Intelligent analysis method for cross-system data collaboration based on business logic

Abstract

The present application relates to the technical field of data processing and analysis, in particular to a cross-system data collaborative intelligent analysis method based on business logic. The method first deploys edge computing nodes at key traffic locations for data collection and preprocessing, then networks and communicates between nodes, then each node extracts data features and performs local analysis, starts and assigns collaborative analysis tasks according to predefined trigger conditions including quantitative and qualitative conditions, nodes share and fuse data according to tasks, use traffic flow and machine learning models to collaboratively analyze and generate decision suggestions based on the fused data set, and feed back to the traffic management system, and finally optimize and adjust according to the feedback information. The corresponding system includes multiple edge computing nodes with corresponding function modules working collaboratively. The present application effectively overcomes the problems of insufficient collaboration and incomplete analysis in existing traffic data processing, can quickly respond to traffic condition changes, improve traffic data analysis quality and the scientific nature of management decisions, and ensure efficient operation of the traffic system.

Description

Technical Field

[0001] This invention relates to the field of data processing and analysis technology, specifically to a cross-system data collaborative intelligent analysis method based on business logic. Background Technology

[0002] With the acceleration of urbanization and the booming development of the transportation industry, transportation systems are becoming increasingly complex, generating massive amounts of real-time traffic data. Traditional traffic data processing methods often face numerous challenges, such as high data transmission latency, high network bandwidth pressure, and insufficient timeliness in responding to sudden traffic situations when relying on centralized cloud computing centers for data processing.

[0003] The emergence of edge computing offers a new approach to solving these problems. By processing data at edge nodes close to the data source, data transmission latency can be effectively reduced and response speed improved. However, in traffic scenarios, the data that a single edge computing node can process and analyze is limited, making it difficult to comprehensively and accurately grasp the real-time situation of the entire traffic system. Furthermore, if the edge nodes lack an effective collaboration mechanism and cannot fully integrate their respective data resources and analytical capabilities, it is difficult to achieve efficient traffic data analysis and reasonable traffic management decisions. Summary of the Invention

[0004] This invention provides a cross-system data collaborative intelligent analysis method and system based on business logic. By collecting, processing, and collaboratively analyzing real-time traffic data at key traffic locations through multiple edge computing nodes, it overcomes the problems of high data processing latency, insufficient collaboration among nodes, and incomplete traffic data analysis in existing technologies. This enables real-time and accurate understanding of traffic conditions, thereby generating effective traffic management decision-making suggestions and improving traffic operation efficiency and management level.

[0005] Intelligent analysis methods for cross-system data collaboration based on business logic include:

[0006] Data Acquisition and Preprocessing: Edge computing nodes are deployed at multiple key traffic locations. Each edge computing node is equipped with at least one data acquisition device to collect real-time traffic data within its area. The collected data is then preprocessed by removing noise, standardizing the format, and correcting outliers.

[0007] Inter-node communication and networking: Each edge computing node constructs a local network through wired or wireless communication. Communication protocols and data interaction rules are determined through identity authentication and network configuration to achieve communication connections between them.

[0008] Data feature extraction and local analysis: Each edge computing node uses deep learning algorithms or traditional statistical analysis methods to extract features from the preprocessed local real-time traffic data, transforming the raw traffic data into feature vectors and conducting local analysis.

[0009] Collaborative analysis task triggering and allocation: Collaborative analysis tasks are initiated based on predefined triggering conditions, and tasks are allocated based on relevant factors of each edge computing node;

[0010] Data collaboration, sharing and fusion: Each edge computing node shares its extracted traffic data feature vectors to the network according to its assigned tasks. The remaining nodes then use weighted fusion and feature splicing methods to fuse the shared data to form a comprehensive data set that reflects the traffic situation.

[0011] Collaborative Analysis and Decision Making: Based on the fused dataset, each edge computing node uses traffic flow models and machine learning prediction models to participate in collaborative analysis, generate traffic management decision suggestions, and feed them back to the traffic management system for execution.

[0012] Feedback and Optimization: Based on the traffic condition feedback information after the traffic management system implements decision-making suggestions, each edge computing node optimizes and adjusts the relevant elements involved in the collaborative analysis.

[0013] Preferably, in the data feature extraction and local analysis step, the deep learning algorithm is a convolutional neural network. It is trained using a large amount of labeled or unlabeled traffic data by constructing a neural network structure containing multiple convolutional layers, pooling layers and fully connected layers to learn and extract feature patterns from the data.

