehousing node A to the electrode preparation production node. Assume the coordinates of the electrode preparation production node are (6, 8). First, calculate the difference in the abscissa as 6 - 3 = 3, its square is 9, the difference in the ordinate is 8 - 4 = 4, its square is 16, and the sum of the two is 25. Take the square root to get 5 kilometers, that is, the distance from warehousing node A to the electrode preparation production node is 5 kilometers. Similarly, calculate the distance from warehousing node B to the supply source. Assume the coordinates of warehousing node B are (2, 3). First, calculate the square of the difference in the abscissa, which is 2 squared equal to 4, and the square of the difference in the ordinate, which is 3 squared equal to 9. The sum of the two is 13, and take the square root to get approximately 3.61 kilometers. Then, calculate the distance from warehousing node B to the electrode preparation production node. The difference in the abscissa is 6 - 2 = 4, its square is 16, the difference in the ordinate is 8 - 3 = 5, its square is 25, and the sum of the two is 41. Take the square root to get approximately 6.40 kilometers. For warehousing node C, calculate its distances to the supply source and the electrode preparation production node in the same way. By comparing these distances, the greedy algorithm will preferentially select the warehousing node with the shortest comprehensive distance to the supply source and the electrode preparation production node. For example, first select warehousing node B, and then continue to search for the node with the shortest distance to the current node and the uncovered urgent demand nodes. Repeat this process until all urgent demand nodes are covered, forming an initial candidate path.

[0057] Step S134, traverse each warehousing node in the initial candidate path, and determine whether the remaining storage capacity of the warehousing node meets the material demand of the corresponding urgent demand node. If not, dynamically allocate materials from adjacent warehousing nodes and update the initial candidate path.

[0058] For example, in the initial candidate path, there is a warehousing node B. The demand for the positive electrode material at the electrode preparation production node is 800 kg, while the remaining storage capacity of the warehousing node B is found to be 500 kg after query, which does not meet the material demand of the electrode preparation production node. At this time, check the adjacent warehousing nodes. Suppose the remaining storage capacity of the adjacent warehousing node C is 400 kg, and the transportation conditions from the warehousing node C to the warehousing node B are good, without special circumstances such as additional transportation costs or time limitations. Then allocate 300 kg of the positive electrode material from the warehousing node C to the warehousing node B, so that the warehousing node B can meet the material requirements of the electrode preparation production node. At the same time, update the initial candidate path and record the material allocation situation from the warehousing node C to the warehousing node B.

[0059] Step S135: Add the initial candidate paths that meet all the emergency demand nodes to the initial distribution path set, and allocate the transportation tool type and the upper limit of the loading capacity for each initial candidate path.

[0060] For example, assume that the initial candidate path that meets all the emergency demand nodes is the path starting from the positive electrode material supply source, passing through the warehousing node B (meeting the material requirements after allocation), and finally reaching the electrode preparation production node. According to the transportation requirements of this path and the transportation equipment situation of the factory, allocate the transportation tool type for this path as an electric forklift. Considering factors such as the load-bearing capacity of the electric forklift, safety specifications, and road conditions, set the upper limit of its loading capacity to 1500 kg.

[0061] In a possible implementation manner, step S140 includes: Step S141: Perform normalization processing on each initial distribution path in the initial distribution path set, and extract the transportation tool type, the upper limit of the loading capacity, the identification sequence of the road segments passed through, and the identification sequence of the associated warehousing nodes corresponding to each initial distribution path as structured path description data.

[0062] For example, in the production of automotive power batteries, there is an initial distribution path starting from the warehouse of the positive electrode material supplier, passing through the transfer warehousing point A and the transfer warehousing point B, and finally reaching the electrode preparation workshop. For this path, the transportation tool type is an electric forklift, and its upper limit of the loading capacity is 1500 kg. The identification sequence of the road segments passed through is the road segments R1, R2, and R3, where R1 represents the road segment from the warehouse of the positive electrode material supplier to the transfer warehousing point A, R2 represents the road segment from the transfer warehousing point A to the transfer warehousing point B, and R3 represents the road segment from the transfer warehousing point B to the electrode preparation workshop. The identification sequence of the associated warehousing nodes is S1 (warehouse of the positive electrode material supplier), S2 (transfer warehousing point A), S3 (transfer warehousing point B), and S4 (electrode preparation workshop). The above data constitutes the structured path description data of this initial distribution path.

[0063] Step S142: Concatenate the structured path description data with the road traffic status matrix and the warehousing capacity status vector in the dynamic environment parameters to generate a multi-dimensional input feature vector for each initial delivery path. The multi-dimensional input feature vector includes path length weight, road traffic score sequence, warehousing node receiving capacity sequence, and transportation tool capacity ratio.

[0064] The multi-dimensional input feature vector includes path length weight, road traffic score sequence, warehousing node receiving capacity sequence, and transportation tool capacity ratio. When calculating the path length weight, first calculate the actual length of the path. For example, according to the actual lengths of road segments R1, R2, and R3 (assuming the length of R1 is 5 kilometers, the length of R2 is 3 kilometers, and the length of R3 is 4 kilometers), the total length is 12 kilometers. Then calculate its weight according to a certain standard (such as based on the maximum length among all possible paths). If the maximum length is assumed to be 20 kilometers, then the length weight of this path is 12 divided by 20, which is equal to 0.6. For the road traffic score sequence, obtain the traffic scores of road segments R1, R2, and R3 in each time interval from the road traffic status matrix. Assume that the traffic score of R1 from 8:00 to 8:30 is 3 points, and the traffic score from 8:30 to 9:00 is 2 points, etc. R2 and R3 also have similar traffic scores in different time intervals, and these scores form the road traffic score sequence. In terms of the warehousing node receiving capacity sequence, obtain the current receivable material quantities of transfer warehousing points A, B, and the electrode preparation workshop from the warehousing capacity status vector. Assume that transfer warehousing point A is 500 kilograms, transfer warehousing point B is 400 kilograms, and the electrode preparation workshop is 1000 kilograms, and these quantities form the warehousing node receiving capacity sequence. The transportation tool capacity ratio is the ratio of the current loading amount (assuming it is 1000 kilograms) to the loading amount limit (1500 kilograms), that is, 1000 divided by 1500, approximately equal to 0.67. By concatenating these data, a multi-dimensional input feature vector is generated.

[0065] Step S143: Extract the spatial correlation features in the multi-dimensional input feature vector through the convolutional layer in the pre-trained path optimization model, and use the time attention mechanism to capture the dynamic change pattern of the road traffic score sequence over time intervals.

[0066] For the convolutional layer to extract spatial correlation features, taking the road passage score sequence as an example, the passage scores of road segments are regarded as a kind of spatially distributed data. Assume that the road passage score sequence is arranged as a three-dimensional tensor in the order of time intervals. The first dimension represents the path identifier (here it is the path identifier from the cathode material supplier's warehouse to the electrode preparation workshop), the second dimension represents the road segment identifier (R1, R2, R3), and the third dimension represents the time interval index (e.g., 8:00 - 8:30 is index 1, 8:30 - 9:00 is index 2, etc.). A three-dimensional convolutional kernel is used to perform a sliding window operation on this three-dimensional tensor, covering a certain range of data blocks each time, thereby generating local spatial feature maps of the road segment passage states at different time scales. A max pooling operation is performed on the local spatial feature maps, that is, the maximum value is selected in the local spatial feature map, and this maximum value represents the most significant passage state feature in this local area. For example, in a certain local spatial feature map, there are multiple passage score values, and the largest one is selected as the key passage state feature in this local area. The key passage state features extracted from each local area are combined to obtain the key passage state features of each road segment within the preset time window. Then these key passage state features are horizontally concatenated with the warehousing node reception capacity sequence to form a spatio-temporal fusion feature vector.

