Vehicle driving electronic information system

Abstract

The invention discloses a vehicle driving electronic information system, which comprises an edge computing node, a central control node, a path planning module, a navigation instruction module and a feedback module, and realizes timely response and dynamic strategy adjustment of sudden abnormal events through a real-time data stream processing technology and a reinforcement learning technology. The response speed and flexibility of the system are improved, the data cleaning algorithm and the multi-source data fusion technology are adopted, the multi-source data are processed through the Kalman filtering algorithm and the Bayesian estimation algorithm, and the accuracy and reliability of the data are improved. Through the deep learning technology and the transfer learning technology, the recognition effect under new vehicle models and special conditions is remarkably improved, and the phenomena of misrecognition and missed recognition are reduced. The automatic and intelligent data cleaning and preprocessing reduces manual intervention, improves the efficiency and accuracy of data processing, updates the global model through the federated learning technology, and adapts to data processing requirements in different scenes.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a vehicle driving electronic information system. Background Art

[0002] With the rapid development of the automotive industry and the continuous advancement of automotive electronic technology, information management and processing during vehicle driving have become increasingly important. Modern cars not only need to monitor and record the vehicle's operating status in real time, but also need to efficiently interact with drivers, passengers and the external environment.

[0003] After searching, the invention patent with Chinese patent number CN116994441A discloses a method and system for processing traffic flow information data. The method first obtains the vehicle passing data of each checkpoint in the target area during the set time period; obtains vehicle information in the motor vehicle information system through the motor vehicle license plate number; then calculates and integrates the obtained passing time, passing speed and vehicle information data to determine the traffic flow coupling data of each checkpoint in the target area during the set time period; finally, traffic anomaly identification is performed based on the vehicle flow passing through each checkpoint, and the traffic anomaly value is corrected. The present application can integrate vehicle driving information, restore the anomaly value, and obtain traffic flow information data of the entire road network in the area at different times. Compared with the prior art, the invention patent with Chinese patent number CN116994441A pre-processes the original data according to 1 hour and 1 day as statistical units, identifies, matches and restores the severely abnormal data; performs fuzzy matching on vehicles whose license plate numbers cannot be correctly identified, and matches the corresponding emission standards according to the average age of similar models.

[0004] However, in actual use, the data processing of the above invention relies on a fixed time window (such as 1 hour and 1 day), and cannot respond to sudden abnormal events in a timely manner; data cleaning and preprocessing rely on manual intervention, and the processing efficiency and accuracy may be limited; license plate recognition relies on limited historical data, and the recognition effect may be poor for new models or in special circumstances.

[0005] Therefore, a vehicle driving electronic information system is proposed. Summary of the invention

[0006] The purpose of the present invention is to solve the shortcomings of the prior art that data processing depends on a fixed time window (such as 1 hour and 1 day) and cannot respond to sudden abnormal events in a timely manner; data cleaning and preprocessing rely on manual intervention, and the processing efficiency and accuracy may be limited; license plate recognition relies on limited historical data, and the recognition effect may be poor for new models or under special circumstances. A vehicle driving electronic information system is proposed.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A vehicle driving electronic information system, comprising:

[0009] Edge computing nodes are used to obtain vehicle location and speed information in real time;

[0010] The central control node is used to receive data from edge computing nodes and perform fusion processing;

[0011] The path planning module is used to generate a real-time traffic situation map based on the fused data and calculate the shortest path;

[0012] A navigation instruction module, used to send navigation instructions to the driver through the vehicle-mounted device;

[0013] Feedback module, used to adjust the path planning according to the driver's driving conditions.

[0014] The above technical solution further includes:

[0015] Preferably, the central control node further includes a data fusion module, which is used to:

[0016] Receive data from multiple edge computing nodes;

[0017] De-noise, verify, convert formats and calibrate timestamps on data;

[0018] Use Kalman filtering algorithm and Bayesian estimation algorithm to process multi-source data;

[0019] Generate real-time traffic situation map.

[0020] Preferably, the path planning module further comprises:

[0021] Use Dijkstra algorithm and A* algorithm to calculate the shortest path;

[0022] Improve path search efficiency.

[0023] Preferably, the system further comprises a distributed machine learning framework for:

[0024] Collect and pre-process vehicle location and speed information in real time;

[0025] Deploy and process data in parallel at edge nodes;

[0026] Use GPU to accelerate the training process and use SGD to update model parameters;

[0027] Upload the model to the central control node for federated learning updates;

[0028] Send the updated model to the edge nodes.

