Logistics transportation optimal path planning method

Abstract

The invention relates to the technical field of path planning, and discloses a logistics transportation optimal path planning method, which comprises the following steps: acquiring real-time traffic flow data, weather forecast information and external factors of accident reports from multiple channels through Internet of Things equipment, a mobile application and an API (Application Program Interface) to form an original data set; preprocessing the original data set and storing the original data set in a distributed database to obtain preprocessed traffic data; and based on the original data set and the preprocessed traffic data, constructing a traffic prediction model based on a machine learning algorithm, and taking the preprocessed traffic data and the original data set as a training set. According to the optimal path planning method for logistics transportation, a neural network architecture comprising a plurality of layers of LSTM units is designed, a sliding window method is adopted to extract a plurality of subsequences from the neural network architecture to serve as training samples, the difference between a predicted value and a true value is measured in combination with a mean square error and a loss function of a penalty term, and more accurate traffic prediction is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and in particular to an optimal path planning method for logistics transportation. Background Art

[0002] Route planning is a key link in the logistics transportation process, aiming to find the best route from the starting point to the end point. The selection of the best route needs to consider not only the distance factor, but also time efficiency, economic cost, environmental friendliness and other aspects. However, in traditional methods, information sources such as traffic flow, weather forecasts and accident reports are scattered and difficult to integrate efficiently, resulting in incomplete or delayed data, affecting the accuracy of route planning. In addition, traditional traffic prediction models often rely on simple statistical analysis and cannot effectively capture the long-term dependencies in traffic data, resulting in inaccurate prediction results. At the same time, when faced with emergencies, traditional methods often have slow responses and cannot adjust route planning in time, increasing transportation risks and costs. Summary of the invention

[0003] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0004] To achieve the above object, the present invention provides the following technical solutions: A method for optimal path planning for logistics transportation comprises: obtaining real-time traffic flow data, weather forecast information and external factors of accident reports from multiple channels through Internet of Things devices, mobile applications and API interfaces to form an original data set; preprocessing the original data set and storing it in a distributed database to obtain preprocessed traffic data; constructing a traffic prediction model based on a machine learning algorithm based on the original data set and the preprocessed traffic data, using the preprocessed traffic data and the original data set as training sets, inputting them into the traffic prediction model, training the model, predicting traffic conditions in a future period of time, and obtaining traffic prediction results; combining the traffic prediction results, using a graph theory algorithm to calculate multiple candidate paths from a starting point to an end point, and obtaining a preliminary optimized path; setting and quantifying an optimization target, based on the preliminary optimized path, selecting a non-dominated solution as a final recommended solution through Pareto frontier analysis, and obtaining an optimal path solution; establishing a real-time monitoring system to monitor the vehicle operation status, identify abnormal conditions and trigger early warning notifications, dynamically adjust the path planning according to actual traffic conditions, and update the optimal path solution.

[0005] As a further solution of the present invention: the real-time traffic flow data, weather forecast information and external factors of accident reports are obtained from multiple channels through IoT devices, mobile applications and API interfaces to form an original data set. The specific steps are: constructing an HTTP GET request for the Amap API, including a timestamp, a geographic area code, and a required parameter list; the received data packet is in JSON format and contains a large amount of redundant information. A preliminary screening method is used to retain only the parts related to the current task, namely, traffic flow, weather conditions and accident location, and the expression is: ; in, Indicates the time interval for the region The accumulated amount of raw data collected, It's time point Time Zone The traffic flow function in The area in the same time period The weather condition function, It is a function of the location where the accident occurred; the data after preliminary screening is checked for integrity to ensure that each data point meets the predefined standards; the data from different time points are integrated into a continuous time series and grouped according to geographical location to form the original data set.

[0006] As a further solution of the present invention: the original data set is preprocessed and stored in a distributed database to obtain preprocessed traffic data. The specific steps are: take the original data set through Spark RDD, apply cleaning rules to remove outliers and duplicate records, and fill in missing data; for data in different formats, use standardized conversion functions to unify the data into a consistent numerical range; based on the cleaned data, extract key features, and use historical data analysis to mine potential patterns and add new feature dimensions. The expression is: ; in, represents the final extracted feature vector, is the average vehicle speed function, is the congestion index function, is an additional characteristic function, and is to adjust the weight parameter; get the preprocessed traffic data .

