Vehicle scheduling management system based on big data analysis

Abstract

The invention discloses a vehicle scheduling management system based on big data analysis, and the system comprises a data collection layer which is used for collecting vehicle position data, vehicle state data, traffic road condition data, driving track data, and business demand data; the data processing layer is used for carrying out cleaning, conversion and distributed storage on the collected data; the data analysis layer is used for constructing a road condition prediction model, a vehicle health assessment model and a scheduling decision model based on the processed data; the application layer is used for vehicle monitoring, intelligent scheduling, report analysis and early warning notification; the data circulation engine is used for establishing data association channels among the functional modules; the space-time road condition prediction module is used for fusing the traffic road condition data and the driving track data according to the space-time sequence model; according to the method, the scheduling strategy is constructed based on the Markov decision process and the deep Q network, the road condition and the vehicle fault can be accurately predicted, and a scientific basis is provided for the scheduling decision.

Description

Technical Field

[0001] This invention relates to the field of vehicle dispatch management, specifically a vehicle dispatch management system based on big data analysis. Background Technology

[0002] In the field of vehicle dispatch management, with the rapid development of industries such as logistics and ride-hailing, traditional vehicle dispatch management systems have gradually exposed their limitations. Traditional systems rely heavily on manual experience or simple rule algorithms for dispatching, making it difficult to process massive and complex data such as vehicle location, status, and traffic conditions in real time and accurately. For example, when faced with sudden traffic congestion or vehicle malfunctions, they cannot adjust dispatch strategies in a timely manner, resulting in unreasonable allocation of vehicle resources, low transportation efficiency, and increased operating costs. At the same time, traditional systems lack effective means of monitoring and predicting vehicle health status, making it difficult to prevent malfunctions in advance and affecting the stability and security of services. In addition, the data interaction and collaboration capabilities between various functional modules are weak, and the value of data cannot be fully explored and utilized. With the development of technologies such as big data and artificial intelligence, how to utilize these technologies to build an intelligent and efficient vehicle dispatch and management system, and achieve optimized allocation of vehicle resources, accurate prediction of road conditions, intelligent early warning of vehicle health, and coordinated operation of various functional modules, has become an urgent problem to be solved. Summary of the Invention

[0003] The purpose of this invention is to solve the problems mentioned above, and therefore proposes a vehicle dispatching and management system based on big data analysis.

[0004] The objective of this invention can be achieved through the following technical solution: a vehicle dispatching and management system based on big data analysis, comprising: The data acquisition layer is used to collect vehicle location data, vehicle status data, traffic condition data, driving trajectory data, and business requirement data. The data processing layer is used to clean, transform, and distribute the collected data. The data analysis layer is used to build traffic prediction models, vehicle health assessment models, and dispatch decision models based on the processed data. The application layer is used for vehicle monitoring, intelligent dispatching, report analysis, and early warning notifications; The data flow engine is used to establish data connection channels between various functional modules, enabling targeted data push and feedback optimization. The spatiotemporal traffic prediction module is used to fuse traffic data and driving trajectory data based on a spatiotemporal sequence model to generate multi-dimensional traffic prediction results. The vehicle health warning module is used to build a vehicle fault prediction model through multi-sensor data fusion and machine learning algorithms; The learning scheduling module is used to build adaptive scheduling strategies based on Markov decision processes to dynamically optimize vehicle resource allocation.

[0005] By collecting data from multiple dimensions, comprehensive information related to vehicle operation is obtained, providing rich and accurate basic data for subsequent data processing and analysis. This ensures the scientific nature and reliability of system decisions. By constructing various models, accurate prediction of road conditions, advance assessment of vehicle health status, and scientific and reasonable scheduling decisions are achieved, providing strong technical support for vehicle scheduling management, improving scheduling efficiency and management level. Combining spatiotemporal characteristics to predict road conditions is more accurate than traditional prediction methods, providing more reliable road condition information for vehicle scheduling, helping to rationally plan routes, reduce traffic congestion, and improve transportation efficiency. Utilizing data fusion and machine learning technologies, advance prediction of vehicle malfunctions is achieved, transforming vehicle maintenance from passive repair to proactive prevention, reducing vehicle malfunction rates, maintenance costs, and vehicle downtime.