[0029] Preferably, the distributed machine learning framework further comprises:

[0030] Transfer learning module to accelerate model training;

[0031] Incremental learning module for updating models in real time;

[0032] Use cross-validation methods to evaluate model performance;

[0033] A grid search method is used to find the optimal hyperparameter combination.

[0034] Preferably, it further includes a situational awareness warning module, which is used to:

[0035] Collect weather and time information;

[0036] Use ensemble learning methods to process multi-source information;

[0037] Use random forest models to assess risk;

[0038] Generate and send warning information through on-board equipment;

[0039] Use message queue technology to ensure reliable message delivery;

[0040] Collect driver feedback to optimize early warning models.

[0041] Preferably, the situational awareness warning module further comprises:

[0042] Use deep learning techniques to improve situational awareness accuracy;

[0043] Optimize decision-making models using reinforcement learning techniques.

[0044] Preferably, it further includes an intelligent traffic signal collaborative control module for:

[0045] Collect vehicle location and speed information in real time;

[0046] Use LSTM model to predict traffic flow;

[0047] Dynamically adjust traffic light timing strategies;

[0048] Use genetic algorithms to determine the best timing solution;

[0049] Send optimized signal timing strategy;

[0050] Real-time monitoring of traffic light effects;

[0051] Use real-time data stream processing platforms for monitoring;

[0052] Use reinforcement learning techniques to optimize traffic light timing strategies;

[0053] Use convolutional neural network (CNN) to improve traffic flow prediction accuracy.

[0054] Preferably, the intelligent traffic signal collaborative control module further includes:

[0055] Use reinforcement learning techniques to optimize traffic light timing strategies;

[0056] Use convolutional neural network (CNN) to improve traffic flow prediction accuracy.

[0057] Preferably, it further comprises a multi-source data stream processing module, which is used to:

[0058] Collect data from multiple sensors in real time;

[0059] Fusion processing data;

[0060] Integrate data using Bayesian estimation algorithms;

[0061] Use stream processing engines to process data in real time;

[0062] Use the MapReduce model to generate real-time analysis results;

[0063] Use window functions to process data;

[0064] Generate and send real-time decision instructions through onboard equipment;

[0065] Collect driver feedback;

[0066] Optimize decision models using incremental learning methods;

[0067] Optimizing decision making using graph neural networks (GNNs).

[0068] The present invention has the following beneficial effects:

[0069] 1. In the present invention, timely response and dynamic adjustment strategies for sudden abnormal events are achieved through real-time data stream processing technology and reinforcement learning technology, which greatly improves the response speed and flexibility of the system. Data cleaning algorithm and multi-source data fusion technology are adopted, and multi-source data are processed through Kalman filtering algorithm and Bayesian estimation algorithm to improve the accuracy and reliability of data. Through deep learning technology (such as CNN) and transfer learning technology, the recognition effect in new models and special circumstances has been significantly improved, reducing the phenomenon of misidentification and missed recognition. Automated and intelligent data cleaning and preprocessing not only reduce manual intervention, but also greatly improve the efficiency and accuracy of data processing.

[0070] 2. In the present invention, multi-dimensional data display is realized through visualization tools, and real-time updated data display function is provided to meet the needs of rapidly changing traffic environments. The combination of real-time data stream processing technology and visualization tools not only enhances the user experience, but also improves the real-time and dynamic nature of data display. The global model is updated through federated learning technology, which further improves the generalization ability and robustness of the model and adapts to the data processing requirements in different scenarios. The application of multi-source data fusion technology ensures the overall quality and reliability of the data, making the system perform better in complex and changing traffic environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a system architecture diagram of a vehicle driving electronic information system proposed by the present invention. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0073] like Figure 1 As shown, a vehicle driving electronic information system proposed by the present invention includes:

[0074] Edge computing nodes are used to obtain vehicle location and speed information in real time;

[0075] The central control node is used to receive data from edge computing nodes and perform fusion processing;

[0076] The path planning module is used to generate a real-time traffic situation map based on the fused data and calculate the shortest path;

[0077] A navigation instruction module, used to send navigation instructions to the driver through the vehicle-mounted device;

[0078] Feedback module, used to adjust the path planning according to the driver's driving conditions.