[0014] Preferably, in the data feature extraction and local analysis step, the local analysis specifically includes:

[0015] Traffic parameter calculations: Calculate local traffic density, measured by the number of vehicles passing through a unit road segment per unit time; calculate local traffic flow rate, determined based on the frequency of vehicles passing through a specific monitoring point within a specific time period; calculate local average speed, derived from the ratio of vehicle distance traveled to the time taken.

[0016] Local event detection: Identifies local events through preset rules. When a vehicle's position does not change within a preset time, it is determined to be a stopped vehicle; when multiple vehicles are traveling at speeds consistently below normal and the distance between them is gradually decreasing, it is determined to be a queue formation; when a vehicle's actual speed exceeds the speed limit of the road segment, it is determined to be speeding.

[0017] Confidence level assessment: Taking into account factors such as camera field of view obstruction, radar signal interference, and induction coil stability during data acquisition, the corresponding confidence levels are assigned to each local analysis result.

[0018] Preferably, the predefined triggering condition includes a quantitative trigger:

[0019] Local analysis confidence: When the confidence of an edge node in its local assessment is lower than a preset threshold, it requests input from neighboring nodes to trigger a collaborative analysis task.

[0020] Resource utilization: If a node's processing load or available bandwidth exceeds predefined limits, the node will offload some analysis tasks or request assistance to trigger collaborative analysis tasks.

[0021] Queue length detection: When the queue length is detected to exceed the critical threshold set based on road type and historical traffic flow data, collaboration with upstream and downstream nodes is triggered to adjust signal timing or replan traffic routes, thereby initiating a collaborative analysis task.

[0022] Traffic flow parameters: When local traffic speed, density, or flow rate shows significant abnormal changes compared to historical or expected patterns, a collaborative analysis task is initiated.

[0023] Preferably, the predefined triggering conditions further include event-based triggers:

[0024] Accident detection: When an edge computing node determines that a major accident has occurred that affects traffic in a wider area, it triggers a collaborative analysis task;

[0025] Requests from neighboring nodes: When receiving information requests from neighboring edge nodes, initiate a collaborative analysis task and participate in collaborative work according to the request content;

[0026] Control system requirements: When the traffic signal control system determines that multi-node data is needed to optimize signal timing, a collaborative analysis task is triggered, in which each node collaboratively provides data and participates in the analysis.

[0027] Preferably, the task allocation based on relevant factors of each edge computing node includes: constructing a collaborative analysis task allocation mechanism based on the computing power of each edge computing node, the coverage of the collected data, and the importance of traffic, reasonably dividing the overall traffic data analysis task, and clarifying the specific responsibilities and tasks of each node in the collaborative analysis.

[0028] Preferably, the machine learning prediction model includes a support vector machine and a long short-term memory network, which are trained using historical traffic data and real-time fused data to predict the occurrence and spread of traffic congestion and analyze the impact of traffic accidents on surrounding traffic.

[0029] Preferably, the collaborative analysis tasks include traffic flow prediction, intersection traffic congestion analysis, road segment traffic bottleneck analysis, and regional traffic situation assessment.

[0030] Preferably, in the feedback and optimization step, the optimization adjustment specifically includes adjusting the weights of data fusion, reallocating collaborative analysis tasks, and correcting the parameters of the collaborative analysis model.

[0031] A cross-system data collaborative intelligent analysis system based on business logic includes multiple edge computing nodes deployed in key transportation locations. Each edge computing node contains:

[0032] The data acquisition module is used to collect real-time traffic data within the area using the equipped data acquisition equipment, and transmit the collected data to the preprocessing module;

[0033] The preprocessing module is used to perform preprocessing operations on the received data, such as removing noise, standardizing data format, and correcting outliers, and then transmits the preprocessed data to the feature extraction module.

[0034] The feature extraction module is used to extract features from the preprocessed data using deep learning algorithms or traditional statistical analysis methods, generate feature vectors, perform local analysis operations, and transmit the feature vectors and related local analysis results to the collaborative analysis module.

[0035] The collaborative analysis module includes a task allocation unit, a data sharing and fusion unit, an analysis and decision-making unit, and a feedback optimization unit;

[0036] The task allocation unit is used to construct a collaborative analysis task allocation mechanism based on the computing power of each edge computing node, the coverage of the collected data, and traffic importance factors. It rationally breaks down the overall traffic data analysis task and clarifies the specific responsibilities and tasks of each node in the collaborative analysis.