[0006] Furthermore, the data acquisition layer includes: Satellite positioning unit: Acquires vehicle latitude and longitude coordinates and driving trajectory data in real time; Onboard sensor network module: collects parameters such as engine speed, water temperature, oil pressure, and tire pressure; Traffic information interface: Connects to the traffic management system to obtain real-time traffic conditions, accident information, and road construction data; Business Order Module: Retrieves order location, time window, and service priority data.

[0007] Furthermore, the data processing layer includes: Data cleaning unit: Removes noisy data based on sliding window algorithm and anomaly detection model; Data standardization unit: transforming multi-source heterogeneous data into a unified spatiotemporal data structure; Distributed storage unit: Utilizes the Hadoop ecosystem to store petabyte-scale historical data.

[0008] Furthermore, the spatiotemporal traffic condition prediction module includes: Spatiotemporal feature extraction unit: Based on the spatiotemporal attention mechanism, the LSTM network extracts the temporal dependence and spatial correlation features of traffic condition data and driving trajectory data. Multimodal fusion unit: weighted fusion of real-time traffic data and historical data from the same period; Traffic prediction unit: Generates a probability distribution of road traffic conditions for the next 30 minutes; Dynamic correction unit: Updates the parameters of the prediction model online based on real-time feedback data.

[0009] It can deeply explore the inherent characteristics of road condition data in time and space, making road condition predictions more in line with the actual situation, improving prediction accuracy and reliability. By comprehensively considering real-time and historical data and making full use of data information, the prediction results are more comprehensive and accurate, providing more effective road condition references for vehicle dispatching. It can intuitively display future road conditions in the form of probability distribution, providing quantitative road condition information for vehicle dispatching, which facilitates the formulation of reasonable dispatching strategies and route planning.

[0010] Furthermore, the vehicle health warning module includes: Multi-sensor fusion unit: The Kalman filter algorithm is used to fuse multi-source data from the vehicle sensor network module; Feature engineering unit: Extracting time-domain features, frequency-domain features, and time-frequency-domain features to construct a health indicator system; Fault prediction unit: Based on the random forest algorithm, a multi-class prediction model is constructed to identify fault types and their probability of occurrence; Early warning classification unit: Generates different levels of early warning information based on the severity and development trend of the fault.

[0011] Furthermore, the learning scheduling module includes: State space construction unit: Maps vehicle location data, business requirement data, road condition prediction and vehicle fault prediction model outputs into state vectors; Action Decision Unit: Generates the optimal scheduling action sequence based on a deep Q-network; Reward design unit: Construct a multi-objective reward function by combining task completion time, driving mileage, energy consumption cost and customer satisfaction; Strategy optimization unit: continuously optimizes the scheduling strategy through experience replay mechanism and policy gradient algorithm.

[0012] Furthermore, the application layer includes: Monitoring interface: A 3D visualization of vehicle location is achieved using WebGL technology; Scheduling Workbench: Provides drag-and-drop task assignment and path planning functions; Data analytics dashboard: Enables multi-dimensional analysis of operational metrics based on BI technology; Early warning center: It realizes graded response to abnormal events through the linkage mechanism of sound and light.

[0013] Furthermore, the data transfer engine includes: Data routing unit: Enables targeted data push between modules based on the publish-subscribe model; Data quality monitoring unit: monitors data integrity, accuracy, and timeliness in real time; Feedback closed-loop unit: Feeds the scheduling execution results as training data back to the data analysis layer; Data security gateway: Implements secure authentication for cross-module data transmission based on a zero-trust architecture.

[0014] Furthermore, the data flow engine is configured to: push the road condition prediction results to the learning and scheduling module as constraints for path planning, and synchronize the vehicle health status data output by the vehicle health warning module to the scheduling decision model to achieve intelligent avoidance of faulty vehicles.

[0015] This enables effective data collaboration across different modules, allowing scheduling decisions to fully consider road conditions and vehicle health, optimize route planning, avoid malfunctioning vehicles, improve the scientific rigor and rationality of scheduling, and reduce operational risks.