[0079] In one embodiment, an edge computing node:

[0080] The main function of the edge computing node is to obtain the vehicle's location and speed information in real time and upload this information to the central control node.

[0081] Data collection:

[0082] GPS module: obtain vehicle location information in real time.

[0083] Sampling frequency: 1Hz

[0084] Accuracy: ±5 meters

[0085] Speed ​​sensor: obtain vehicle speed information in real time.

[0086] Sampling frequency: 10Hz

[0087] Accuracy: ±0.5km / h

[0088] Environmental sensor: obtain environmental information, such as temperature, humidity, etc.

[0089] Sampling frequency: 1Hz

[0090] Accuracy: ±1℃, ±5%RH

[0091] Camera: Get images of the surrounding environment in real time.

[0092] Resolution: 1080p

[0093] Frame rate: 30fps

[0094] Radar: Get real-time information about obstacles ahead.

[0095] Detection range: 0-200 meters

[0096] Accuracy: ±1 meter

[0097] Data upload:

[0098] Wireless communication module: Use 4G or 5G wireless network to upload the collected location information, speed information, environmental information, image information and obstacle information to the central control node.

[0099] Bandwidth requirement: at least 5Mbps

[0100] Delay requirement: ≤50ms

[0101] Data packaging module: Packs different types of sensor data into a unified format for easy uploading and processing.

[0102] Data format: JSON

[0103] Encryption: Use AES-256 encryption algorithm

[0104] In one embodiment, the central control node:

[0105] The main function of the central control node is to receive data from the edge computing nodes, perform fusion processing, and generate a real-time traffic situation map.

[0106] Data reception and preprocessing:

[0107] Data receiving module: receives data from multiple edge computing nodes.

[0108] Receiving frequency: 1Hz

[0109] Storage capacity: at least 1TB

[0110] Data preprocessing module: denoises, verifies and converts data formats to ensure data consistency and accuracy.

[0111] Denoising module: Use the median filter algorithm to remove invalid or abnormal data.

[0112] Data verification module: Use CRC check code to check the integrity and accuracy of data.

[0113] Data format conversion module: converts data in different formats into a unified format (such as JSON).

[0114] Data fusion and processing:

[0115] Data fusion module:

[0116] The timestamps of each data source are calibrated to ensure that all data are fused and processed under the same time reference.

[0117] Time synchronization: Use the NTP protocol for time synchronization.

[0118] The Kalman filter algorithm is used to fuse data from multiple data sources.

[0119] Kalman filter formula:

[0120]

[0121] P k|k-1 =AP k-1|k-1 A T +Q

[0122] K k =P k|k-1 H T (HP k|k-1 H T +R) -1

[0123]

[0124] P k|k =(IK k H)P k|k-1

[0125] State transfer: represents the position and velocity prediction at the current time k.

[0126] Prediction error covariance: P k|k-1 Describes the uncertainty in position and velocity prediction errors.

[0127] Kalman gain: K k Indicates the weight of the fused measurement (such as GPS position).

[0128] Status Update: It is the state estimate after integrating GPS measurements.

[0129] Error covariance update: P k|k Describes the updated error covariance.

[0130] Use Bayesian estimation algorithm to integrate multi-source data and generate comprehensive road condition information.

[0131] Bayesian estimation formula:

[0132]

[0133] Prior probability: p(x) represents the probability distribution of position and velocity without considering the current measurement data.

[0134] Likelihood function: p(y|x) represents the probability of a measurement (such as a GPS position) given its position and velocity.

[0135] Posterior probability: p(x|y) represents the probability distribution of the state after taking into account the current measurement value.

[0136] Traffic situation generation module:

[0137] Based on the fused data set, a real-time traffic situation map is generated, including information such as traffic flow, congestion, and average speed of each road section.

[0138] Map data: Use OpenStreetMap data to generate traffic situation maps.

[0139] In one embodiment, the path planning module:

[0140] The main function of the path planning module is to generate a real-time traffic situation map based on the fused data and calculate the shortest path.

[0141] Shortest path calculation:

[0142] Dijkstra algorithm module: Use Dijkstra algorithm to calculate the shortest path.

[0143] Initialization: The distance of the starting point is 0, and the distances of the remaining nodes are infinite.

[0144] Priority Queue: Use Fibonacci heap for priority sorting.

[0145] Iterative process: Each time, take out the node with the smallest distance from the priority queue and update the distance of its neighboring nodes.