[0067] In terms of the time attention mechanism, trainable attention weight coefficients are assigned to each time interval. For example, for the time interval 8:00 - 8:30, based on the analysis of historical data and the current traffic pattern, an attention weight coefficient is assigned, assumed to be 0.3. For the time interval 8:30 - 9:00, the assigned attention weight coefficient is 0.25, etc. Then these attention weight coefficients are normalized by the softmax function. The calculation process of the softmax function is that for each attention weight coefficient, first calculate the power with this attention weight coefficient as the base and the natural constant e as the exponent. For example, for 0.3, calculate e to the power of 0.3, and for 0.25, calculate e to the power of 0.25, etc. Then divide these power results by the sum of all power results respectively to obtain the normalized attention weight coefficients. These normalized attention weight coefficients characterize the influence degree of different time intervals on the path feasibility. The normalized attention weight coefficients are weighted and summed with the spatio-temporal fusion feature vector. For example, assume that an element value in the spatio-temporal fusion feature vector is 10, and the corresponding normalized attention weight coefficient is 0.3, then the weighted result is 10 multiplied by 0.3 equals 3. After weighted summing all elements in this way, a dynamic path feature vector with temporal weights is output.

[0068] Step S144: Based on the spatial association features and dynamic change patterns, calculate the path feasibility score for each initial delivery path. The path feasibility score is positively correlated with the average value of the road passage score sequence and negatively correlated with the minimum value of the receiving capacity sequence of the warehousing nodes.

[0069] When calculating the average value of the road passage score sequence, first add up the passage scores of road segments R1, R2, and R3 in each time interval. Assume that the total passage score of R1 in different time intervals is 20 points, R2 is 15 points, and R3 is 18 points, with a total of 53 points. Then divide by the number of time intervals (assume it is 10 time intervals) to get an average value of 5.3 points. The minimum value of the receiving capacity sequence of the warehousing nodes is 400 kg at transfer warehousing point A, 400 kg at transfer warehousing point B, and 1000 kg at the electrode preparation workshop, and the minimum value is 400 kg. According to a certain calculation rule (for example, set a basic value related to production demand, assume it is 1000), calculate the path feasibility score. Assume that the calculation method positively correlated with the average value of the road passage score sequence is to directly multiply by a coefficient, and this coefficient is set to 0.5 according to experience. Then the part related to the road passage score is 5.3 multiplied by 0.5, which equals 2.65. The calculation method negatively correlated with the minimum value of the receiving capacity sequence of the warehousing nodes is to multiply the result of subtracting the minimum value from the basic value by another coefficient. Assume this coefficient is 0.1, then this part is (1000 - 400) multiplied by 0.1, which equals 60. Add these two parts together (2.65 + 60) to get the path feasibility score of 62.65.

[0070] Step S145: Filter the set of initial delivery paths according to the path feasibility scores, eliminate the initial delivery paths with scores lower than the preset threshold, and sort the remaining paths in descending order of scores to generate a candidate optimized path queue.

[0071] Assume the preset threshold is 50 points. Then for those initial delivery paths with path feasibility scores lower than 50 points, such as an initial delivery path starting from the negative electrode material supplier's warehouse with a path feasibility score of 45 points, this initial delivery path will be eliminated. Sort the remaining paths from high to low scores. For example, if there are three remaining paths with path feasibility scores of 65 points, 60 points, and 55 points respectively, arrange them in this order to form a candidate optimized path queue.

[0072] Step S146: Traverse each candidate optimized path in the candidate optimized path queue, and adjust the path feasibility score in combination with the demand priority coefficient in the material demand information to generate a comprehensive path evaluation value after priority weighting.

[0073] In the production of automotive power batteries, positive electrode materials are critical to the electrode preparation process, and their demand priority is relatively high, so the demand priority coefficient is set to 1.2. For a candidate optimization path involving the distribution of positive electrode materials, its original path feasibility score is 60 points, and the comprehensive path evaluation value after adjustment of the demand priority coefficient is 60 times 1.2, which equals 72 points. For the candidate optimization path for the distribution of negative electrode materials, since the demand priority of negative electrode materials is relatively low, the demand priority coefficient is set to 0.8. If its original path feasibility score is 55 points, then the adjusted comprehensive path evaluation value is 55 times 0.8, which equals 44 points.

[0074] Step S147, selecting the candidate optimization path with the highest comprehensive path evaluation value as the target delivery path, and analyzing its corresponding transportation tool type and loading capacity upper limit.

[0075] In the candidate optimization path queue, it is found through calculation that a candidate optimization path starting from the cathode material supplier warehouse, passing through a specific transfer storage point and reaching the electrode preparation workshop has the highest comprehensive path evaluation value, for example, 75 points. The transportation tool type corresponding to this path is an electric forklift, and the upper limit of the load is 1,500 kilograms, so this path is determined as the target delivery path.

[0076] Step S148, dynamically calculate the maximum allowable speed of the transportation tool on each road section and the optimal stay time at the storage node based on the real-time traffic score of the road traffic status matrix in the target delivery path and the remaining capacity data of the storage capacity state vector.

[0077] Extract the real-time traffic score of the road segment that the target delivery path passes through in the current time interval from the road traffic status matrix. For example, the real-time traffic score of road segment R1 from 9:00 to 9:30 is 3 points. Calculate the maximum allowable speed according to the preset score-speed mapping table. Assuming that the traffic score of 1 point in the score-speed mapping table corresponds to a speed of 10 km / h, 2 points corresponds to 20 km / h, 3 points corresponds to 30 km / h, etc., then the maximum allowable speed of road segment R1 is 30 km / h. Obtain the upper limit of the benchmark speed corresponding to the type of transport tool (electric forklift) (assuming it is 40 km / h), and take the minimum value of the road segment speed limit and the upper limit of the benchmark speed as the maximum allowable speed, which is 30 km / h here. When calculating the optimal stay time at the storage node, factors such as the remaining capacity data of the storage node and the loading and unloading efficiency of the materials should be considered. Assume that the remaining capacity of transit storage point A is 500 kg, the loading and unloading efficiency of the electric forklift is 100 kg of materials every 10 minutes, and 300 kg of materials currently need to be unloaded at transit storage point A, then the optimal stay time is 300 divided by 100 times 10, which equals 30 minutes.

[0078] Step S149: Generate the dynamic scheduling instruction based on the allowed maximum speed and the optimal residence duration. The dynamic scheduling instruction includes a speed adjustment instruction sequence and a loading quantity redistribution instruction, where the loading quantity redistribution instruction dynamically allocates the unloading ratio of each node according to the remaining capacity data of the passing storage nodes.

[0079] For example, if the allowed maximum speed for road segment R1 is 30 km / h, then the part of the speed adjustment instruction sequence corresponding to R1 is the instruction to adjust the speed to 30 km / h. For the loading quantity redistribution instruction, at each storage node on the target delivery path, obtain the remaining capacity data of the node at the current moment and the capacity overflow status of the adjacent node. For example, at transfer storage point A, the current remaining capacity is 500 kg, and the adjacent transfer storage point B is full (capacity overflow status). Calculate the difference between the theoretical unloading quantity (assumed to be 400 kg) pre-allocated to this node (transfer storage point A) in the current loading quantity of the transport vehicle and the remaining capacity data, which is -100 kg (400 - 500). Since the difference is less than zero, no redistribution is required. If the difference is greater than zero, then redistribute the excess part according to the capacity overflow status ratio of the adjacent node, and update the theoretical unloading quantity of the subsequent nodes. Generate the actual unloading instruction for each node according to the redistributed unloading quantity, and synchronize the instruction execution timestamp with the working status of the material access equipment at the storage node. In this way, the entire process from the initial delivery path set to the output of the target delivery path and the dynamic scheduling instruction is completed, ensuring the efficient and accurate delivery of materials in the production process of automotive power batteries.