[0016] Furthermore, when constructing the traffic prediction model, vehicle health assessment model, and scheduling decision model, the data analysis layer employs a transfer learning algorithm to transfer model parameters trained in similar regions or time periods to the current model training, thereby accelerating model convergence. Combined with a reinforcement learning algorithm, the model parameters are dynamically optimized based on feedback from actual scheduling effects, achieving adaptive adjustment of the model.

[0017] Compared with the prior art, the beneficial effects of the present invention are: The data analysis layer constructs models for road condition prediction, vehicle health assessment, and scheduling decision-making; the spatiotemporal road condition prediction module adopts a spatiotemporal attention mechanism LSTM network and multimodal fusion technology; the vehicle health warning module combines multi-sensor fusion and random forest algorithm; the learning scheduling module constructs scheduling strategies based on Markov decision process and deep Q network, accurately predicts road conditions and vehicle faults, and provides a scientific basis for scheduling decisions; it dynamically optimizes vehicle resource allocation, improves scheduling efficiency, and reduces operating costs.

[0018] The data analysis layer employs transfer learning and learning algorithms to accelerate model convergence and achieve adaptive adjustment; the data flow engine links road conditions, vehicle health data, and scheduling decisions, shortening model training time, improving model prediction accuracy and adaptability, enabling intelligent avoidance of faulty vehicles, and enhancing scheduling safety and rationality. Attached Figure Description

[0019] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0020] Figure 1 This is a diagram showing the composition of the vehicle dispatching and management system based on big data analysis according to the present invention. Detailed Implementation

[0021] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figure 1 As shown, the vehicle dispatch management system based on big data analytics includes: The data acquisition layer is used to collect vehicle location data, vehicle status data, traffic condition data, driving trajectory data, and business requirement data. The data processing layer is used to clean, transform, and distribute the collected data. The data analysis layer is used to build traffic prediction models, vehicle health assessment models, and dispatch decision models based on the processed data. The application layer is used for vehicle monitoring, intelligent dispatching, report analysis, and early warning notifications; The data flow engine is used to establish data connection channels between various functional modules, enabling targeted data push and feedback optimization. The spatiotemporal traffic prediction module is used to fuse traffic data and driving trajectory data based on a spatiotemporal sequence model to generate multi-dimensional traffic prediction results. The vehicle health warning module is used to build a vehicle fault prediction model through multi-sensor data fusion and machine learning algorithms; The learning scheduling module is used to build adaptive scheduling strategies based on Markov decision processes to dynamically optimize vehicle resource allocation.

[0023] The data acquisition layer includes: Satellite positioning unit: Acquires vehicle latitude and longitude coordinates and driving trajectory data in real time; Onboard sensor network module: collects parameters such as engine speed, water temperature, oil pressure, and tire pressure; Traffic information interface: Connects to the traffic management system to obtain real-time traffic conditions, accident information, and road construction data; Business Order Module: Retrieves order location, time window, and service priority data.

[0024] The data processing layer includes: Data cleaning unit: Removes noisy data based on sliding window algorithm and anomaly detection model; Data standardization unit: transforming multi-source heterogeneous data into a unified spatiotemporal data structure; Distributed storage unit: Utilizes the Hadoop ecosystem to store petabyte-scale historical data.

[0025] The spatiotemporal traffic condition prediction module includes: Spatiotemporal feature extraction unit: Based on the spatiotemporal attention mechanism, the LSTM network extracts the temporal dependence and spatial correlation features of traffic condition data and driving trajectory data. Multimodal fusion unit: weighted fusion of real-time traffic data and historical data from the same period; Traffic prediction unit: Generates a probability distribution of road traffic conditions for the next 30 minutes; Dynamic correction unit: Updates the parameters of the prediction model online based on real-time feedback data.

[0026] The vehicle health warning module includes: Multi-sensor fusion unit: The Kalman filter algorithm is used to fuse multi-source data from the vehicle sensor network module; Feature engineering unit: Extracting time-domain features, frequency-domain features, and time-frequency-domain features to construct a health indicator system; Fault prediction unit: Based on the random forest algorithm, a multi-class prediction model is constructed to identify fault types and their probability of occurrence; Early warning classification unit: Generates different levels of early warning information based on the severity and development trend of the fault.