[0146] Dijkstra algorithm formula:

[0147] d[v]=min(d[v],d[u]+w(u,v))

[0148] Distance update: d[v] represents the shortest path length from the starting point to node v.

[0149] Distance calculation: d[u]+w(u,v) represents the distance from the starting point to the node u plus the distance from u to v.

[0150] A* algorithm module: Use the A* algorithm to improve path search efficiency.

[0151] Heuristic function: Use Manhattan distance as the heuristic function.

[0152] Priority Queue: Use a min-heap for priority sorting.

[0153] A* algorithm formula:

[0154] f(n)=g(n)+ (n)

[0155] Actual distance: g(n) represents the actual distance from the starting point to the current node n.

[0156] Estimated distance: (n) represents the estimated distance from the current node n to the target node (such as Manhattan distance).

[0157] Path backtracking:

[0158] Path backtracking module: Starting from the end point, according to the recorded predecessor node information, backtrack to the starting point one by one to obtain the shortest path.

[0159] Predecessor node record: record the predecessor node before each node arrives.

[0160] In one embodiment, the navigation instruction module:

[0161] The main function of the navigation instruction module is to send navigation instructions to the driver through the on-board equipment.

[0162] Navigation command generation:

[0163] Instruction generation module: Generates navigation instructions based on the calculated shortest path, including specific operations such as turning and changing lanes.

[0164] Instruction format: text and voice prompts.

[0165] Specific instructions: such as "turn left 100 meters ahead."

[0166] Navigation command sending:

[0167] Instruction sending module: sends navigation instructions to the driver through on-board equipment (such as display screen or voice prompt system).

[0168] Display: Use LCD display screen.

[0169] Voice prompt system: Use TTS (Text-to-Speech) technology for voice prompts.

[0170] In one embodiment, the feedback module:

[0171] The main function of the feedback module is to adjust the path planning according to the driver's driving conditions.

[0172] Feedback information collection:

[0173] Driving status monitoring module: monitors in real time whether the driver is driving according to navigation instructions.

[0174] Monitoring frequency: 1Hz

[0175] Data upload: Use 4G / 5G network to upload data to the central control node.

[0176] Data upload module: upload the driver's driving conditions to the central control node.

[0177] Data format: JSON

[0178] Encryption: Use TLS encryption protocol.

[0179] Feedback information processing:

[0180] Feedback processing module: adjusts the path planning according to the uploaded driving conditions to ensure the real-time and accuracy of the path planning.

[0181] Data parsing: Parse the uploaded data.

[0182] Path Adjustment: If deviation is detected, the path is recalculated and new navigation instructions are sent.

[0183] Adjustment algorithm: Use dynamic programming algorithm to adjust the path.

[0184] In one embodiment, the distributed machine learning framework:

[0185] The main function of the distributed machine learning framework is to perform data preprocessing and model training on edge computing nodes, and update the global model through federated learning.

[0186] Data preprocessing:

[0187] Data preprocessing module: preprocesses the collected data on the edge computing node, including data cleaning and normalization.

[0188] Data cleaning: fill missing values ​​with the mean.

[0189] Data normalization: Use Z-score to normalize the data.

[0190] Data format: Convert to a data format supported by TensorFlow.

[0191] Model training:

[0192] Model training module:

[0193] Deploy distributed machine learning frameworks (such as TensorFlow) on edge nodes.

[0194] Process data in parallel and use GPU to accelerate model training.

[0195] Stochastic gradient descent (SGD) is used for parameter updates.

[0196] Training algorithm: Use Adam optimizer for parameter update.

[0197] Adam optimizer formula:

[0198]

[0199] First moment estimate: m t Used to estimate the gradient of model parameters.

[0200] Second moment estimate: v t Used to estimate the square of the gradient.

[0201] Unbiased estimate: and It is an unbiased estimate of the first and second order moments.

[0202] Parameter update: θ t+1 are the model parameters updated using unbiased estimates.

[0203] Among them, θ t is a parameter, is the gradient of the loss function with respect to the parameters, α is the learning rate, β1 and β2 are exponential decay rates, and ∈ is a small constant for numerical stability.

[0204] Model upload and update:

[0205] Model upload module: upload the trained model to the central control node.

[0206] Upload frequency: upload once every hour.

[0207] Data format: The model is saved in HDF5 format.