[0080] For example, in a possible implementation manner, step S143 includes: Step S1431: Arrange the road passage score sequence in chronological order as a three-dimensional tensor, where the first dimension represents the path identifier, the second dimension represents the road segment identifier, and the third dimension represents the time interval index.

[0081] For example, the first dimension represents the path identifier, which refers to a specific delivery path from the cathode material supplier to the electrode preparation workshop. The second dimension represents the road segment identifier. For example, R1 represents the road segment from the supplier's warehouse to the first transfer storage point, and R2 represents the road segment from the transfer storage point to the next location, etc. The third dimension represents the time interval index, such as 8:00 - 8:30 is index 1, 8:30 - 9:00 is index 2, etc.

[0082] Step S1432: Perform a sliding window operation on the three-dimensional tensor using a three-dimensional convolutional kernel to generate a local spatial feature map of the road segment passage status at different time scales.

[0083] The sliding window operation is similar to a sliding window moving on this three-dimensional data structure, covering a certain range each time. Taking road segment R1 as an example, the passing scores in different time intervals form a two-dimensional surface. The sliding window slides on this surface. When the window covers the score data of a certain number of time intervals and road segments, a local spatial feature map of the passing state of road segment R1 at different time scales can be generated. This local spatial feature map reflects the local characteristics of the passing state of road segment R1 at a specific time scale.

[0084] Step S1433: Perform a max pooling operation on the local spatial feature map to extract the key passing state features of each road segment within a preset time window.

[0085] For example, select the maximum value among them. This maximum value is the key passing state feature of each road segment within the preset time window. For example, if there are multiple score values in this local spatial feature map representing the combined influence of different traffic flow, vehicle speed, etc., select the largest score value among them. This value represents the most significant passing state feature in this local area.

[0086] Step S1434: Horizontally splice the key passing state features with the storage node receiving capacity sequence to form a spatio-temporal fusion feature vector.

[0087] Suppose the key passing state feature is a passing state score value of 3 for road segment R1 within a specific time window (this passing state score value represents a comprehensive passing condition assessment). In the storage node receiving capacity sequence, the receiving capacity of transfer storage point A is 500 kilograms. Splice the two data of 3 points and 500 kilograms together in order to form a part of the spatio-temporal fusion feature vector. The data of other road segments and storage nodes are spliced in the same way.

[0088] Step S1435: In the time attention mechanism, assign a trainable attention weight coefficient to each time interval, and normalize the attention weight coefficient through the softmax function to characterize the influence degree of different time intervals on the path feasibility.

[0089] For example, for the time interval from 8:00 to 8:30, assign an attention weight coefficient according to various factors such as historical traffic data, current production plan, and road surrounding environment. Suppose after analysis, the traffic condition in this time interval has a greater impact on the entire delivery path, and the assigned attention weight coefficient is 0.3. For the time interval from 8:30 to 9:00, since the traffic flow may be relatively small, the assigned attention weight coefficient is 0.2.

[0090] The calculation process of the softmax function is as follows: For each attention weight coefficient, first calculate the power with this coefficient as the base and the natural constant e as the exponent. For the coefficient 0.3, calculate e to the power of 0.3 to obtain a value. For the coefficient 0.2, calculate e to the power of 0.2 to obtain another value. Then divide these power results by the sum of all power results respectively. For example, assuming there are only these two coefficients, the sum of e to the power of 0.3 and e to the power of 0.2 is obtained. Dividing e to the power of 0.3 by this sum gives a normalized coefficient, and dividing e to the power of 0.2 by this sum gives another normalized coefficient. These normalized attention weight coefficients characterize the influence degree of different time intervals on the path feasibility.

[0091] Step S1436, perform weighted summation on the normalized attention weight coefficients and the spatio-temporal fusion feature vector, and output a dynamic path feature vector with temporal weights.

[0092] Suppose an element value in the spatio-temporal fusion feature vector is 10, and the corresponding normalized attention weight coefficient is 0.3. Then the weighted result is 10 multiplied by 0.3, which equals 3. Perform such weighted calculations on each element in the spatio-temporal fusion feature vector. After adding up all the weighted results, a dynamic path feature vector with temporal weights is output. This dynamic path feature vector comprehensively considers the spatial correlation features and the temporal dynamic change patterns, and can more accurately reflect the characteristics of the delivery path.

[0093] For example, in a possible implementation manner, step S148 includes: Step S1481, extract the real-time traffic score of the road segments passed by the target delivery path in the current time interval from the road traffic state matrix.

[0094] For example, the target delivery path is from the cathode material supplier to the electrode preparation workshop, and the real-time traffic score of road segment R1 in the time interval from 9:00 to 9:30 is 3 points.

[0095] Step S1482, convert the real-time traffic score into a speed limit value for the road segment according to a preset score-speed mapping table, where the higher the traffic score in the score-speed mapping table, the greater the corresponding speed limit value.

[0096] Suppose the score-speed mapping table stipulates that a traffic score of 1 point corresponds to a speed of 10 km / h, 2 points correspond to 20 km / h, 3 points correspond to 30 km / h, etc. Then the speed limit value corresponding to the real-time traffic score of 3 points for road segment R1 is 30 km / h.

[0097] Step S1483: Obtain the benchmark speed limit corresponding to the type of the transportation vehicle, and take the minimum value of the road section speed limit and the benchmark speed limit as the allowed maximum speed.

[0098] For example, for a transportation vehicle such as an electric forklift, its benchmark speed limit is determined to be 40 km / h according to factors such as vehicle performance, safety regulations, and internal road conditions of the factory. Then, take the minimum value of the road section speed limit and the benchmark speed limit as the allowed maximum speed. For road section R1, the road section speed limit of 30 km / h is less than the benchmark speed limit of 40 km / h, so the allowed maximum speed of R1 is 30 km / h.

[0099] Step S1484: When it is detected that the real-time traffic score of the front road section is updated in the next time interval, re-execute the score-speed conversion operation and update the allowed maximum speed of the subsequent road sections.

[0100] For example, the real-time traffic score of road section R2 in the time interval from 9:30 to 10:00 is updated from the previously predicted 2 points to 1 point. According to the score-speed mapping table, the road section speed limit corresponding to 1 point is 10 km / h. If the benchmark speed limit of R2 is 30 km / h, then the allowed maximum speed of R2 is updated to 10 km / h, and the allowed maximum speeds in the subsequent path planning related to R2 should be updated accordingly.

[0101] Step S1485: Arrange the allowed maximum speeds of consecutive road sections in chronological order to generate a speed adjustment instruction sequence, and embed it in the dynamic scheduling instruction for the transportation vehicle controller to read in real time.

[0102] For example, the allowed maximum speed of road section R1 is 30 km / h, and the allowed maximum speed of road section R2 is 10 km / h. Generate a speed adjustment instruction sequence in order. This speed adjustment instruction sequence contains the speed adjustment information of each road section, and the transportation vehicle controller can adjust the speed in time during driving according to this speed adjustment instruction sequence.

[0103] For example, in a possible implementation manner, step S149 includes: Step S1491: At each warehousing node of the target delivery path, obtain the remaining capacity data of the node at the current moment and the capacity overflow status of the adjacent node.

[0104] For example, in the target delivery path, the current remaining capacity of the transfer warehouse point A is 500 kg, and the adjacent transfer warehouse point B is full (capacity overflow status).