[0037] The data sharing and fusion unit is used to share the relevant traffic data feature vectors extracted by itself to the network according to the task allocation, and to receive data shared by other nodes. It uses weighted fusion and feature splicing methods to fuse these data to form a comprehensive and more complete data set reflecting the traffic situation, and then transmits the data set to the analysis and decision-making submodule.

[0038] The analysis and decision-making unit is used to analyze traffic conditions based on the fused data set, using traffic flow models and machine learning prediction models, and to generate corresponding traffic management decision-making suggestions based on the analysis results, and then feed the decision-making suggestions back to the corresponding traffic management system for execution.

[0039] The feedback optimization unit is used to receive traffic condition change information after the traffic management system implements decision suggestions, and to optimize and adjust the model parameters, task allocation mechanism, and data fusion method of collaborative analysis.

[0040] Compared with the prior art, the advantages of this invention are:

[0041] By enabling multiple edge computing nodes to work collaboratively, the data resources collected by each node are fully integrated, avoiding the limitations of data from a single node. Combined with local analysis and collaborative analysis mechanisms, the real-time situation of the entire traffic system can be grasped more comprehensively and accurately, effectively improving the quality of traffic data analysis and providing a more reliable basis for traffic management decisions.

[0042] By leveraging the proximity of edge computing nodes to data sources for data processing and collaborative analysis, the latency of data transmission to the cloud is significantly reduced. This enables faster responses to traffic emergencies, timely generation and feedback of effective traffic management decision-making suggestions, which helps improve traffic efficiency, alleviate traffic congestion, and ensure the smooth operation of the transportation system.

[0043] Based on predefined triggering conditions and a scientifically sound task allocation mechanism, the system can dynamically adjust the tasks and methods of collaborative analysis according to actual traffic conditions. At the same time, by combining feedback optimization steps, it continuously adapts to the complex and ever-changing traffic environment, ensuring the long-term effectiveness and stability of the entire system. This enables it to continue to play a role in different traffic scenarios and improve the level of intelligence in traffic management. Attached Figure Description

[0044] Figure 1 This is a flowchart of the cross-system data collaborative intelligent analysis method based on business logic proposed in this invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0046] Example 1:

[0047] Suppose that the business logic-based cross-system data collaborative intelligent analysis system of the present invention is constructed in a bustling commercial district and surrounding road area of ​​a city. This area includes multiple intersections (such as crossroads, T-junctions, etc.), main roads, secondary roads, and branch roads connecting the parking lot of the commercial district, etc., with large traffic flow and complex and changeable traffic conditions.

[0048] Five edge computing nodes, labeled A, B, C, D, and E, were deployed at key traffic locations within this area. Node A is located at a major intersection connecting two main roads and surrounded by several large shopping malls and office buildings, resulting in very high vehicle and pedestrian traffic. Node B is near the middle section of a main road, surrounded by small shops and bus stops, making it a section with concentrated vehicle traffic. Node C is located at the intersection of a secondary road and a main road, where traffic congestion frequently occurs. Node D covers a side road area leading to the commercial area's parking lot, where vehicles entering and exiting the parking lot significantly impact surrounding traffic during weekday commuting hours and weekends and holidays. Node E is located on a ring road within the commercial area, primarily serving the internal circulation of vehicles and traffic management at various entrances to and from the commercial area.

[0049] Each edge computing node is equipped with corresponding data acquisition devices. For example, cameras are installed at intersections and key locations on the road to capture image data such as vehicle trajectories, license plate information, and traffic flow; radar equipment is deployed along the roadside to monitor vehicle speed, distance, and other information; and induction coils are buried under the road surface to count data such as the frequency of vehicle passage.

[0050] Reference Figure 1 The methodology and steps outlined provide specific implementation steps for a cross-system data collaboration and intelligent analysis method based on business logic in this scenario, including:

[0051] Data Acquisition and Preprocessing:

[0052] Node A:

[0053] The camera collects image data according to a preset shooting angle, resolution (e.g., 1080P), and frame rate (25 frames per second). The radar acquires relevant data about the target object at a detection frequency of once every 0.1 seconds and a detection radius of 100 meters. The induction coil collects vehicle passage information based on a sensing sensitivity of 0.5 seconds and a sampling period. All collected data is accompanied by a precise timestamp and corresponding collection location identifier (e.g., specific intersection direction, lane position, etc.).