[0007] As a further solution of the present invention: the traffic prediction model based on the original data set and the preprocessed traffic data is constructed based on the machine learning algorithm, and the specific steps are: merging the preprocessed traffic data and the original data set into a complete training set; using the sliding window method to extract multiple subsequences as training samples, each subsequence contains data of several time steps; designing a neural network architecture containing a multi-layer LSTM unit, each LSTM unit is responsible for processing the data of one time step, and passing the context information through the hidden state, and adding a fully connected layer for finally outputting the prediction result, the expression is: ; in, Indicates at time The hidden state of is the input feature vector, is the long short-term memory network function; the loss function combining the mean square error and the penalty term is used to measure the difference between the predicted value and the true value, and the expression is: ; in, is the total loss function, Based on the hidden state The predicted traffic conditions after the fully connected layer and activation function, is the actual observed traffic condition, is the sample size, is the penalty coefficient, is the feature dimension, is the gradient of the loss function with respect to the predicted value; using the defined loss function The LSTM model is trained with the Adam optimizer. During the training process, the model parameters are updated through the back propagation algorithm to gradually reduce the loss function. The trained LSTM model is used to predict the traffic conditions in the future. The prediction results for the next few time steps are expressed as: ; in, Indicates future time The predicted traffic conditions in the area, is the time after LSTM processing The hidden state of is the weight matrix of the fully connected layer, is the bias term, is an activation function.

[0008] As a further solution of the present invention: the above-mentioned combination of traffic prediction results and the use of graph theory algorithm to calculate multiple candidate paths from the starting point to the end point to obtain a preliminary optimized path, the specific steps are: based on the pre-processed traffic data and traffic prediction results, construct a weighted directed graph Represent the traffic network, initialize the A* search algorithm parameters, design a heuristic function that combines geographic distance and predicted traffic conditions, and for each node, calculate the actual cost from the starting node to the current node. The expression is as follows: ; ; ; in, It is the edge The weight of is the length of the road section, is the predicted traffic condition, and is the adjustment coefficient, is a function of weather conditions, Is a node The heuristic estimate of From the starting node To Node The straight-line distance, is the adjustment coefficient, Is a node Forecasted traffic conditions in your area, From the starting node To Node The actual cost of From the starting node To Node A collection of paths, Is the path The weight of each edge in ; using weighted graph and the heuristic function , execute the A* search algorithm to find the starting node To the target node The multiple candidate paths of are expressed as: ; in, Is a node According to the results of the A* search algorithm, multiple candidate paths from the starting point to the end point are generated, and the cost of each path is calculated. One or more paths with the lowest cost are selected as the preliminary optimization path. The expression is: ; in, is the initial optimization path, is the set of all candidate paths from the starting point to the end point, Is the path The weight of each edge in .

[0009] As a further solution of the present invention: the optimization target is set and quantified, and based on the preliminary optimization path, the non-dominated solution is selected as the final recommended solution through Pareto front analysis to obtain the optimal path solution. The specific steps are: clarifying the optimization target in the logistics transportation process, including time efficiency, economic cost and environmental friendliness; constructing a comprehensive evaluation function to integrate the three optimization targets, and introducing interaction rules to achieve a reasonable balance between different targets. The expression is: ; in, is a candidate path variable, is the time distribution function along the path, represents the path length, is the time sensitivity coefficient, and are the economic cost weight and its elasticity index, It is the environmental influencing factor. is the cost per trip, is the carbon dioxide emissions per kilometer, is the number of road sections, m is the total mileage, is the average economic cost, is the average carbon dioxide emissions per kilometer, is a comprehensive evaluation function; using the comprehensive evaluation function, all preliminary optimization paths are scored, and a group of non-dominated solutions are screened out from the paths using the Pareto frontier analysis method; after determining the non-dominated solutions, the optimal path solution is selected based on time efficiency > economic cost > environmental friendliness.

[0010] As a further solution of the present invention: the real-time monitoring system is established to monitor the vehicle operation status, identify abnormal situations and trigger early warning notifications, dynamically adjust the path planning according to the actual traffic conditions, and update the optimal path plan. The specific steps are: based on the original data set, the abnormal situation that affects driving safety or efficiency is identified through the abnormal detection model; when an abnormal warning is received or a significant change in traffic conditions is detected, the system immediately starts the dynamic path adjustment algorithm and dynamically adjusts the path planning; based on the results of the dynamic path adjustment algorithm, the system selects one or more new paths with the lowest cost as the final recommended plan, and sends it to the driver for execution in a timely manner. When the new best path is significantly better than the current path, the optimal path plan is updated immediately, otherwise the original path remains unchanged.