[0105] Step S1492: Calculate the difference between the theoretical unloading quantity pre-allocated to the node in the current loading quantity of the transportation vehicle and the remaining capacity data.

[0106] Assume that the current loading capacity of the transportation vehicle is 1000 kg, and the theoretical unloading quantity pre-allocated to transfer storage point A is 600 kg. Then the difference is 600 - 500 = 100 kg.

[0107] Step S1493, if the difference is greater than zero, re-allocate the excess part according to the capacity overflow status ratio of adjacent nodes, and update the theoretical unloading quantity of subsequent nodes.

[0108] For example, since transfer storage point B is full and cannot receive the overflow from transfer storage point A, it is necessary to consider other adjacent nodes or adjust the subsequent unloading plan. If there is another adjacent node, transfer storage point C, with a remaining capacity of 800 kg and no risk of capacity overflow, then these 100 kg can be allocated to transfer storage point C according to a certain ratio, and the theoretical unloading quantity of subsequent nodes is updated.

[0109] Step S1494, generate the actual unloading instructions for each node according to the re-allocated unloading quantity, and synchronize the instruction execution timestamp with the working status of the material access device at the storage node.

[0110] For example, according to the re-allocation, the actual unloading quantity of transfer storage point A is 500 kg, and the actual unloading quantity of transfer storage point C is 100 kg. Convert these unloading quantities into specific instructions and ensure that the execution time of these instructions matches the working status of the material access device (such as forklifts, automatic guided vehicles, etc.) at the storage node. For example, if the material access device is handling other materials during a certain period, the execution time of the unloading instruction needs to avoid this period to ensure efficient material handling operations.

[0111] Step S1495, when it is detected that the actual unloading quantity at a certain storage node is not completed due to insufficient capacity, automatically trigger the emergency temporary storage allocation process for the remaining materials, and append the uncompleted quantity to the unloading task of the next reachable node.

[0112] For example, during the actual unloading process at transfer storage point A, due to some reasons (such as temporarily finding that part of the cargo storage area is occupied), only 400 kg can be actually unloaded, and the uncompleted 100 kg of materials needs to trigger the emergency temporary storage allocation process.

[0113] For example, in a possible implementation manner, step S1495 includes: Step S1495-1, obtain the quantity and type of the materials with uncompleted unloading in the target distribution path.

[0114] In the above example, the quantity of the materials with uncompleted unloading is 100 kg, and the material type is the positive electrode material.

[0115] Step S1495-2: Query the remaining loading capacity of all transportation vehicles that intersect with the current path within a preset time range and the planned storage nodes to be passed through.

[0116] Suppose that within the next hour (preset time range), there are two other transportation vehicles, T1 and T2, that intersect with the current path. The remaining loading capacity of transportation vehicle T1 is 200 kg, and the planned storage points to be passed through are transfer storage point D and the electrode preparation workshop; the remaining loading capacity of transportation vehicle T2 is 150 kg, and the planned storage points to be passed through are transfer storage point C and transfer storage point D.

[0117] Step S1495-3: Screen out the target transportation vehicles whose remaining loading capacity is greater than the quantity of unfinished materials and the storage nodes to be passed through include available temporary storage nodes.

[0118] The remaining loading capacities of both transportation vehicles T1 and T2 are greater than 100 kg. However, transfer storage point D passed through by transportation vehicle T1 can be used as a temporary storage node, while for transfer storage points C and D passed through by transportation vehicle T2, there may be other material storage arrangements or they may not be suitable for temporarily storing the positive electrode material. Therefore, transportation vehicle T1 is screened out as the target transportation vehicle.

[0119] Step S1495-4: Allocate the unfinished materials to the loading queue of the target transportation vehicle and update the unloading instructions of its associated storage nodes.

[0120] Allocate 100 kg of positive electrode material to the loading queue of transportation vehicle T1, and at the same time update the unloading instructions of transfer storage point D so that it can receive this 100 kg of positive electrode material when transportation vehicle T1 arrives.

[0121] Step S1495-5: Send a priority queue-jumping instruction to the target transportation vehicle so that it preferentially executes the unloading operation of the unfinished materials when arriving at the temporary storage node.

[0122] For example, in a possible implementation manner, the speed adjustment instruction sequence in the dynamic scheduling instruction is generated as follows: Set a speed adjustment trigger point at a preset distance before the starting point of each road segment of the target delivery path. For example, for road segment R1, set a speed adjustment trigger point 50 meters from the starting point.

[0123] When the transportation vehicle positioning module detects entry into the speed adjustment trigger point, extract the maximum allowable speed of the current road segment and the predicted passing score of the next road segment from the speed adjustment instruction sequence.

[0124] Assume the vehicle is an electric forklift. When its positioning system detects the speed adjustment trigger point for entering road segment R1, it obtains the maximum allowed speed for R1 as 30 km / h from the speed adjustment instruction sequence, and simultaneously obtains a predicted passing score of 2 for the next road segment R2.

[0125] Calculate the acceleration adjustment amount based on the difference between the current speed and the maximum allowed speed, and generate a control signal including the target speed value and the acceleration time interval.

[0126] Assume the current speed of the electric forklift is 20 km / h and the maximum allowed speed is 30 km / h. According to the speed change formula (speed change amount = final speed - initial speed), the speed change amount is 30 - 20 = 10 km / h. Based on the acceleration performance of the electric forklift (assuming it can accelerate by 10 km / h every 5 minutes), the acceleration time interval is calculated as 5 minutes, and a control signal including the target speed value of 30 km / h and the acceleration time interval of 5 minutes is generated.

[0127] In the transition interval before entering the next road segment, dynamically smooth the speed change curve based on the predicted passing score to avoid path deviation caused by sudden deceleration.

[0128] For example, since the predicted passing score for the next road segment R2 is 2, the corresponding speed limit is relatively low, and it is not possible to suddenly decelerate from 30 km / h to the speed limit of R2 directly. According to the predicted passing score and the deceleration performance of the electric forklift, in the transition interval before entering R2 (such as the interval 10 meters from the starting point of R2), gradually reduce the speed to make the speed change curve smooth, ensuring that the vehicle can accurately follow the path and avoiding path deviation caused by sudden deceleration.

[0129] For example, in a possible implementation, the method further includes: After outputting the target delivery path, real-time monitor the update frequency of the road traffic state matrix in the dynamic environment parameters.

[0130] For example, continuously collect data such as traffic flow and vehicle speed on the road through a sensor network, update the road traffic state matrix every 10 minutes, and count the update frequency.

[0131] When the update frequency exceeds a preset threshold, activate the online learning mode of the path optimization model, and use the latest path execution result as a training sample to incrementally update the model parameters.

[0132] Assume that the preset threshold is 3 updates per hour. When it is found that the update frequency of the road traffic status matrix reaches 4 times per hour within a certain time period, the online learning mode is activated. The execution result of a just-completed path segment (including information such as the actual driving speed, the residence time at each warehousing node, and whether arriving on time, etc.) is used as a new training sample to update the parameters of the path optimization model.

[0133] Re-evaluate the comprehensive path evaluation values of the unexecuted paths in the candidate optimized path queue according to the updated model parameters.

[0134] For example, there are three unexecuted paths in the candidate optimized path queue, namely path P1, P2, and P3. Using the updated model parameters, recalculate the comprehensive path evaluation values of these three paths. For path P1, recalculate the impact of factors such as its road traffic scoring sequence and the receiving capacity sequence of warehousing nodes on the path feasibility, and combine the new demand priority coefficient, etc., to obtain the new comprehensive path evaluation value.

[0135] When it is detected that the difference between the highest comprehensive path evaluation value after re-evaluation and the evaluation value of the original target delivery path exceeds the tolerance range, forcibly interrupt the current path execution and switch to the new target delivery path.