[0054] For the collected data, firstly, filtering algorithms are used to remove noise data caused by weather (such as raindrop noise in images during heavy rain) and equipment itself (such as occasional pixel failures of cameras). Then, camera image data, radar data, and induction coil data are uniformly converted into the specified data format. Finally, based on the conventional value range and statistical rules of traffic data (such as the reasonable range of vehicle speed from 0 to 120 km / h), data that obviously exceeds the reasonable range is corrected. The preprocessed data is then transmitted to the feature extraction module of node A.

[0055] Nodes B, C, D, and E: These nodes also collect data according to their respective set acquisition device parameters and perform preprocessing operations similar to those of node A, such as noise removal, format unification, and outlier correction, to ensure the quality of the collected data. Then, the preprocessed data is transmitted to their respective feature extraction modules.

[0056] Inter-node communication and networking:

[0057] These five edge computing nodes form a self-organizing network via a 5G network. During network construction, each node performs device authentication using an encrypted authentication method based on the SSL / TLS protocol, ensuring that only legitimately authorized nodes can access the network. Simultaneously, network parameters are configured to define data communication rules based on the TCP / IP protocol, such as specifying data transmission port numbers, packet size limits, and timeout retransmission mechanisms. This ensures secure, stable, and efficient data interaction between nodes, establishing communication connections and laying the foundation for subsequent collaborative data processing.

[0058] Data feature extraction and local analysis:

[0059] Node A:

[0060] Feature extraction: A convolutional neural network (CNN) algorithm is used for feature extraction. A neural network structure is constructed, which includes three convolutional layers (with kernel sizes of 3x3, 5x5, and 3x3 and a stride of 1), two pooling layers (using max pooling with a pooling window of 2x2), and two fully connected layers. The network is trained using a large amount of labeled and unlabeled traffic data of the intersection accumulated over the past few months. This enables the network to learn and extract data feature vectors such as peak and trough traffic flow characteristics in different directions of the intersection, vehicle turning ratio characteristics in each lane, and vehicle type distribution characteristics at different times.

[0061] Local analysis:

[0062] Traffic parameter calculation: Calculate local traffic density by counting the number of vehicles passing through a unit road segment (each lane is 100 meters as a statistical unit road segment) every 5 minutes; traffic flow rate is determined based on the frequency of vehicles passing through a specific monitoring point at the intersection every 10 minutes; local average speed is obtained by calculating the time it takes for vehicles in each lane to travel a fixed length (e.g., 200 meters) of road segment.

[0063] Local event detection: When a vehicle's position does not change significantly in the image for 30 consecutive seconds, it is determined to be a stopped vehicle; if more than 5 vehicles in the same lane are found to have a speed continuously below 20 km / h and the distance between them gradually decreases to within 5 meters, it is determined to be a queue; compare the actual speed of the vehicles with the speed limits of each lane at the intersection, and if the speed limit is exceeded, it is determined to be speeding.

[0064] Confidence Level Assessment: Considering that tall buildings around the intersection may obstruct the camera's field of view, and that the radar signal may be unstable due to electromagnetic interference from electronic devices at nearby bus stops, corresponding confidence levels are assigned to each local analysis result. For example, the confidence level for vehicle feature analysis in an obstructed lane is reduced to 0.7, and the confidence level for vehicle speed data detected by the radar when interference occurs is adjusted to 0.8. This is used to measure the reliability of the local analysis results, and the feature vectors and related local analysis results are then transmitted to the collaborative analysis module of node A.

[0065] Nodes B, C, D, and E: These nodes also employ similar deep learning algorithms or traditional statistical analysis methods (e.g., when analyzing data at the intersection of secondary and main roads, node C uses the mean to calculate the average vehicle speed over different time periods and uses variance to analyze speed fluctuations) to extract features and generate their respective feature vectors. Simultaneously, they conduct local analysis based on the traffic characteristics of their respective areas. For example, node D, located in a parking lot side road area, focuses on the changes in queue lengths of vehicles entering and exiting the parking lot to calculate traffic density, determines traffic flow rate based on the frequency of vehicles entering and exiting the parking lot gate, and derives average speed by calculating the average time it takes for vehicles to travel from the side road entrance to the parking lot entrance. In terms of local event detection, a queue exceeding 5 minutes and reaching a length of 20 meters at the parking lot exit is considered a queue formation event. Furthermore, the confidence level of the local analysis results is evaluated based on the actual working conditions of each node's data acquisition equipment (e.g., if the camera's view is partially obstructed by roadside billboards on a ring road within a commercial area, etc.), and the relevant data is transmitted to their respective collaborative analysis modules.