[0011] As a further solution of the present invention: when receiving an abnormal warning or detecting a significant change in traffic conditions, the system immediately starts a dynamic path adjustment algorithm and dynamically adjusts the path planning expression to: ; in, It is a new path The cost, is a set of candidate paths from the current location to the destination, is the weight of each edge in the original path, is the traffic prediction influencing factor, It is the change in the traffic conditions of the corresponding road section in the future time period.

[0012] Compared with the prior art, the present invention has the following beneficial effects: By designing a neural network architecture containing multiple layers of LSTM units and using the sliding window method to extract multiple subsequences as training samples, the difference between the predicted value and the true value is measured by combining the loss function of the mean square error and the penalty term. This provides more accurate traffic forecasts, helps to plan the optimal route in advance, reduce transportation time and cost, and improve logistics efficiency. Based on the anomaly detection model, abnormal situations that affect driving safety or efficiency are identified. Once an abnormal warning is received or a significant change in traffic conditions is detected, the dynamic path adjustment algorithm is immediately started and the path planning is dynamically adjusted, which enhances the robustness and adaptability of the system. It not only improves the ability to respond to emergencies, but also can quickly adjust the path according to real-time traffic changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The figure is a flow chart of an optimal path planning method for logistics transportation. DETAILED DESCRIPTION

[0014] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, the specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings.

[0015] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0016] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0017] Example 1 See also Figure 1 , which is the first embodiment of the present invention, and provides a method for optimal path planning for logistics transportation, comprising: S1. Obtain real-time traffic flow data, weather forecast information, and external factors of accident reports from multiple channels through IoT devices, mobile applications, and API interfaces to form an original data set; Specifically, real-time traffic flow data, weather forecast information, and external factors of accident reports are obtained from multiple channels through IoT devices, mobile applications, and API interfaces to form an original data set. The specific steps are as follows: construct an HTTP GET request for the Amap API, including a timestamp, geographic area code, and a required parameter list; the received data packet is in JSON format and contains a lot of redundant information. A preliminary screening method is used to retain only the parts related to the current task, namely traffic flow, weather conditions, and accident locations. The expression is: ; in, Indicates the time interval for the region The accumulated amount of raw data collected, It's time point Time Zone The traffic flow function in The area in the same time period The weather condition function, It is a function of the location where the accident occurred; the data after preliminary screening is checked for integrity to ensure that each data point meets the predefined standards; the data from different time points are integrated into a continuous time series and grouped according to geographical location to form the original data set.

[0018] It should be noted that the data obtained through IoT devices, mobile applications and API interfaces not only covers information such as traffic flow, weather forecasts and accident reports, but also contains key metadata such as timestamps and geographic area codes. Metadata is crucial for subsequent time series analysis and geographic location grouping. In addition, the use of preliminary screening methods to remove redundant information ensures the streamlining and pertinence of the data and improves data processing efficiency. The final raw data set provides high-quality basic information for subsequent steps, ensuring the accuracy and reliability of the entire path planning system.

[0019] S2, preprocessing the original data set and storing it in a distributed database to obtain preprocessed traffic data; Specifically, the original data set is preprocessed and stored in a distributed database to obtain preprocessed traffic data. The specific steps are: take the original data set through Spark RDD, apply cleaning rules to remove outliers and duplicate records, and fill in missing data; for data in different formats, use standardized conversion functions to unify the data into a consistent numerical range; based on the cleaned data, extract key features, and use historical data analysis to mine potential patterns and add new feature dimensions. The expression is: ; in, represents the final extracted feature vector, is the average vehicle speed function, is the congestion index function, is an additional characteristic function, and is to adjust the weight parameter; get the preprocessed traffic data .

[0020] It should be noted that when preprocessing the original data set, Spark RDD technology was used to achieve efficient large-scale data processing capabilities, which can quickly remove outliers and duplicate records and fill in missing data, thereby improving the quality of the data. The application of standardized conversion functions unifies the data formats from different sources, facilitating subsequent feature extraction and model training. The extracted key features are combined with new feature dimensions mined through historical data analysis to enhance the model's ability to predict future traffic conditions. At the same time, the use of distributed databases ensures data security and access speed, providing strong support for logistics and transportation with high real-time requirements.