[0027] The learning scheduling module includes: State space construction unit: Maps vehicle location data, business requirement data, road condition prediction and vehicle fault prediction model outputs into state vectors; Action Decision Unit: Generates the optimal scheduling action sequence based on a deep Q-network; Reward design unit: Construct a multi-objective reward function by combining task completion time, driving mileage, energy consumption cost and customer satisfaction; Strategy optimization unit: continuously optimizes the scheduling strategy through experience replay mechanism and policy gradient algorithm.

[0028] The application layer includes: Monitoring interface: A 3D visualization of vehicle location is achieved using WebGL technology; Scheduling Workbench: Provides drag-and-drop task assignment and path planning functions; Data analytics dashboard: Enables multi-dimensional analysis of operational metrics based on BI technology; Early warning center: It realizes graded response to abnormal events through the linkage mechanism of sound and light.

[0029] The data transfer engine includes: Data routing unit: Enables targeted data push between modules based on the publish-subscribe model; Data quality monitoring unit: monitors data integrity, accuracy, and timeliness in real time; Feedback closed-loop unit: Feeds the scheduling execution results as training data back to the data analysis layer; Data security gateway: Implements secure authentication for cross-module data transmission based on a zero-trust architecture.

[0030] The data flow engine is configured to: push the road condition prediction results to the learning and scheduling module as constraints for path planning; and synchronize the vehicle health status data output by the vehicle health warning module to the scheduling decision model to achieve intelligent avoidance of faulty vehicles.

[0031] When constructing the traffic prediction model, vehicle health assessment model, and scheduling decision model, the data analysis layer employs a transfer learning algorithm to transfer model parameters trained in similar regions or time periods to the current model training, thereby accelerating model convergence. Combined with a reinforcement learning algorithm, the model parameters are dynamically optimized based on feedback from actual scheduling effects, achieving adaptive adjustment of the model.

[0032] Specifically, in practical applications, the data acquisition layer serves as the foundation of the system's operation, collecting data through diverse devices and interfaces. The satellite positioning unit utilizes the Global Navigation Satellite System to acquire real-time vehicle latitude and longitude coordinates and driving trajectory data, providing a basis for vehicle location monitoring and route planning. The vehicle-mounted sensor network module collects status parameters such as engine speed, water temperature, oil pressure, and tire pressure through sensors installed in key parts of the vehicle, enabling real-time monitoring of vehicle operation. The traffic information interface connects to the traffic management system to obtain real-time road conditions, accident information, and road construction data, aiding in understanding the external traffic environment. The business order module retrieves order location, time window, and service priority data from the business system to clarify scheduling requirements. This data encompasses vehicle location data, vehicle status data, traffic condition data, driving trajectory data, and business requirement data, providing rich information for subsequent system processing.

[0033] The data processing layer performs in-depth processing on the collected data to ensure its usability. The data cleaning unit removes noise and outliers from the data based on the sliding window algorithm and anomaly detection model; the data standardization unit transforms multi-source heterogeneous data into a unified spatiotemporal data structure to facilitate subsequent analysis; and the distributed storage unit adopts the Hadoop ecosystem to achieve petabyte-level historical data storage, providing data support for data analysis and model training.

[0034] The data analysis layer constructs key models based on the processed data. In the spatiotemporal traffic prediction module, the spatiotemporal feature extraction unit uses an LSTM network with a spatiotemporal attention mechanism to extract the temporal dependence and spatial correlation features of traffic data; the multimodal fusion unit weightedly fuses real-time traffic data with historical data from the same period; the traffic prediction unit generates a probability distribution of road traffic status for the next 30 minutes; and the dynamic correction unit updates the prediction model's parameters online based on real-time feedback data to construct the traffic prediction model. The vehicle health warning module uses a multi-sensor fusion unit to fuse multi-source data using a Kalman filter algorithm; the feature engineering unit extracts time-domain, frequency-domain, and time-frequency-domain features to construct a health indicator system; the fault prediction unit constructs a multi-classification prediction model based on a random forest algorithm; and the warning grading unit generates different levels of warning information based on fault severity and development trends to construct a vehicle health assessment model. The learning scheduling module maps relevant data into state vectors through a state space construction unit; the action decision unit generates the optimal scheduling action sequence based on a deep Q-network; the reward design unit constructs a reward function combining multiple objectives; and the policy optimization unit optimizes the scheduling policy through an experience playback mechanism and a policy gradient algorithm to construct a scheduling decision model.