[0066] Collaborative analysis task triggering and allocation:

[0067] Example of a quantitative trigger:

[0068] Local analysis confidence level: On a weekday afternoon, due to construction in the surrounding area, the intersection where node A is located is dusty, and the camera images become blurry, causing the confidence level of the analysis of some vehicle behavior at the intersection to drop to 0.5 (below the preset threshold of 0.6). At this time, according to the local analysis confidence level trigger condition, node A sends a request input message to the neighboring nodes B and C to start a collaborative analysis task, hoping to use the data they collect to analyze the traffic conditions at the intersection more accurately.

[0069] Resource utilization: During weekends and holidays, as more vehicles head towards the commercial area, the amount of data collected by node B increases significantly, causing its CPU utilization to exceed 90% for 10 consecutive minutes (exceeding the predefined resource utilization limit, which is 85%). Due to the excessive processing load, node B offloads some traffic flow prediction analysis tasks and sends assistance requests to nodes A, D, and E, triggering collaborative analysis tasks. Each node shares the corresponding traffic prediction analysis work according to the task allocation mechanism, jointly ensuring the stable operation of the system and the timeliness of data analysis.

[0070] Queue length detection: During the weekday evening rush hour, node C detected that the queue length of vehicles merging from the secondary arterial road into the main arterial road at the intersection reached 100 meters (exceeding the critical threshold of 80 meters set based on the road type and historical traffic flow data). Furthermore, based on the traffic flow model, it was predicted that the queue might overflow to the upstream road section within the next 3 minutes. Node C immediately triggered collaboration with upstream node B and downstream node D to initiate a collaborative analysis task, jointly adjusting signal timing (such as extending the green light duration of the secondary arterial road to speed up the merging of vehicles into the main arterial road) and replanning traffic routes (guiding some vehicles to divert from the side roads) to alleviate traffic pressure.

[0071] Traffic flow parameters: On a weekday morning, the traffic flow on the side road where node D is located, which was originally stable, suddenly experienced a drop in vehicle speed of more than 40% (significantly different from the historical average vehicle speed for the same period and the prediction of normal traffic flow model), and this lasted for more than 8 minutes. Node D judged that there was an abnormal change in local traffic speed, initiated a collaborative analysis task, shared data with nodes A, B, and E, and conducted collaborative analysis to find the cause of the speed drop (such as whether there was a temporary malfunction at the parking lot entrance causing vehicle backlog, etc.) in order to take appropriate traffic management measures.

[0072] Example of an event-based trigger:

[0073] Accident Detection: Node E detected a collision between two vehicles on the ring road within the commercial area using camera image recognition technology. This caused partial road congestion, affecting the normal circulation of vehicles within the area and the traffic order for entering and exiting the commercial area. This is a major accident affecting traffic in a wider area. Node E immediately triggered a collaborative analysis task and sent the accident information to other nodes. The nodes collaboratively analyzed the scope of the accident's impact on the surrounding roads (such as which intersections and road sections would experience traffic congestion) and jointly formulated traffic management plans (such as setting up temporary traffic control and guiding vehicles to avoid the accident section).

[0074] Requests from neighboring nodes: Suppose node A, wanting to more accurately analyze the traffic flow patterns at an intersection during different time periods of a holiday, sends information requests to nodes B, C, D, and E, hoping to obtain detailed traffic flow data for their respective areas during the corresponding time periods. Upon receiving the requests, nodes B, C, D, and E initiate a collaborative analysis task, sharing their respective collected and analyzed traffic flow data with node A according to the request, assisting node A in completing a more comprehensive traffic flow analysis.

[0075] Control System Requirements: As traffic flow in and around the commercial area dynamically changes throughout the day, the traffic signal control system determines the need to optimize signal timing at intersections based on real-time traffic conditions to improve overall traffic efficiency. At this point, the traffic signal control system sends demand signals to each edge computing node, triggering a collaborative analysis task. Each node collaboratively provides its collected traffic data (such as traffic volume, queue length, and vehicle flow direction at each intersection) to participate in the analysis and assist the traffic signal control system in dynamically adjusting parameters such as signal light duration and phase based on real-time traffic conditions, thereby achieving optimized traffic signal configuration.