[0021] S3. Based on the original data set and the preprocessed traffic data, a traffic prediction model based on a machine learning algorithm is constructed. The preprocessed traffic data and the original data set are used as training sets and input into the traffic prediction model. The model is trained to predict the traffic conditions in the future and obtain traffic prediction results. Specifically, based on the original data set and the preprocessed traffic data, a traffic prediction model based on a machine learning algorithm is constructed. The specific steps are: merge the preprocessed traffic data and the original data set into a complete training set; use the sliding window method to extract multiple subsequences as training samples, each subsequence contains several time steps of data; design a neural network architecture containing multiple layers of LSTM units, each LSTM unit is responsible for processing one time step of data, and passes context information through hidden states, and adds a fully connected layer to finally output the prediction result. The expression is: ; in, Indicates at time The hidden state of is the input feature vector, is the long short-term memory network function; the loss function combining the mean square error and the penalty term is used to measure the difference between the predicted value and the true value, and the expression is: ; in, is the total loss function, Based on the hidden state The predicted traffic conditions after the fully connected layer and activation function, is the actual observed traffic condition, is the sample size, is the penalty coefficient, is the feature dimension, is the gradient of the loss function with respect to the predicted value; using the defined loss function The LSTM model is trained with the Adam optimizer. During the training process, the model parameters are updated through the back propagation algorithm to gradually reduce the loss function. The trained LSTM model is used to predict the traffic conditions in the future. The prediction results for the next few time steps are expressed as: ; in, Indicates future time The predicted traffic conditions in the area, is the time after LSTM processing The hidden state of is the weight matrix of the fully connected layer, is the bias term, is an activation function.

[0022] It should be noted that in the process of constructing a traffic prediction model based on machine learning algorithms, the effective application of the sliding window method enables the model to capture short-term fluctuations and long-term trends in traffic data, thereby improving prediction accuracy. The design of multi-layer LSTM units can effectively handle complex dependencies in time series data, while the fully connected layer is responsible for mapping hidden states to specific prediction results. The introduction of the loss function of mean square error and penalty terms not only measures the difference between the predicted value and the true value, but also prevents the occurrence of overfitting. The selection of the Adam optimizer and the application of the back-propagation algorithm further accelerate the convergence of the model and improve the training efficiency. The final traffic prediction results provide an important decision-making basis for path planning and significantly improve the predictability and controllability of logistics transportation.

[0023] S4. Combined with the traffic prediction results, a graph theory algorithm is used to calculate multiple candidate paths from the starting point to the end point to obtain a preliminary optimized path; Specifically, combined with the traffic prediction results, a graph theory algorithm is used to calculate multiple candidate paths from the starting point to the end point to obtain a preliminary optimized path. The specific steps are: based on the preprocessed traffic data and traffic prediction results, a weighted directed graph is constructed. Represent the traffic network, initialize the A* search algorithm parameters, design a heuristic function that combines geographic distance and predicted traffic conditions, and for each node, calculate the actual cost from the starting node to the current node. The expression is as follows: ; ; ; in, It is the edge The weight of is the length of the road section, is the predicted traffic condition, and is the adjustment coefficient, is a function of weather conditions, Is a node The heuristic estimate of From the starting node To Node The straight-line distance, is the adjustment coefficient, Is a node Forecasted traffic conditions in your area, From the starting node To Node The actual cost of From the starting node To Node A collection of paths, Is the path The weight of each edge in ; using weighted graph and the heuristic function , execute the A* search algorithm to find the starting node To the target node The multiple candidate paths of are expressed as: ; in, Is a node According to the results of the A* search algorithm, multiple candidate paths from the starting point to the end point are generated, and the cost of each path is calculated. One or more paths with the lowest cost are selected as the preliminary optimization path. The expression is: ; in, is the initial optimization path, is the set of all candidate paths from the starting point to the end point, Is the path The weight of each edge in .

[0024] It should be noted that in the process of calculating the preliminary optimized path using graph theory algorithms in combination with traffic forecast results, a weighted directed graph is constructed and the A* search algorithm parameters are initialized to ensure the scientificity and rationality of the path calculation. The design of the heuristic function comprehensively considers the geographical distance and predicted traffic conditions, improves the search efficiency, and calculates the actual cost of each node, which fully reflects the actual difficulty of the road section and ensures the optimality of the path selection. The generated multiple candidate paths and their cost evaluations lay the foundation for the subsequent Pareto frontier analysis, making the final recommended plan more reasonable and diversified, and meeting the path optimization under different logistics and transportation needs.