[0136] Assume that the tolerance range is set to 10%, the comprehensive path evaluation value of the original target delivery path is 80 points, and the comprehensive path evaluation value of path P1 after re-evaluation is 90 points. The difference is (90 - 80) / 80 = 12.5%, which exceeds the 10% tolerance range. At this time, forcibly interrupt the currently executing path and switch to path P1 as the new target delivery path to ensure the efficiency and accuracy of material delivery.

[0137] In a possible implementation manner, step S150 includes: Step S151, after the transportation tool starts, continuously receive the real-time traffic flow data uploaded by the sensor network and the capacity update information of the warehousing nodes.

[0138] For example, an electric forklift departs from the cathode material supplier and heads to the electrode preparation workshop. During the driving process, the control system of the electric forklift continuously receives the data from the sensor network deployed in the factory area. The sensor network precisely monitors the traffic flow situation on the road section, such as the number of vehicles passing through per minute, vehicle types, etc. information, and also closely monitors the capacity update information of each warehousing node. For a warehousing node, such as the transfer warehousing point A, the remaining capacity of the stored cathode material will change with the production process and the loading and unloading operations of other transportation tools, and these change information will be uploaded in real time and received by the control system of the electric forklift.

[0139] Step S152: When it is detected that the congestion level of any road segment in the current path exceeds the preset threshold or the remaining capacity of any warehousing node is lower than the critical value, a path replanning event is triggered.

[0140] Suppose the preset congestion level threshold is level 3 (the congestion level is set from 1 to 5, with 5 being the most congested). When the electric forklift travels to road segment R1 connecting transfer warehousing point A and transfer warehousing point B, it is learned from the data uploaded by the sensor network that the congestion level of this road segment has reached level 4, which exceeds the preset threshold. Or when the electric forklift is about to reach transfer warehousing point B, it receives the updated information on the remaining capacity of transfer warehousing point B and finds that its remaining capacity is lower than the critical value (suppose the critical value is 200 kg and the current remaining capacity is 150 kg). In this case, a path replanning event will be triggered.

[0141] Step S153: Pause the travel of the transport vehicle, and recalculate the set of alternative paths for the target delivery path according to the latest dynamic environment parameters.

[0142] The electric forklift stops at the current position, and the control system starts to replan the path. Based on the latest road traffic state matrix (including real-time congestion levels, average vehicle speeds, etc. of each road segment) and warehousing capacity state vector (latest remaining capacities of each warehousing node, working states of material handling equipment, etc.), the set of alternative paths is calculated. For example, the original target delivery path is from the cathode material supplier through transfer warehousing point A, transfer warehousing point B to the electrode preparation workshop. When replanning now, other possible paths will be considered, such as directly from the cathode material supplier to transfer warehousing point C and then to the electrode preparation workshop; or from the cathode material supplier through transfer warehousing point D, transfer warehousing point E to the electrode preparation workshop, etc. These possible paths constitute the set of alternative paths.

[0143] Step S154: Score each alternative path in the set of alternative paths by using a multi-objective evaluation strategy, and the multi-objective evaluation strategy includes delivery delay risk, transportation cost increment, and warehousing node load balance degree.

[0144] Step S155: Select the alternative path with the highest score as the updated target delivery path, and generate the corresponding dynamic scheduling instruction to adjust the travel direction and speed of the transport vehicle.

[0145] In a possible implementation manner, step S154 includes: Step S1541: Calculate the total estimated delivery time of each alternative path, and the total estimated delivery time is the sum of the travel times of each road segment and the loading and unloading times of the warehousing nodes.

[0146] For example, for the alternative path from the cathode material supplier directly to transfer storage point C and then to the electrode preparation workshop, first calculate the travel time of the road segments. Assume that the length of road segment R2 from the cathode material supplier to transfer storage point C is 5 kilometers. According to the real-time traffic flow data, the average vehicle speed on this road segment is 30 kilometers per hour. Then the travel time of this road segment is the road segment length divided by the average vehicle speed, that is, 5 divided by 30, and through the written calculation process, it is approximately 0.17 hours (here, for the convenience of expression, the time unit is converted to hours). Assume that the loading and unloading time at transfer storage point C is 0.5 hours according to past experience and the current working efficiency of the loading and unloading equipment. The length of road segment R3 from transfer storage point C to the electrode preparation workshop is 4 kilometers, and the average vehicle speed is 25 kilometers per hour. The travel time of this road segment is 4 divided by 25 which equals 0.16 hours. Assume that the loading and unloading time at the electrode preparation workshop is 0.3 hours. Then the total estimated delivery time for this alternative path is 0.17 + 0.5 + 0.16 + 0.3, and through the written calculation process of adding them in sequence, it is 1.13 hours.

[0147] Step S1542, estimate the transportation energy consumption cost of the alternative path according to the type of the transportation vehicle and the path length, where the transportation energy consumption cost is proportional to the path length and the unit distance energy consumption coefficient of the transportation vehicle.

[0148] For example, the unit distance energy consumption coefficient of an electric forklift is determined according to factors such as its performance and battery capacity, and assume it is 0.1 degree of electricity per kilometer. For the above alternative path from the cathode material supplier directly to transfer storage point C and then to the electrode preparation workshop, the path length is the sum of the lengths of road segments R2 and R3, that is, 5 + 4 = 9 kilometers. Then the transportation energy consumption cost is the path length multiplied by the unit distance energy consumption coefficient, that is, 9 multiplied by 0.1, and through the written calculation process, it is 0.9 degrees of electricity.

[0149] Step S1543, analyze the changing trend of the remaining capacity of the storage nodes involved in the alternative path. If the remaining capacity is lower than the safety threshold after the delivery is completed, increase the load balancing penalty value.

[0150] For example, for this alternative path, the current remaining capacity of transfer storage point C is 400 kilograms. Assume that the demand for cathode materials at the electrode preparation workshop is 300 kilograms. After the delivery is completed, the remaining capacity of transfer storage point C is 400 - 300 = 100 kilograms. If the safety threshold is set at 150 kilograms, then the remaining capacity of transfer storage point C is lower than the safety threshold after the delivery is completed. According to the pre-set rules, increase the load balancing penalty value. Assume that this penalty value is set at 10 (this value is determined according to the factory's emphasis on the load balancing of the storage nodes and the actual situation).

[0151] Step S1544: Normalize the total estimated delivery time, transportation energy consumption cost, and load balancing penalty value, and perform weighted summation according to preset weights to obtain the comprehensive score of each alternative path.

[0152] Suppose the preset weights are: the weight of the total estimated delivery time is 0.4, the weight of the transportation energy consumption cost is 0.3, and the weight of the load balancing penalty value is 0.3. For the total estimated delivery time of 1.13 hours, assume the maximum total estimated delivery time among all alternative paths is 2 hours. Then the normalized total estimated delivery time is 1.13 divided by 2, which equals 0.565. For the transportation energy consumption cost of 0.9 degrees of electricity, assume the maximum transportation energy consumption cost among all alternative paths is 1.5 degrees of electricity. Then the normalized transportation energy consumption cost is 0.9 divided by 1.5, which equals 0.6. For the load balancing penalty value of 10, assume the maximum load balancing penalty value among all alternative paths is 20. Then the normalized load balancing penalty value is 10 divided by 20, which equals 0.5. Then perform weighted summation according to the preset weights. The comprehensive score = 0.565 multiplied by 0.4 + 0.6 multiplied by 0.3 + 0.5 multiplied by 0.3. First, calculate the multiplication parts: 0.565 multiplied by 0.4 gives 0.226, 0.6 multiplied by 0.3 gives 0.18, and 0.5 multiplied by 0.3 gives 0.15. Then add these three results together to get 0.556.