[0076] After initiating collaborative analysis tasks based on the aforementioned triggering conditions, a collaborative analysis task allocation mechanism is constructed according to the computing power of each edge computing node (e.g., node A has stronger processor performance and more powerful data analysis capabilities; node D has relatively smaller storage capacity, but the collected data is crucial for traffic analysis around the parking lot), the coverage of the collected data (node ​​A covers the entire intersection, while node D mainly focuses on the parking lot side road area), and the traffic importance (the main road nodes A and B have relatively high traffic importance, while side road nodes D and E also play a key role in traffic management during specific periods). This mechanism clarifies the specific responsibilities and tasks of each node in the collaborative analysis. For example, in the collaborative analysis task triggered by the aforementioned intersection accident, node E is responsible for continuously providing detailed image data, vehicle location, and status information of the accident scene. Node A uses its powerful analysis capabilities combined with traffic data from surrounding intersections to assess the impact of the accident on the overall traffic flow in the area. Nodes B, C, and D assist in analyzing feasible vehicle diversion routes and predicting the degree of congestion on surrounding roads based on traffic data from their respective areas, jointly completing the traffic management-related analysis tasks.

[0077] Data collaboration, sharing, and integration:

[0078] Each edge computing node, according to its task allocation, shares its extracted traffic data feature vectors with the network via encrypted network transmission (using the AES encryption algorithm to encrypt the data), ensuring the security of data transmission. For example, node A sends out data such as traffic flow characteristics at an intersection and vehicle turning characteristics in each direction, while node D shares vehicle entry and exit flow characteristics and queue length characteristics of parking lot side streets with other nodes.

[0079] After receiving the shared data, the remaining nodes use weighted fusion and feature splicing methods to fuse the shared data into a comprehensive dataset reflecting the traffic situation. For weighted fusion, considering that node A's data acquisition equipment has higher accuracy and its local confidence level for intersection traffic analysis is relatively high (e.g., 0.8), node A's data is assigned a higher weight (e.g., 0.6) when fusing traffic flow feature data, while the parking lot branch road data of node D has a relatively lower weight (e.g., 0.4). Based on this weight allocation, similar traffic flow feature data from different nodes are fused and calculated. For feature splicing, the intersection vehicle turning features of node A are spliced ​​and combined with the main road vehicle flow direction features of node B to form a more comprehensive description of the traffic flow situation. This allows each node to obtain more complete traffic information from the fused dataset, thus providing strong data support for subsequent collaborative analysis. The fused dataset is then transmitted to the analysis and decision-making units of each node.

[0080] Collaborative Analysis and Decision Making:

[0081] The analysis and decision-making units of each edge computing node participate in collaborative analysis based on the fused dataset, using traffic flow models (models built on fluid dynamics principles and traffic engineering theories, which simulate the formation, propagation, and dissipation of traffic flow throughout the region by inputting parameters such as fused traffic density, speed, and flow rate) and machine learning prediction models (such as using support vector machines to classify and predict traffic congestion, using long short-term memory networks to analyze the time series change trend of traffic flow, and using historical traffic data from the past few years and real-time fused data for model training).

[0082] For example, if collaborative analysis predicts that a large number of vehicles will emerge from the parking lot of a shopping mall in a commercial area within the next 30 minutes due to the end of a promotional event, potentially causing congestion on surrounding roads, each node generates traffic management decision-making suggestions based on the analysis results. For instance, node A suggests adjusting the duration of the traffic lights at its intersection leading to the parking lot, appropriately increasing the green light duration to expedite vehicle dispersal; node B proposes guiding vehicles on the main road to adjacent side roads in advance to alleviate pressure on the main road; and node D decides to set up temporary traffic guides at the parking lot exit to direct vehicles to exit in an orderly manner. These decision-making suggestions are then fed back to the corresponding traffic management system in standardized XML format for execution. The traffic management department can then take timely measures based on these suggestions to optimize traffic conditions.

[0083] Feedback and optimization:

[0084] After the traffic management system executes the decision suggestions fed back by each node, each edge computing node continuously monitors changes in traffic conditions and collects feedback information. For example, node A monitors whether the average vehicle transit time at the intersection has been shortened and whether congestion has been alleviated after adjusting the traffic light duration; node D observes whether the vehicle queue length at the parking lot exit has been effectively controlled.