[0025] S5. Set and quantify the optimization goal, select the non-dominated solution as the final recommended solution based on the preliminary optimization path through Pareto frontier analysis, and obtain the optimal path solution; Specifically, the optimization goal is set and quantified. Based on the preliminary optimization path, the non-dominated solution is selected as the final recommended solution through Pareto frontier analysis to obtain the optimal path solution. The specific steps are: clarify the optimization goals in the logistics transportation process, including time efficiency, economic cost and environmental friendliness; construct a comprehensive evaluation function to integrate the three optimization goals, and introduce interactive rules to achieve a reasonable balance between different goals. The expression is: ; in, is a candidate path variable, is the time distribution function along the path, represents the path length, is the time sensitivity coefficient, and are the economic cost weight and its elasticity index, It is the environmental influencing factor. is the cost per trip, is the carbon dioxide emissions per kilometer, is the number of road sections, m is the total mileage, is the average economic cost, is the average carbon dioxide emissions per kilometer, is a comprehensive evaluation function; using the comprehensive evaluation function, all preliminary optimization paths are scored, and a group of non-dominated solutions are screened out from the paths using the Pareto frontier analysis method; after determining the non-dominated solutions, the optimal path solution is selected based on time efficiency > economic cost > environmental friendliness.

[0026] It should be noted that when setting and quantifying the optimization goals, the three major goals of time efficiency, economic cost and environmental friendliness are clearly defined, which reflects the multiple pursuits of speed, economy and environmental protection in modern logistics and transportation. The constructed comprehensive evaluation function not only integrates the three optimization goals, but also achieves a balance between different goals by introducing interaction rules. The Pareto frontier analysis method is used to screen non-dominated solutions to ensure that the recommended path has the best comprehensive performance in all optimization goals. The optimal path plan is selected according to priority, which not only takes into account the most urgent needs at present (such as time efficiency), but also takes into account long-term interests (such as environmental impact), providing a comprehensive and flexible path planning solution for logistics and transportation.

[0027] S6. Establish a real-time monitoring system to monitor the vehicle operation status, identify abnormal situations and trigger early warning notifications, dynamically adjust the route planning according to the actual traffic conditions, and update the optimal route plan; Specifically, a real-time monitoring system is established to monitor the vehicle operation status, identify abnormal situations and trigger early warning notifications, dynamically adjust the path planning according to the actual traffic conditions, and update the optimal path plan. The specific steps are: based on the original data set, identify abnormal situations that affect driving safety or efficiency through the anomaly detection model; when receiving an abnormal warning or detecting a significant change in traffic conditions, the system immediately starts the dynamic path adjustment algorithm and dynamically adjusts the path planning; based on the results of the dynamic path adjustment algorithm, the system selects one or more new paths with the lowest cost as the final recommended solution and sends them to the driver for execution in a timely manner. When the new best path is significantly better than the current path, the optimal path plan is updated immediately, otherwise the original path remains unchanged; when receiving an abnormal warning or detecting a significant change in traffic conditions, the system immediately starts the dynamic path adjustment algorithm and dynamically adjusts the path planning. The expression is: ; in, It is a new path The cost, is a set of candidate paths from the current location to the destination, is the weight of each edge in the original path, is the traffic prediction influencing factor, It is the change in the traffic conditions of the corresponding road section in the future time period.

[0028] It should be noted that the established real-time monitoring system has greatly enhanced the safety and controllability of the transportation process by continuously monitoring the vehicle's operating status and identifying and warning of abnormal situations. Once an abnormality is detected or traffic conditions change significantly, the dynamic path adjustment algorithm that is immediately activated can recalculate the optimal path in a short time to ensure the continuity and efficiency of the transportation task. The cost evaluation mechanism of the new path ensures that the adjusted path still maintains a low-cost advantage, and is sent to the driver in a timely manner for execution, reducing response delays. The design of this system not only improves the ability to respond to emergencies, but also can quickly adjust the path according to real-time traffic changes, thereby minimizing transportation risks and costs and improving customer satisfaction and service quality.