[0208] Model update module: The central control node updates the global model through federated learning and sends the updated model to the edge nodes.

[0209] Federated Learning: Aggregate model parameters using Secure Aggregation technology.

[0210] Model delivery: Use FTP protocol to deliver the model.

[0211] In one embodiment, transfer learning and incremental learning:

[0212] The main functions of the transfer learning and incremental learning modules are to accelerate model training and update the model in real time.

[0213] Transfer Learning:

[0214] Transfer learning module: Use existing pre-trained models to accelerate the training process of new models.

[0215] Pre-trained model: Use the ResNet model pre-trained on ImageNet.

[0216] Fine-tuning: Perform fine-tuning training on a new dataset.

[0217] Fine-tuning formula:

[0218]

[0219] Parameter optimization: θ new are the model parameters optimized on a new dataset (such as local traffic data).

[0220] Loss function: It is a loss function used to evaluate the gap between the model output and the actual label (such as traffic conditions).

[0221] Among them, f θ (x) is the model output and y is the true label.

[0222] Incremental Learning:

[0223] Incremental learning module: Updates model parameters based on real-time data to achieve continuous optimization of the model.

[0224] Incremental data processing: Use Mini-BatchSGD for incremental learning.

[0225] Online Update: Use online learning algorithms to update the model in real time.

[0226] Online learning formula:

[0227]

[0228] Parameter update: θ t+1 are the model parameters updated using the gradient of the current sample.

[0229] Learning rate: η t is the learning rate that varies over time.

[0230] in, is the gradient of the loss function with respect to the current parameters.

[0231] In one embodiment, the situational awareness warning module

[0232] The main function of the situational awareness warning module is to generate warning information and send it to the driver by collecting situational information such as weather and time.

[0233] Situational Information Collection:

[0234] Data collection module: collects weather information and time information.

[0235] Weather station data: Use the API provided by the China Meteorological Administration to obtain weather information.

[0236] Time information: Use the RTC module to obtain the current time.

[0237] Data fusion module: integrates information from multiple data sources to generate comprehensive situational information.

[0238] Data fusion algorithm: Use Bagging ensemble learning method to fuse multiple data sources.

[0239] Comprehensive situational information: including weather, time and other information.

[0240] risk assessment:

[0241] Risk Assessment Module:

[0242] Risk assessment was performed using random forest models.

[0243] Each scenario is categorized and a risk score is generated.

[0244] Random Forest Formula:

[0245]

[0247] Predicted value: is the average of all tree predictions.

[0248] Number of trees: B is the number of trees in the random forest.

[0249] Among them, fb (X) is the output of the b-th tree.

[0250] Generation and sending of early warning information:

[0251] Warning information generation module: Generates warning information based on the evaluation results, such as "Heavy rain is expected on the road ahead, please slow down."

[0252] Warning information format: text and voice prompts.

[0253] Warning information sending module: Send warning information through vehicle-mounted equipment and use message queue technology to ensure reliable message delivery.

[0254] In-vehicle equipment: Use TTS technology for voice prompts.

[0255] Message Queue: Use RabbitMQ to ensure reliable message delivery.

[0256] Feedback and Optimization:

[0257] Feedback information collection module: collects drivers' feedback on warning information.

[0258] Feedback mechanism: Use on-board equipment to collect driver feedback.

[0259] Data upload: Use 4G / 5G network to upload data to the central control node.

[0260] Model optimization module: Use online learning methods to optimize the early warning model to improve the accuracy and practicality of the early warning.

[0261] Online learning: Use online learning algorithms to gradually optimize the early warning model.

[0262] Model update: Update model parameters in real time.

[0263] In one embodiment, deep learning and reinforcement learning:

[0264] The main functions of the deep learning and reinforcement learning modules are to improve the accuracy of situational awareness and optimize decision-making models.

[0265] Deep Learning:

[0266] Deep Learning Module: Use deep learning technology to improve situational awareness accuracy.

[0267] Convolutional Neural Network (CNN): Use the VGG16 model for context recognition.

[0268] Training data: Use the ImageNet dataset for pre-training.

[0269] CNN formula:

[0270]

[0271] Activation value: is the activation value of the ith unit in layer l.

[0272] Weighted sum: is the result of weighted summation.

[0273] Weights and biases: and are weights and bias terms.

[0274] 10.2 Reinforcement Learning

[0275] Reinforcement Learning Module: Use reinforcement learning techniques to optimize decision-making models.