[0035] The application layer provides users with intuitive and practical functions. The monitoring interface uses WebGL technology to achieve a 3D visualization of vehicle locations, facilitating real-time monitoring of vehicle dynamics; the dispatching workbench provides drag-and-drop task allocation and route planning functions to improve dispatching efficiency; the data analysis dashboard uses BI technology to achieve multi-dimensional analysis of operational indicators, providing data support for decision-making; and the early warning center uses an audio-visual linkage mechanism to achieve graded response to abnormal events and promptly handle emergencies.

[0036] The data flow engine acts as a data bridge in the system. The data routing unit uses a publish-subscribe model to push data between modules, such as pushing traffic prediction results to the learning and scheduling module as constraints for path planning, and synchronizing vehicle health status data output by the vehicle health warning module to the scheduling decision model to achieve intelligent avoidance of faulty vehicles. The data quality monitoring unit monitors data integrity, accuracy, and timeliness in real time. The feedback loop unit feeds back scheduling execution results as training data to the data analysis layer. The data security gateway uses a zero-trust architecture to achieve secure authentication of cross-module data transmission, ensuring efficient and secure data flow and interaction between functional modules.

[0037] The spatiotemporal traffic prediction module fuses traffic data and driving trajectory data using a spatiotemporal sequence model to generate multi-dimensional traffic prediction results, helping the dispatch system plan better routes in advance. The vehicle health warning module constructs a vehicle fault prediction model through multi-sensor data fusion and machine learning algorithms, promptly identifying and issuing warnings of potential vehicle problems to ensure driving safety. The learning-based dispatch module constructs an adaptive dispatch strategy based on Markov decision processes, dynamically optimizing vehicle resource allocation and improving vehicle utilization efficiency and service quality.

[0038] Meanwhile, when constructing the traffic prediction model, vehicle health assessment model, and scheduling decision model, the data analysis layer adopts the transfer learning algorithm, which transfers the model parameters trained in similar areas or similar time periods to the current model training to accelerate model convergence. Combined with the reinforcement learning algorithm, the model parameters are dynamically optimized based on the feedback of actual scheduling effect, so as to realize the adaptive adjustment of the model and continuously improve the system performance. Through close collaboration among the above components, the vehicle dispatch management system based on big data analytics enables vehicle monitoring, intelligent dispatching, report analysis, and early warning notification functions, providing an efficient and intelligent solution for vehicle dispatch management.

[0039] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A vehicle dispatching and management system based on big data analytics, characterized in that, include: The data acquisition layer is used to collect vehicle location data, vehicle status data, traffic condition data, driving trajectory data, and business requirement data. The data processing layer is used to clean, transform, and distribute the collected data. The data analysis layer is used to build traffic prediction models, vehicle health assessment models, and dispatch decision models based on the processed data. The application layer is used for vehicle monitoring, intelligent dispatching, report analysis, and early warning notifications; The data flow engine is used to establish data connection channels between various functional modules, enabling targeted data push and feedback optimization. The spatiotemporal traffic prediction module is used to fuse traffic data and driving trajectory data based on a spatiotemporal sequence model to generate multi-dimensional traffic prediction results. The vehicle health warning module is used to build a vehicle fault prediction model through multi-sensor data fusion and machine learning algorithms; The learning scheduling module is used to build adaptive scheduling strategies based on Markov decision processes to dynamically optimize vehicle resource allocation.