[0029] In summary, by designing a neural network architecture containing multi-layer LSTM units, and using the sliding window method to extract multiple subsequences as training samples, and combining the loss function of the mean square error and the penalty term to measure the difference between the predicted value and the true value, a more accurate traffic forecast is provided, which is helpful to plan the optimal route in advance, reduce transportation time and cost, and improve logistics efficiency. The abnormal conditions that affect driving safety or efficiency are identified based on the anomaly detection model. Once an abnormal warning is received or a significant change in traffic conditions is detected, the dynamic path adjustment algorithm is immediately started, and the path planning is dynamically adjusted, which enhances the robustness and adaptability of the system, not only improves the ability to respond to emergencies, but also can quickly adjust the path according to real-time traffic changes.

[0030] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for optimal path planning for logistics transportation, characterized in that: include: Obtain real-time traffic flow data, weather forecast information, and external factors of accident reports from multiple channels through IoT devices, mobile applications, and API interfaces to form raw data sets; Preprocess the original data set and store it in a distributed database to obtain preprocessed traffic data Based on the original data set and the preprocessed traffic data, a traffic prediction model based on a machine learning algorithm is constructed. The preprocessed traffic data and the original data set are used as training sets and input into the traffic prediction model. The model is trained to predict the traffic conditions in the future and obtain traffic prediction results. Combined with the traffic prediction results, a graph theory algorithm is used to calculate multiple candidate paths from the starting point to the end point to obtain a preliminary optimized path; Set and quantify the optimization goal, select the non-dominated solution as the final recommended solution based on the preliminary optimization path through Pareto frontier analysis, and obtain the optimal path solution; Establish a real-time monitoring system to monitor vehicle operating status, identify abnormal situations and trigger early warning notifications, dynamically adjust route planning according to actual traffic conditions, and update the optimal route plan.

2. The method for optimal logistics transportation path planning according to claim 1, characterized in that: The specific steps of obtaining real-time traffic flow data, weather forecast information, and external factors of accident reports from multiple channels through IoT devices, mobile applications, and API interfaces to form an original data set are as follows: Construct an HTTP GET request for the Amap API, including a timestamp, geographic region code, and a list of required parameters; The received data packet is in JSON format and contains a lot of redundant information. A preliminary screening method is used to retain only the parts related to the current task, namely traffic flow, weather conditions and accident location. The expression is: ; in, Indicates the time interval for the region The accumulated amount of raw data collected, It's time point Time Zone The traffic flow function in The area in the same time period The weather condition function, is a function of the location where the accident occurred; Perform integrity checks on the data after initial screening to ensure that each data point meets the predefined standards; The data from different time points are integrated into a continuous time series and grouped according to geographical location to form the original data set.

3. The method for optimal logistics transportation path planning according to claim 1, characterized in that: The original data set is preprocessed and stored in a distributed database to obtain preprocessed traffic data. The specific steps are as follows: Take the original data set through Spark RDD, apply cleaning rules to remove outliers, duplicate records, and fill in missing data; For data in different formats, a standardized conversion function is used to unify the data into a consistent numerical range; Extract key features based on cleaned data, and use historical data analysis to mine potential patterns and add new feature dimensions; Get the preprocessed traffic data.

4. The method for optimal logistics transportation path planning according to claim 1, characterized in that: The traffic prediction model based on the machine learning algorithm is constructed based on the original data set and the preprocessed traffic data. The specific steps are as follows: Combine the preprocessed traffic data and the original dataset into a complete training set; The sliding window method is used to extract multiple subsequences as training samples, and each subsequence contains data of several time steps; Design a neural network architecture containing multiple layers of LSTM units. Each LSTM unit is responsible for processing the data of one time step and passing the context information through the hidden state. Add a fully connected layer to finally output the prediction result. The expression is: ; in, Indicates at time The hidden state of is the input feature vector, is the long short-term memory network function; The loss function that combines the mean square error and the penalty term is used to measure the difference between the predicted value and the true value. The expression is: ; in, is the total loss function, Based on the hidden state The predicted traffic conditions after the fully connected layer and activation function, is the actual observed traffic condition, is the sample size, is the penalty coefficient, is the feature dimension, is the gradient of the loss function with respect to the predicted value; Using the defined loss function The LSTM model is trained with the Adam optimizer. During the training process, the model parameters are updated through the back-propagation algorithm to gradually reduce the loss function. Use the trained LSTM model to predict the traffic conditions in the future. The prediction results for the next few time steps are expressed as: ； in, Indicates future time The predicted traffic conditions in the area, is the time after LSTM processing The hidden state of is the weight matrix of the fully connected layer, is the bias term, is an activation function.