[0085] Each node statistically analyzes the collected feedback information, comparing the actual results with the expected decision-making outcomes. If adjusting traffic light durations fails to alleviate congestion as expected, the feedback optimization unit in the collaborative analysis module where node A resides will optimize and adjust the relevant elements involved in the collaborative analysis. For example, it might adjust the weights of data fusion, perhaps due to insufficient attention previously paid to the main road speed data of node B, and reallocate the weights to increase their proportion in the fused data; it might also reallocate collaborative analysis tasks, considering allowing node C to participate more in the analysis related to traffic management at the intersection; it might correct the parameters of the collaborative analysis model, fine-tuning the flow coefficient for the intersection in the traffic flow model, and retraining the weight parameters in the machine learning prediction model based on actual traffic data. Through this feedback and optimization process, the accuracy and effectiveness of collaborative analysis are continuously improved, enabling the entire system to better adapt to complex and ever-changing traffic environments and continuously provide high-quality decision support for traffic management.

[0086] The above specific implementation examples demonstrate the application of a business logic-based cross-system data collaborative intelligent analysis method and system in real-world traffic scenarios, showcasing the advantages and practicality of this invention in addressing complex traffic conditions, achieving efficient data collaborative analysis, and supporting traffic management decisions. In different traffic scenarios, the system deployment and specific operations of each step can be flexibly adjusted according to actual road layouts, traffic flow characteristics, and other factors to meet diverse traffic management needs.

[0087] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0088] 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 cross-system data collaborative intelligent analysis method based on business logic, characterized in that: Includes the following steps: Data Acquisition and Preprocessing: Edge computing nodes are deployed at multiple key traffic locations. Each edge computing node is equipped with at least one data acquisition device to collect real-time traffic data within its area. The collected data is then preprocessed by removing noise, standardizing the format, and correcting outliers. Inter-node communication and networking: Each edge computing node constructs a local network through wired or wireless communication. Communication protocols and data interaction rules are determined through identity authentication and network configuration to achieve communication connections between them. Data feature extraction and local analysis: Each edge computing node uses deep learning algorithms or statistical analysis methods to extract features from the preprocessed local real-time traffic data, transforming the raw traffic data into feature vectors and conducting local analysis. Collaborative analysis task triggering and allocation: Collaborative analysis tasks are initiated based on predefined triggering conditions, and tasks are allocated based on relevant factors of each edge computing node; Data collaboration, sharing and fusion: Each edge computing node shares its extracted traffic data feature vectors to the network according to its assigned tasks. The remaining nodes then use weighted fusion and feature splicing methods to fuse the shared data to form a comprehensive data set that reflects the traffic situation. Collaborative Analysis and Decision Making: Based on the fused dataset, each edge computing node uses traffic flow models and machine learning prediction models to participate in collaborative analysis, generate traffic management decision suggestions, and feed them back to the traffic management system for execution. Feedback and Optimization: Based on the traffic condition feedback information after the traffic management system implements decision-making suggestions, each edge computing node optimizes and adjusts the relevant elements involved in the collaborative analysis; The predefined triggering conditions include quantitative triggers: Local analysis confidence: When the confidence of an edge node in its local assessment is lower than a preset threshold, it requests input from neighboring nodes to trigger a collaborative analysis task. Resource utilization: If a node's processing load or available bandwidth exceeds predefined limits, the node will offload some analysis tasks or request assistance to trigger collaborative analysis tasks. Queue length detection: When the queue length is detected to exceed the critical threshold set based on road type and historical traffic flow data, collaboration with upstream and downstream nodes is triggered to adjust signal timing or replan traffic routes, thereby initiating a collaborative analysis task. Traffic flow parameters: When local traffic speed, density, or flow rate shows significant abnormal changes compared to historical or expected patterns, a collaborative analysis task is initiated; The predefined triggering conditions also include event-based triggers: Accident detection: When an edge computing node determines that a major accident has occurred that affects traffic in a wider area, it triggers a collaborative analysis task; Requests from neighboring nodes: When receiving information requests from neighboring edge nodes, initiate a collaborative analysis task and participate in collaborative work according to the request content; Control system requirements: When the traffic signal control system determines that multi-node data is needed to optimize signal timing, a collaborative analysis task is triggered, in which each node collaboratively provides data and participates in the analysis.