[0276] Q-Learning: Use the Q-Learning algorithm to optimize the decision model.

[0277] Reward function: Set the reward function based on context information.

[0278] Q-Learning formula:

[0279]

[0280] application:

[0281] Value function update: Q(s,a) is the value function of a state-action pair.

[0282] Immediate reward: r is the immediate reward (such as the reward for avoiding congestion).

[0283] Discount factor: γ is the importance of future rewards.

[0284] Learning rate: α is the learning rate.

[0285] In one embodiment, the intelligent traffic signal coordinated control module

[0286] The main function of the intelligent traffic signal collaborative control module is to predict traffic flow and dynamically adjust the signal light timing strategy.

[0287] Traffic flow prediction:

[0288] Data collection module: collects vehicle location and speed information in real time.

[0289] Acquisition frequency: 1Hz

[0290] Data upload: Use 4G / 5G network to upload data to the central control node.

[0291] Traffic prediction module: Use the LSTM model to predict traffic flow in the future.

[0292] Model training: Use the past month’s traffic data for training.

[0293] Forecast time window: predict traffic flow within the next 10 minutes.

[0294] LSTM formula:

[0295] i t =σ(W ix x t +W it-1 +b i )

[0296] f t =σ(W fx x t +W ft-1 +b f )

[0297]

[0298] o t =σ(W ox x t +W ot-1 +b o )

[0299]

[0300] Input gate: i t Control input information (such as historical traffic flow data).

[0301] Forget gate: f t Control what information is retained.

[0302] Unit status: C t is the state of the memory cell.

[0303] Output gate: o t Controls the output information.

[0304] Hidden state: t is the hidden state at the current moment (such as predicted traffic flow).

[0305] Traffic light timing strategy adjustment:

[0306] Timing strategy adjustment module:

[0307] Dynamically adjust traffic light timing strategies based on predicted traffic flow.

[0308] Use genetic algorithms to determine the best timing solution.

[0309] Genetic Algorithms: Implemented Genetic Algorithms using the DEAP library in Python.

[0310] Delivery strategy: Use the FTP protocol to deliver the optimized signal timing strategy.

[0311] Real-time monitoring and optimization:

[0312] Real-time monitoring module: monitors the effect of traffic lights in real time and continuously adjusts the timing strategy based on feedback information.

[0313] Monitoring frequency: monitoring is performed once every minute.

[0314] Monitoring data: including traffic flow, waiting time and other information.

[0315] Optimization module: Use reinforcement learning technology to optimize traffic light timing strategies.

[0316] Reinforcement Learning: Use Q-Learning algorithm to optimize traffic light timing strategy.

[0317] CNN module: Use convolutional neural network (CNN) to improve the accuracy of traffic flow prediction.

[0318] In one embodiment, the multi-source data stream processing engine:

[0319] The main function of the multi-source data stream processing engine is to collect data from multiple sensors in real time, perform fusion processing, and generate real-time decision instructions.

[0320] Data collection and integration:

[0321] Data collection module: collects data from multiple sensors such as cameras, radars, and global positioning systems (GPS) in real time.

[0322] Acquisition frequency: 1Hz

[0323] Data upload: Use 4G / 5G network to upload data to the central control node.

[0324] Data fusion module: Fusion processing data, using Bayesian estimation algorithm to integrate data.

[0325] Data fusion algorithm: Use Bayesian estimation algorithm to integrate multi-source data.

[0326] Comprehensive traffic information: including traffic flow, congestion and other information.

[0327] Data processing and analysis:

[0328] Data processing module:

[0329] Use a stream processing engine such as Apache Flink to process data in real time.

[0330] Use the MapReduce model to generate real-time analysis results.

[0331] Use window functions to process data.

[0332] Processing frequency: Data is processed once per second.

[0333] Processing result: Generate a real-time traffic situation map.

[0334] Decision instruction generation and sending:

[0335] Decision instruction generation module: Generates real-time decision instructions based on the processed comprehensive road condition information.

[0336] Decision instruction formats: text and voice prompts.

[0337] Specific instructions: such as "There is an obstacle ahead, please take a detour."

[0338] Instruction sending module: sends instructions through the on-board equipment and ensures that the driver can receive them in time.

[0339] In-vehicle device: Use the display screen and voice prompt system to send instructions.

[0340] Command sending frequency: Send a command once per second.