2. The vehicle dispatching and management system based on big data analysis according to claim 1, characterized in that, The data acquisition layer includes: Satellite positioning unit: Acquires vehicle latitude and longitude coordinates and driving trajectory data in real time; Onboard sensor network module: collects parameters such as engine speed, water temperature, oil pressure, and tire pressure; Traffic information interface: Connects to the traffic management system to obtain real-time traffic conditions, accident information, and road construction data; Business Order Module: Retrieves order location, time window, and service priority data.

3. The vehicle dispatching and management system based on big data analysis according to claim 1, characterized in that, The data processing layer includes: Data cleaning unit: Removes noisy data based on sliding window algorithm and anomaly detection model; Data standardization unit: transforming multi-source heterogeneous data into a unified spatiotemporal data structure; Distributed storage unit: Utilizes the Hadoop ecosystem to store petabyte-scale historical data.

4. The vehicle dispatching and management system based on big data analysis according to claim 1, characterized in that, The spatiotemporal traffic condition prediction module includes: Spatiotemporal feature extraction unit: Based on the spatiotemporal attention mechanism, the LSTM network extracts the temporal dependence and spatial correlation features of traffic condition data and driving trajectory data. Multimodal fusion unit: weighted fusion of real-time traffic data and historical data from the same period; Traffic prediction unit: Generates a probability distribution of road traffic conditions for the next 30 minutes; Dynamic correction unit: Updates the parameters of the prediction model online based on real-time feedback data.

5. The vehicle dispatching and management system based on big data analysis according to claim 1, characterized in that, The vehicle health warning module includes: Multi-sensor fusion unit: The Kalman filter algorithm is used to fuse multi-source data from the vehicle sensor network module; Feature engineering unit: Extracting time-domain features, frequency-domain features, and time-frequency-domain features to construct a health indicator system; Fault prediction unit: Based on the random forest algorithm, a multi-class prediction model is constructed to identify fault types and their probability of occurrence; Early warning classification unit: Generates different levels of early warning information based on the severity and development trend of the fault.

6. The vehicle dispatching and management system based on big data analysis according to claim 1, characterized in that, The learning scheduling module includes: State space construction unit: Maps vehicle location data, business requirement data, road condition prediction and vehicle fault prediction model outputs into state vectors; Action Decision Unit: Generates the optimal scheduling action sequence based on a deep Q-network; Reward design unit: Construct a multi-objective reward function by combining task completion time, driving mileage, energy consumption cost and customer satisfaction; Strategy optimization unit: continuously optimizes the scheduling strategy through experience replay mechanism and policy gradient algorithm.

7. The vehicle dispatching and management system based on big data analysis according to claim 1, characterized in that, The application layer includes: Monitoring interface: A 3D visualization of vehicle location is achieved using WebGL technology; Scheduling Workbench: Provides drag-and-drop task assignment and path planning functions; Data analytics dashboard: Enables multi-dimensional analysis of operational metrics based on BI technology; Early warning center: It realizes graded response to abnormal events through the linkage mechanism of sound and light.

8. The vehicle dispatching and management system based on big data analysis according to claim 1, characterized in that, The data transfer engine includes: Data routing unit: Enables targeted data push between modules based on the publish-subscribe model; Data quality monitoring unit: monitors data integrity, accuracy, and timeliness in real time; Feedback closed-loop unit: Feeds the scheduling execution results as training data back to the data analysis layer; Data security gateway: Implements secure authentication for cross-module data transmission based on a zero-trust architecture.

9. The vehicle dispatching and management system based on big data analysis according to claim 1, characterized in that, The data flow engine is configured to: push the road condition prediction results to the learning and scheduling module as constraints for path planning; and synchronize the vehicle health status data output by the vehicle health warning module to the scheduling decision model to achieve intelligent avoidance of faulty vehicles.

10. The vehicle dispatching and management system based on big data analysis according to claim 1, characterized in that, When constructing the traffic prediction model, vehicle health assessment model, and scheduling decision model, the data analysis layer employs a transfer learning algorithm to transfer model parameters trained in similar regions or time periods to the current model training, thereby accelerating model convergence. Combined with a reinforcement learning algorithm, the model parameters are dynamically optimized based on feedback from actual scheduling effects, achieving adaptive adjustment of the model.

Images