2. The intelligent cross-system data collaboration analysis method based on business logic according to claim 1, characterized in that, In the data feature extraction and local analysis steps, the deep learning algorithm is a convolutional neural network. It is trained using labeled or unlabeled traffic data by constructing a neural network structure containing multiple convolutional layers, pooling layers and fully connected layers to learn and extract feature patterns from the data.

3. The intelligent cross-system data collaboration analysis method based on business logic according to claim 1, characterized in that, The data feature extraction and local analysis steps include conducting local analysis, specifically including: Traffic parameter calculations: Calculate local traffic density, measured by the number of vehicles passing through a unit road segment per unit time; calculate local traffic flow rate, determined based on the frequency of vehicles passing through a specific monitoring point within a specific time period; calculate local average speed, derived from the ratio of vehicle distance traveled to the time taken. Local event detection: Identifies local events through preset rules. When a vehicle's position does not change within a preset time, it is determined to be a stopped vehicle; when multiple vehicles are traveling at speeds consistently below normal and the distance between them is gradually decreasing, it is determined to be a queue formation; when a vehicle's actual speed exceeds the speed limit of the road segment, it is determined to be speeding. Confidence level assessment: Taking into account factors such as camera field of view obstruction, radar signal interference, and induction coil stability during data acquisition, the corresponding confidence levels are assigned to each local analysis result.

4. The intelligent cross-system data collaboration analysis method based on business logic according to claim 1, characterized in that, The task allocation based on relevant factors of each edge computing node includes: constructing a collaborative analysis task allocation mechanism based on the computing power of each edge computing node, the coverage of the collected data, and the importance of traffic, reasonably breaking down the overall traffic data analysis task, and clarifying the specific responsibilities and tasks of each node in the collaborative analysis.

5. The intelligent cross-system data collaboration analysis method based on business logic according to claim 1, characterized in that, The machine learning prediction model includes support vector machines and long short-term memory networks. It is trained using historical traffic data and real-time fused data to predict the occurrence and spread of traffic congestion and analyze the impact of traffic accidents on surrounding traffic.

6. The intelligent cross-system data collaboration analysis method based on business logic according to claim 1, characterized in that, The collaborative analysis tasks include traffic flow prediction, intersection traffic congestion analysis, road segment traffic bottleneck analysis, and regional traffic situation assessment.

7. The intelligent cross-system data collaboration analysis method based on business logic according to claim 1, characterized in that, In the feedback and optimization steps, the optimization adjustment specifically includes adjusting the weights of data fusion, reallocating collaborative analysis tasks, and correcting the parameters of the collaborative analysis model.

8. A cross-system data collaboration intelligent analysis system based on business logic, designed based on the cross-system data collaboration intelligent analysis method based on business logic as described in any one of claims 1-7, characterized in that, This includes multiple edge computing nodes deployed in key transportation locations. Each edge computing node contains: The data acquisition module is used to collect real-time traffic data within the area using the equipped data acquisition equipment, and transmit the collected data to the preprocessing module; The preprocessing module is used to perform preprocessing operations on the received data, such as removing noise, standardizing data format, and correcting outliers, and then transmits the preprocessed data to the feature extraction module. The feature extraction module is used to extract features from the preprocessed data using deep learning algorithms or traditional statistical analysis methods, generate feature vectors, perform local analysis operations, and transmit the feature vectors and related local analysis results to the collaborative analysis module. The collaborative analysis module includes a task allocation unit, a data sharing and fusion unit, an analysis and decision-making unit, and a feedback optimization unit; The task allocation unit is used to construct a collaborative analysis task allocation mechanism based on the computing power of each edge computing node, the coverage of the collected data, and traffic importance factors. It rationally breaks down the overall traffic data analysis task and clarifies the specific responsibilities and tasks of each node in the collaborative analysis. The data sharing and fusion unit is used to share the relevant traffic data feature vectors extracted by itself to the network according to the task allocation, and to receive data shared by other nodes. It uses weighted fusion and feature splicing methods to fuse these data to form a comprehensive and more complete data set reflecting the traffic situation, and then transmits the data set to the analysis and decision-making submodule. The analysis and decision-making unit is used to analyze traffic conditions based on the fused data set, using traffic flow models and machine learning prediction models, and to generate corresponding traffic management decision-making suggestions based on the analysis results, and then feed the decision-making suggestions back to the corresponding traffic management system for execution. The feedback optimization unit is used to receive traffic condition change information after the traffic management system implements decision suggestions, and to optimize and adjust the model parameters, task allocation mechanism, and data fusion method of collaborative analysis.

Images