[0153] Step S1545: Sort the set of alternative paths from high to low according to the comprehensive score, and select the alternative path ranked first as the updated target delivery path.

[0154] For example, according to the above calculation method, after calculating the comprehensive scores of all alternative paths, these alternative paths can be sorted from high to low according to the comprehensive score. Suppose after calculation and sorting, the alternative path from the cathode material supplier directly to transfer warehouse C and then to the electrode preparation workshop has the highest comprehensive score. Then select this path as the updated target delivery path. Then, according to the real-time traffic score of the road traffic state matrix and the remaining capacity data of the warehouse capacity state vector of this updated target delivery path, generate corresponding dynamic scheduling instructions to adjust the driving direction and speed of the transportation vehicle (electric forklift) to ensure that the electric forklift can efficiently deliver the cathode material to the electrode preparation workshop according to the new path.

[0155] In a possible implementation manner, the method further includes: Step S210: Monitor the actual material consumption rate of each production node of the target factory and compare it with the estimated consumption rate in the material demand information.

[0156] In this embodiment, during the production process of automotive power batteries, each production node has a clear material requirements plan, which includes an estimated consumption rate. Taking the electrode preparation production node as an example, for the positive electrode material, according to the production plan and past production experience, the estimated consumption rate per hour is 50 kilograms. During the actual production process, through the material monitoring system installed on the production line, the quantity of the positive electrode material actually consumed per hour is accurately counted. Suppose that within a certain consecutive few hours, the quantities of the positive electrode material actually consumed per hour are 60 kilograms, 62 kilograms, and 58 kilograms respectively. This indicates that the actual material consumption rate is higher than the estimated consumption rate.

[0157] Step S220, when the actual material consumption rate continuously exceeds the preset percentage of the estimated consumption rate, it is determined that there is a demand deviation in the corresponding production node, and a correction request is generated.

[0158] Continuing with the positive electrode material of the electrode preparation production node as an example, the preset percentage is set to 20%. The estimated consumption rate is 50 kilograms per hour, so the 20% excess is 50 multiplied by 20% which equals 10 kilograms, that is, when the actual consumption rate exceeds 60 kilograms per hour, it is regarded as having a demand deviation. Since the previously counted actual consumption rate continuously exceeds this value, it is determined that there is a demand deviation in the electrode preparation production node. At this time, a correction request is generated, and this correction request contains detailed deviation information, such as the quantity exceeding the estimated consumption rate per hour on average, etc.

[0159] Step S230, according to the deviation amount in the correction request and the current delivery progress, calculate the quantity of materials to be additionally delivered and the latest delivery time.

[0160] Suppose in the current delivery plan, the next delivery of the positive electrode material to the electrode preparation production node is expected to arrive in 2 hours. According to the situation where the previously counted actual consumption rate exceeds the estimated consumption rate, on average, it exceeds by 6 kilograms per hour (Calculation process: the total sum of the previously counted excess amounts divided by the number of counted hours, that is, (10 + 12 + 8) divided by 3 = 10, here taking the approximate value of 6 kilograms). Then the quantity of the positive electrode material expected to be additionally needed within these 2 hours is 6 multiplied by 2 = 12 kilograms, and this is the quantity of materials to be additionally delivered. The latest delivery time is the time when the next scheduled delivery arrives, that is, 2 hours later.

[0161] Step S240, screen out the paths that can insert additional materials midway from the initial delivery path set, and evaluate whether the total delay time of the path after insertion exceeds the preset tolerance range.

[0162] Step S250, if there is an insertable path, update the material distribution plan and the loading capacity of the transportation vehicle for the corresponding path, otherwise generate a new emergency delivery task and assign it to an idle transportation vehicle.

[0163] If there is an insertable path, such as the path mentioned above that passes by the storage point that can provide the positive electrode material after 0.5 hours. Update the material distribution plan for this path. Originally, it was planned to unload a certain amount of positive electrode material at a certain storage point. Now, it needs to be adjusted, and the additional 12 kg of positive electrode material should be reasonably distributed into the loading and unloading plans of each storage node. At the same time, the loading capacity of the transportation tool (electric forklift) is also updated accordingly, increasing by 12 kg on the basis of the original loading capacity.

[0164] If there is no insertable path, then a new emergency delivery task is generated. Query for idle transportation tools. Suppose an idle electric forklift is found. Assign the task to this emergency delivery task, set to obtain 12 kg of positive electrode material from the nearest positive electrode material storage point and send it to the electrode preparation production node at the fastest speed. At the same time, considering dynamic environment parameters such as the road traffic status matrix and the storage capacity status vector, plan the optimal delivery path.

[0165] In a possible implementation manner, step S240 includes: Step S241, obtain the remaining loadable capacity of all currently executing paths and the estimated time to reach each storage node.

[0166] For example, there is a delivery path from the positive electrode material supplier to the electrode preparation workshop that is currently being executed, and the transportation tool is an electric forklift. The electric forklift has currently loaded 1000 kg, and its upper loading limit is 1500 kg, so the remaining loadable capacity is 500 kg. Along this path, it is estimated to reach the transfer storage point A after 1 hour and the electrode preparation workshop after 3 hours.

[0167] Step S242, according to the latest delivery time of the additional material, calculate backward the latest time point when the material needs to be loaded onto the transportation tool.

[0168] According to the latest delivery time of the additional material (after 2 hours), calculate backward the latest time point when the material needs to be loaded onto the transportation tool. Since it takes 1 hour for the transportation tool to reach the transfer storage point A from the current position, then the additional material needs to be loaded onto the electric forklift within 1 hour at the latest.

[0169] Step S243, traverse the current positions of the transportation tools on each path and the remaining paths, and determine whether there is a transportation tool that passes by the storage node that can provide the additional material before the latest time point.

[0170] Suppose on another delivery path from the positive electrode material supplier to other production workshops, the transportation tool, the electric forklift, will pass by a storage point that can provide positive electrode material after 0.5 hours, and there is enough positive electrode material at this storage point for additional supply.

[0171] Step S244, if it exists, further verify whether the remaining loading capacity of the transportation vehicle meets the demand for additional materials.

[0172] If such a situation exists, further verify whether the remaining loading capacity of the transportation vehicle meets the demand for additional materials. The remaining loading capacity of the electric forklift on this path is 300 kg, while 12 kg of cathode material needs to be added. 300 kg is greater than 12 kg, meeting the capacity requirement.

[0173] Step S245, add the paths that simultaneously meet the time constraint and the capacity constraint to the candidate path list, and sort them in ascending order of the remaining time of the paths for selection.

[0174] Suppose there are other paths that meet the conditions. Sort them in ascending order according to the remaining time for each path to reach the destination. For example, if the remaining time of one path is 1.5 hours and that of another is 2 hours, the path with a remaining time of 1.5 hours will be ranked first.

[0175] In a possible implementation manner, the method further includes: Step S310, collect the operating parameters of the transportation vehicle in real time, where the operating parameters include the current position, remaining power, load status, and fault code.

[0176] During the distribution process of automotive power batteries, each transportation vehicle (such as an electric forklift) is equipped with a monitoring system. Taking an electric forklift that is performing the distribution task from the cathode material supplier to the electrode preparation workshop as an example, its current position can be accurately obtained through the positioning system on the vehicle, such as on a certain road section after departing from the supplier, and it is still 2 km away from the transfer storage point A. At the same time, its remaining power can be accurately obtained through the power monitoring sensor. Suppose the remaining power is 30% of the total power (assuming the total power is 100 units). In terms of the load status, it is known through the load sensor that 1200 kg of cathode material is currently loaded (the upper limit of the loading capacity is 1500 kg). And the vehicle control system will detect whether there is a fault code in real time. If a slight fault occurs in a certain motor of the electric forklift, a corresponding fault code will be generated.