5. The method for optimal logistics transportation path planning according to claim 1, characterized in that: The above-mentioned traffic prediction results are combined to use a graph theory algorithm to calculate multiple candidate paths from the starting point to the end point to obtain a preliminary optimized path. The specific steps are as follows: Based on the preprocessed traffic data and traffic prediction results, a weighted directed graph is constructed. Represent the traffic network, initialize the A* search algorithm parameters, design a heuristic function that combines geographic distance and predicted traffic conditions, and for each node, calculate the actual cost from the starting node to the current node. The expression is as follows: ； ； ； in, It is the edge The weight of is the length of the road section, is the predicted traffic condition, and is the adjustment coefficient, is a function of weather conditions, Is a node The heuristic estimate of From the starting node To Node The straight-line distance, is the adjustment coefficient, Is a node Forecasted traffic conditions in your area, From the starting node To Node The actual cost, From the starting node To Node A collection of paths, Is the path The weight of each edge in ; Using weighted graphs and the heuristic function , execute the A* search algorithm to find the starting node To the target node The multiple candidate paths of are expressed as: ； in, Is a node the estimated total cost of According to the results of the A* search algorithm, multiple candidate paths from the starting point to the end point are generated, and the cost of each path is calculated. One or more paths with the lowest cost are selected as the preliminary optimization path. The expression is: ； in, is the initial optimization path, is the set of all candidate paths from the starting point to the end point, Is the path The weight of each edge in .

6. The method for optimal logistics transportation path planning according to claim 1, characterized in that: The optimization goal is set and quantified, and based on the preliminary optimization path, the non-dominated solution is selected as the final recommended solution through Pareto frontier analysis to obtain the optimal path solution. The specific steps are as follows: Clarify the optimization goals in the logistics and transportation process, including time efficiency, economic cost and environmental friendliness; Construct a comprehensive evaluation function to integrate the three optimization objectives, and introduce interaction rules to achieve a reasonable balance between different objectives; All preliminary optimization paths are scored using a comprehensive evaluation function, and a set of non-dominated solutions are selected from the paths using the Pareto frontier analysis method. After determining the non-dominated solution, the optimal path solution is selected according to time efficiency > economic cost > environmental friendliness.

7. The method for optimal logistics transportation path planning according to claim 1, characterized in that: The real-time monitoring system is established to monitor the vehicle operation status, identify abnormal situations and trigger early warning notifications, dynamically adjust the path planning according to the actual traffic conditions, and update the optimal path plan. The specific steps are as follows: Based on the original data set, anomaly detection models are used to identify abnormal situations that affect driving safety or efficiency; When receiving an abnormal warning or detecting a significant change in traffic conditions, the system immediately activates the dynamic path adjustment algorithm and dynamically adjusts the path planning; Based on the results of the dynamic path adjustment algorithm, the system selects one or more new paths with the lowest cost as the final recommended solutions and sends them to the driver for execution in a timely manner. When the new best path is significantly better than the current path, the optimal path solution is updated immediately, otherwise the original path remains unchanged.

8. The method for optimal logistics transportation path planning according to claim 7, characterized in that: When an abnormal warning is received or a significant change in traffic conditions is detected, the system immediately starts the dynamic path adjustment algorithm and dynamically adjusts the path planning expression as follows: ； in, It is a new path The cost, is a set of candidate paths from the current location to the destination, is the weight of each edge in the original path, is the traffic prediction influencing factor, It is the change in the traffic conditions of the corresponding road section in the future time period.

9. The method for optimal logistics transportation path planning according to claim 6, characterized in that: The comprehensive evaluation function is constructed to integrate the three optimization objectives and to achieve a reasonable balance between different objectives by introducing interaction rules. The expression is: ; in, is a candidate path variable, is the time distribution function along the path, represents the path length, is the time sensitivity coefficient, and are the economic cost weight and its elasticity index, It is the environmental influencing factor. is the cost per trip, is the carbon dioxide emissions per kilometer, is the number of road sections, m is the total mileage, is the average economic cost, is the average carbon dioxide emissions per kilometer, is a comprehensive evaluation function.

10. The method for optimal logistics transportation path planning according to claim 3, characterized in that: Based on the cleaned data, key features are extracted, and historical data analysis is used to mine potential patterns and add new feature dimensions. The expression is: ; in, represents the final extracted feature vector, is a function of average vehicle speed, is the congestion index function, is an additional characteristic function, and is the adjustment weight parameter.

Images