[0341] Feedback information collection module: collects the driver's feedback on instructions.

[0342] Feedback mechanism: Use on-board equipment to collect driver feedback.

[0343] Data upload: Use 4G / 5G network to upload data to the central control node.

[0344] Model optimization module: Use incremental learning methods to optimize decision models and use graph neural networks (GNNs) to optimize decision generation.

[0345] Incremental learning: Use online learning algorithms to gradually optimize the decision model.

[0346] Graph Neural Network (GNN): Implementing Graph Neural Network using PyTorch Geometric library.

[0347] GNN formula:

[0348]

[0349] Node representation: It is the representation of the k-th layer node v (such as the node representation in the traffic network).

[0350] Weight matrix: W (k) is the weight matrix.

[0351] Activation function: σ is the activation function.

[0352] Neighbor Set: is the set of neighbors of node v (such as adjacent nodes in a road network).

[0353] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A vehicle driving electronic information system, characterized in that: include: Edge computing nodes are used to obtain vehicle location and speed information in real time; The central control node is used to receive data from edge computing nodes and perform fusion processing; The path planning module is used to generate a real-time traffic situation map based on the fused data and calculate the shortest path; A navigation instruction module, used to send navigation instructions to the driver through the vehicle-mounted device; Feedback module, used to adjust the path planning according to the driver's driving conditions.

2. A vehicle driving electronic information system according to claim 1, characterized in that: The central control node further includes a data fusion module, which is used to: Receive data from multiple edge computing nodes; De-noise, verify, convert formats and calibrate timestamps on data; Use Kalman filtering algorithm and Bayesian estimation algorithm to process multi-source data; Generate real-time traffic situation map.

3. A vehicle driving electronic information system according to claim 1, characterized in that: The path planning module further comprises: Use Dijkstra algorithm and A* algorithm to calculate the shortest path; Improve path search efficiency.

4. A vehicle driving electronic information system according to claim 1, characterized in that: Further included is a distributed machine learning framework for: Collect and pre-process vehicle location and speed information in real time; Deploy and process data in parallel at edge nodes; Use GPU to accelerate the training process and use SGD to update model parameters; Upload the model to the central control node for federated learning updates; Send the updated model to the edge nodes.

5. A vehicle driving electronic information system according to claim 4, characterized in that: The distributed machine learning framework further includes: Transfer learning module to accelerate model training; Incremental learning module for updating models in real time; Use cross-validation methods to evaluate model performance; A grid search method is used to find the optimal hyperparameter combination.

6. A vehicle driving electronic information system according to claim 1, characterized in that: Further comprising a situational awareness warning module for: Collect weather and time information; Use ensemble learning methods to process multi-source information; Use random forest models to assess risk; Generate and send warning information through on-board equipment; Use message queue technology to ensure reliable message delivery; Collect driver feedback to optimize early warning models.

7. A vehicle driving electronic information system according to claim 6, characterized in that: The situation awareness warning module further includes: Use deep learning techniques to improve situational awareness accuracy; Optimize decision-making models using reinforcement learning techniques.

8. The vehicle driving electronic information system according to claim 1, characterized in that: Further comprising an intelligent traffic signal coordinated control module for: Collect vehicle location and speed information in real time; Use LSTM model to predict traffic flow; Dynamically adjust traffic light timing strategies; Use genetic algorithms to determine the best timing solution; Send optimized signal timing strategy; Real-time monitoring of traffic light effects; Use real-time data stream processing platforms for monitoring; Use reinforcement learning techniques to optimize traffic light timing strategies; Use convolutional neural network (CNN) to improve traffic flow prediction accuracy.

9. A vehicle driving electronic information system according to claim 8, characterized in that: The intelligent traffic signal collaborative control module further includes: Use reinforcement learning techniques to optimize traffic light timing strategies; Use convolutional neural network (CNN) to improve traffic flow prediction accuracy.

10. The vehicle driving electronic information system according to claim 1, characterized in that: Further comprising a multi-source data stream processing module for: Collect data from multiple sensors in real time; Fusion processing data; Integrate data using Bayesian estimation algorithms; Use stream processing engines to process data in real time; Use MapReduce model to generate real-time analysis results; Use window functions to process data; Generate and send real-time decision instructions through onboard equipment; Collect driver feedback; Optimize decision models using incremental learning methods; Optimizing decision making using graph neural networks (GNNs).

Images