[0177] Step S320, when it is detected that the remaining power is lower than the preset threshold, calculate the reachable range of the transportation vehicle according to the current path length of the target distribution path and the distance to the destination.

[0178] For example, assume that the preset threshold is 20% of the total power. When the remaining power of the electric forklift is detected to be 30%, although it has not yet dropped below the threshold, for early planning, the calculation of the reachable range is carried out. It is known that the distance from the current location to the destination (electrode preparation workshop) is 5 kilometers, and the power consumption per kilometer of the electric forklift at its current load state is 5 units (this value is determined based on factors such as the performance and load of the electric forklift). The current remaining power is 30 units (100 multiplied by 30%), so theoretically it can still travel 6 kilometers (30 divided by 5) at a power consumption of 5 units per kilometer.

[0179] Step S330, if the reachable range does not include the destination, generate a detour instruction from the nearest charging station or alternative transport tool dispatching point in the target delivery route.

[0180] For example, since it can theoretically still travel 6 kilometers and the distance to the destination is 5 kilometers, the reachable range includes the destination at this time, and no detour instruction needs to be generated. However, assume that if the distance to the destination is 7 kilometers, then the reachable range does not include the destination. At this time, query the nearest charging station or alternative transport tool dispatching point in the target delivery route. Assume that there is a charging station 3 kilometers away from the current location, and the system will generate a detour instruction to direct the electric forklift to go to this charging station for charging, and then continue to go to the destination, and re-plan the route during the detour to avoid congested sections, so as to reach the charging station at the fastest speed and reduce the overall delivery delay.

[0181] Step S340, when a fault code is received, analyze the fault type and match a predefined emergency handling plan, where the emergency handling plan includes a material transfer path, the priority of calling alternative tools, and the estimated time for fault repair.

[0182] After the control system of the electric forklift receives the fault code of a minor motor fault, the system analyzes the fault type and determines it as a fault type where the motor performance has declined but it can still run for a short distance. Match the predefined emergency handling plan, which stipulates the material transfer path in this plan. For example, there is a temporary material storage point near the current location, and some of the positive electrode materials on the electric forklift can be transferred to this temporary storage point first. Regarding the priority of calling alternative tools, if there are other idle electric forklifts, according to the pre-set priority, give priority to calling the electric forklift that is the nearest to the current location and has an appropriate loading capacity. The estimated time for fault repair is estimated to be 1 hour based on the past maintenance experience of similar faults and the arrangement of maintenance personnel.

[0183] Step S350, start the material transfer process according to the emergency handling plan, and update the dynamic environment parameters of the affected target delivery route to trigger global path re-planning.

[0184] Specifically, in accordance with the emergency response plan, initiate the material transfer process, and arrange for staff to transfer some of the positive electrode materials (such as 500 kilograms) on the electric forklift to the temporary storage point. At the same time, due to the occurrence of this incident, the dynamic environmental parameters of the target delivery route have changed. For example, the road traffic status has been affected by the temporary occupation of the material transfer vehicle, and the capacity status of the warehousing node (temporary storage point) has also changed. Feed these updated dynamic environmental parameters back into the system to trigger global route replanning. Recalculate the optimal delivery route from the temporary storage point to the electrode preparation workshop, considering the new road traffic status, the capacity of the warehousing node, and the scheduling of other transportation tools, to ensure that the positive electrode materials can be delivered to the electrode preparation production node on time and efficiently, and to ensure that the production process of automotive power batteries is not affected too much.

[0185] Figure 2 FIG. shows a schematic diagram of exemplary hardware and software components of an intelligent distribution decision-making system 100 for production materials that can implement the ideas of the present application provided in some embodiments of the present application. For example, the processor 120 can be used on the intelligent distribution decision-making system 100 for production materials and is used to execute the functions in the present application.

[0186] The intelligent distribution decision-making system 100 for production materials can be a general-purpose server or a special-purpose server, both of which can be used to implement the intelligent distribution decision-making method for production materials of the present application. Although only one server is shown in the present application, for convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0187] For example, the intelligent distribution decision-making system 100 for production materials can include a network port 110 connected to the network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as disks, ROM, or RAM, or any combination thereof. Exemplarily, the intelligent distribution decision-making system 100 for production materials can also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The intelligent distribution decision-making system 100 for production materials also includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0188] For ease of description, only one processor is described in the intelligent distribution decision-making system 100 for production materials. However, it should be noted that the intelligent distribution decision-making system 100 for production materials in this application may also include multiple processors. Therefore, the steps performed by one processor described in this application can also be jointly performed or separately performed by multiple processors. For example, if the processor of the intelligent distribution decision-making system 100 for production materials performs step A and step B, it should be understood that step A and step B can also be jointly performed by two different processors or separately performed in one processor. For example, the first processor performs step A, the second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0189] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are preset. When the processor executes the computer-executable instructions, the above intelligent distribution decision-making method for production materials is implemented.

[0190] It should be noted that, in order to simplify the description of the present invention disclosure and thus help the understanding of one or more embodiments of the invention, in the previous description of the embodiments of the present invention, sometimes multiple features are merged into one embodiment, drawing or description thereof.

Claims

1. An intelligent distribution decision method for production materials, characterized in that: The method comprises: Obtaining material demand information of each production node of the target factory within a preset period, wherein the material demand information includes material type, demand priority and demand time window; Based on the real-time traffic data and storage layout data of the area where the target factory is located, determining dynamic environmental parameters corresponding to the material demand information, wherein the dynamic environmental parameters include road traffic status and storage node capacity; Generate an initial distribution path set according to the demand priority and demand time window in the material demand information, each initial distribution path in the initial distribution path set is associated with at least one transportation tool and a corresponding material allocation plan; Inputting the dynamic environment parameters corresponding to the initial distribution path set and the material demand information into a pre-trained path optimization model, outputting a target distribution path and a dynamic scheduling instruction, wherein the dynamic scheduling instruction is used to adjust the loading capacity and driving speed of the transport vehicle; According to the target delivery path and the dynamic dispatching instructions, the transport vehicle is controlled to execute material delivery, and during the delivery process, the path is replanned based on the dynamic environmental parameters updated in real time.

2. The intelligent distribution decision method for production materials according to claim 1 is characterized in that: The determining of the dynamic environmental parameters corresponding to the material demand information based on the real-time traffic data and warehouse layout data of the area where the target factory is located includes: Collecting real-time traffic flow data through a sensor network deployed in the target factory area, and extracting the average vehicle speed and congestion level of each road section within a preset time interval; Obtaining the location coordinates, remaining storage capacity and working status of material storage and retrieval equipment of each storage node in the storage layout data; Constructing a road traffic status matrix according to the congestion level and average vehicle speed in the real-time traffic flow data, wherein the rows of the road traffic status matrix represent road segment identifiers, the columns represent time intervals, and the matrix elements are traffic scores of the corresponding road segments in the time intervals; Based on the remaining storage capacity of the storage node and the working status of the material storage and access equipment, a storage capacity state vector is generated, each element of the storage capacity state vector represents the amount of material that can be received by the corresponding storage node at the current moment; The road traffic state matrix and the storage capacity state vector are associated and mapped to obtain the road traffic state and storage node capacity in the dynamic environment parameters.

3. The intelligent distribution decision method for production materials according to claim 1 is characterized in that: The generating of an initial delivery path set according to the demand priority and demand time window in the material demand information includes: According to the demand priority in the material demand information, each production node is divided into an emergency demand node and a non-emergency demand node, and the start time of the demand time window of the emergency demand node is extracted; Based on the difference between the start time of the demand time window and the current system time, the delivery urgency of each emergency demand node is calculated, and the delivery urgency is inversely proportional to the difference; According to the delivery urgency and the storage node positions in the storage layout data, a greedy algorithm is used to generate an initial candidate path, wherein the initial candidate path covers all the urgent demand nodes and has the shortest path length; Traversing each storage node in the initial candidate path, determining whether the remaining storage capacity of the storage node meets the material demand of the corresponding emergency demand node, and if not, dynamically allocating materials from adjacent storage nodes and updating the initial candidate path; The initial candidate paths that meet all the urgent demand nodes are added to the initial distribution path set, and a transportation tool type and a loading capacity upper limit are allocated to each initial candidate path.

4. The intelligent distribution decision method for production materials according to claim 1 is characterized in that: The step of inputting the dynamic environment parameters corresponding to the initial delivery path set and the material demand information into a pre-trained path optimization model and outputting a target delivery path and a dynamic scheduling instruction includes: Normalizing each initial delivery path in the initial delivery path set, extracting the type of transport tool, the upper limit of the load, the identification sequence of the road segments passed through, and the identification sequence of the associated storage nodes corresponding to each initial delivery path as structured path description data; The structured path description data is concatenated with the road traffic state matrix and storage capacity state vector in the dynamic environment parameters to generate a multi-dimensional input feature vector for each initial delivery path, wherein the multi-dimensional input feature vector includes a path length weight, a road traffic score sequence, a storage node receiving capacity sequence, and a transportation tool capacity ratio; Extracting spatial correlation features in the multi-dimensional input feature vector through the convolutional layer in the pre-trained path optimization model, and capturing the dynamic change pattern of the road traffic score sequence over time intervals using a temporal attention mechanism; Based on the spatial correlation characteristics and dynamic change patterns, a path feasibility score of each initial delivery path is calculated, wherein the path feasibility score is positively correlated with the average value of the road traffic score sequence and negatively correlated with the minimum value of the storage node receiving capacity sequence; The initial delivery path set is filtered according to the path feasibility score, the initial delivery paths with scores lower than a preset threshold are eliminated, and the remaining paths are sorted in descending order of scores to generate a candidate optimization path queue; Traversing each candidate optimization path in the candidate optimization path queue, adjusting the path feasibility score in combination with the demand priority coefficient in the material demand information, and generating a priority-weighted comprehensive path evaluation value; Select the candidate optimization path with the highest comprehensive path evaluation value as the target delivery path, and analyze its corresponding transportation tool type and loading capacity upper limit; According to the real-time traffic score of the road traffic state matrix in the target delivery path and the remaining capacity data of the storage capacity state vector, the maximum allowable speed of the transportation tool on each road section and the optimal stay time at the storage node are dynamically calculated; The dynamic scheduling instruction is generated based on the maximum allowed speed and the optimal stay time, and the dynamic scheduling instruction includes a speed adjustment instruction sequence and a load redistribution instruction, wherein the load redistribution instruction dynamically allocates the unloading ratio of each node according to the remaining capacity data of the storage nodes passed through.

5. The intelligent distribution decision method for production materials according to claim 2 is characterized in that: The controlling the transport vehicle to perform material distribution according to the target distribution path and the dynamic dispatching instruction, and performing path replanning based on the real-time updated dynamic environmental parameters during the distribution process, includes: After the transportation tool is started, continuously receiving the real-time traffic flow data and capacity update information of the storage nodes uploaded by the sensor network; When it is detected that the congestion level of any road segment in the current path exceeds the preset threshold or the remaining capacity of any storage node is lower than the critical value, the path replanning event is triggered; Pause the travel of the transport vehicle, and recalculate the set of candidate routes for the target delivery route based on the latest dynamic environmental parameters; Using a multi-objective evaluation strategy to score each alternative path in the set of alternative paths, the multi-objective evaluation strategy includes delivery delay risk, transportation cost increment, and storage node load balance; The alternative route with the highest score is selected as the updated target delivery route, and the corresponding dynamic scheduling instructions are generated to adjust the driving direction and speed of the transportation tool.

6. The intelligent distribution decision method for production materials according to claim 5 is characterized in that: The adopting of a multi-objective evaluation strategy to score each candidate path in the candidate path set includes: Calculate the total estimated delivery time for each alternative route, where the total estimated delivery time is the cumulative sum of the travel time of each road segment and the loading and unloading time of the storage node; According to the type of the transport tool and the path length, the transport energy consumption cost of the alternative path is estimated, wherein the transport energy consumption cost is proportional to the path length and the unit distance energy consumption coefficient of the transport tool; Analyze the change trend of the remaining capacity of the storage nodes involved in the alternative path, and if the remaining capacity is lower than the safety threshold after the delivery is completed, increase the load balancing penalty value; The total estimated delivery time, transportation energy consumption cost and load balancing penalty value are normalized and weighted summed according to preset weights to obtain a comprehensive score for each alternative route; The set of candidate routes is sorted from high to low according to the comprehensive score, and the candidate route ranked first is selected as the updated target delivery route.

7. The intelligent distribution decision method for production materials according to claim 1 is characterized in that: The method further comprises: Monitor the actual material consumption rate of each production node of the target factory and compare it with the estimated consumption rate in the material demand information; When the actual material consumption rate continuously exceeds the preset percentage of the estimated consumption rate, it is determined that there is a demand deviation in the corresponding production node, and a correction request is generated; Calculate the quantity of materials to be added and the latest delivery time according to the deviation in the correction request and the current delivery progress; Filtering out a path where additional materials can be inserted midway from the initial delivery path set, and evaluating whether the total delay time of the path after the insertion exceeds a preset tolerance range; If there is an insertable path, the material allocation plan and the loading capacity of the transport vehicle of the corresponding path will be updated; otherwise, a new emergency delivery task will be generated and assigned to an idle transport vehicle.

8. The intelligent distribution decision method for production materials according to claim 7 is characterized in that: The step of selecting a path where additional materials can be inserted midway from the initial delivery path set includes: Obtain the remaining loading capacity of all currently executing paths and the estimated arrival time at each storage node; According to the latest delivery time of the additional materials, reversely calculate the latest time point when the materials need to be loaded onto the transportation vehicle; Traverse the current position of the transport tool and the remaining path of each path, and determine whether there is a transport tool that passes by a storage node that can provide additional materials before the latest time point; If so, further verify whether the remaining loading capacity of the transport vehicle can meet the demand for additional materials; The paths that meet both the time constraint and the capacity constraint are added to the candidate path list and arranged in ascending order according to the remaining time of the path for selection.

9. The intelligent distribution decision method for production materials according to claim 1 is characterized in that: The method further comprises: Collecting the operating parameters of the transportation vehicle in real time, the operating parameters including current location, remaining power, load status and fault code; When it is detected that the remaining power is lower than a preset threshold, the reachable range of the transportation tool is calculated according to the current path length of the target delivery path and the distance to the destination; If the reachable range does not include the destination, generating a detour instruction from the nearest charging station or alternate transportation tool dispatch point in the target delivery path; When a fault code is received, the fault type is analyzed and matched with a predefined emergency treatment plan, which includes a material transfer path, a priority for calling spare tools, and an estimated time for fault repair; The material transfer process is initiated according to the emergency handling plan, and the dynamic environmental parameters of the affected target distribution path are updated to trigger global path replanning.

10. An intelligent distribution decision system for production materials, characterized in that: The intelligent distribution decision system for production materials includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the intelligent distribution decision method for production materials described in any one of claims 1 to 9 above.

Images