An intelligent early warning system and method for highway traffic safety based on artificial intelligence

Abstract

The present invention relates to the technical field of traffic safety, and specifically relates to an intelligent early warning system and method for highway traffic safety based on artificial intelligence. It consists of a data acquisition module, a data preprocessing module, an intelligent analysis module, an early warning decision-making module, and an interactive early warning interface. The data acquisition module collects multi-parameter data through three channels: sensors, monitoring systems, and weather stations. The data preprocessing module cleans, standardizes, normalizes, and supplements missing values for the collected multi-parameter data. The intelligent analysis module uses a feedforward neural network FFNN + self-attention mechanism to train the extracted features, constructs a multi-dimensional risk prediction model, automatically identifies changes in traffic patterns, predicts potential dangerous events, and the interactive response interface displays early warning information and multi-parameter data collected in real time. The present invention can comprehensively understand the road conditions and traffic risk situations, provide strong data support for the decision-making of the early warning system, and significantly improve the accuracy and intelligent level of early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic safety, and specifically relates to an intelligent early warning system and method for highway traffic safety based on artificial intelligence. Background Art

[0002] The existing traffic safety early warning systems based on traditional algorithms in the current market mainly rely on vehicle driving data and road conditions for safety risk prediction. These systems use simple threshold judgment methods, which can improve the safety of highway traffic to a certain extent, but there are obvious deficiencies. It is difficult to adapt to complex and changeable traffic environments, especially in dealing with atypical or sudden situations, the effect is limited. And due to the lack of support from deep learning and big data analysis, the early warning accuracy is limited, and the false alarm and missed alarm rates are relatively high. It is difficult to accurately identify and analyze potential traffic risk patterns. In addition, these systems do not fully consider the behavior habits and risk preferences of different drivers, and the early warning information lacks pertinence and practicality, making it difficult to meet the personalized service needs. Therefore, the present invention provides an intelligent early warning system and method for highway traffic safety based on artificial intelligence to overcome the deficiencies of the prior art and improve the safety of highway traffic. Summary of the Invention

[0003] The present invention provides an intelligent early warning system and method for highway traffic safety based on artificial intelligence to solve the technical problems that the existing intelligent early warning systems for traffic safety are difficult to adapt to complex and changeable traffic environments and have relatively low early warning accuracy.

[0004] To achieve the above object, the technical solutions adopted by the present invention are as follows: An intelligent early warning system for highway traffic safety based on artificial intelligence is composed of a data acquisition module, a data preprocessing module, an intelligent analysis module, an early warning decision module, and an interactive early warning interface. The data acquisition module collects multi-parameter data through three channels: sensors, monitoring systems, and meteorological stations. The data preprocessing module cleans, standardizes, normalizes, and supplements missing values for the collected multi-parameter data. After the feature extraction of the preprocessed multi-parameter data, the intelligent analysis module uses a feed-forward neural network FFNN + self-attention mechanism to train the extracted features, constructs a multi-dimensional risk prediction model, and captures the correlation between different positions in the multi-parameter data through the self-attention mechanism. By calculating the query, key, and value vectors of each element in the sequence, and assigning weights to each element through similarity calculation, and performing weighted summation on the value vectors according to the weights to obtain the context representation of each element, and inputting the context representation into the subsequent layers of the feed-forward neural network FFNN model, so as to automatically identify changes in traffic patterns and predict potential dangerous events. The interactive response interface displays early warning information and real-time collected multi-parameter data.

[0005] Further, the multi-parameter data includes historical accident records, real-time traffic flow, weather conditions, road conditions, driver behavior data, and random data.

[0006] Further, the multi-parameter data cleaning uses the standard deviation method to identify and process outliers.

[0007] Further, the improved Z-score normalization is used to standardize the multi-parameter data so that it can be compared and analyzed on the same scale.

[0008] Further, the standardized multi-parameter data is processed using min-max normalization to convert the multi-parameter data into a dimensionless value between 0 and 1.

[0009] Further, the intelligent analysis module constructs a driver profile based on the driver behavior data. The profile includes the driver's driving habits, reaction speed, and risk preference. Then, according to the driver profile, a personalized warning threshold is set for each driver.

[0010] Further, the warning decision module timely issues warning information according to the prediction result of the model, including the warning type, warning level, and recommended measures, and issues the warning information to the driver and the traffic management department through the roadside display screen, in-vehicle terminal, and mobile phone APP.

[0011] A method for an intelligent warning system for highway traffic safety based on artificial intelligence includes the following steps:

[0012] S01. Data collection: Collect multi-parameter data through three channels: sensors, monitoring systems, and weather stations;

[0013] S02. Data preprocessing: Clean, standardize, normalize, and supplement missing values for the multi-parameter data;

[0014] S03. Intelligent analysis: Use the feedforward neural network FFNN + self-attention mechanism to train the extracted features, construct a multi-dimensional risk prediction model, automatically identify changes in traffic patterns, and predict potential dangerous events;

[0015] S04. Warning decision: Based on the prediction analysis result, automatically generate a warning signal, and provide a diagnostic report and recommended measures according to the warning level;

[0016] S05. Interactive response: Display the warning information and the multi-parameter data collected in real time.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] (1)Through the integration and analysis of multi-parameter data, the present invention utilizes a feedforward neural network (FFNN) + self-attention mechanism to more comprehensively understand road conditions and traffic risk situations, providing strong data support for the decision-making of the early warning system, thereby achieving adaptive learning in complex traffic environments and significantly improving the accuracy and intelligence level of early warnings.

[0019] (2)By analyzing and mining a large amount of driver information, the present invention constructs an accurate driver profile and customizes personalized early warning thresholds and reminder methods for each driver according to the profile, thereby significantly improving the pertinence and effectiveness of early warning information and reducing the false alarm and missed alarm rates. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a structural diagram of an intelligent road traffic safety early warning system for artificial intelligence of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] The present invention will be described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that those of ordinary skill in the art can make several changes and improvements without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0022] As Figure 1 shown, an intelligent road traffic safety early warning system based on artificial intelligence consists of a data acquisition module, a data preprocessing module, an intelligent analysis module, an early warning decision-making module, and an interactive early warning interface. The data acquisition module collects multi-parameter data through three channels: sensors, monitoring systems, and meteorological stations. The data preprocessing module cleans, standardizes, normalizes, and supplements missing values for the collected multi-parameter data. After feature extraction of the preprocessed multi-parameter data, the intelligent analysis module uses a feedforward neural network (FFNN) + self-attention mechanism to train the extracted features, constructs a multi-dimensional risk prediction model, automatically identifies changes in traffic patterns, predicts potential dangerous events, and the interactive response interface displays early warning information and real-time collected multi-parameter data.

[0023] Specifically, the data acquisition module collects a large amount of traffic data including historical accident records, real-time traffic flow, and weather conditions. These multi-parameter data are obtained through various channels such as sensors, monitoring systems, and meteorological stations.

[0024] The main data includes historical accident records , real-time traffic flow , weather conditions , road conditions , driver behavior data , Random data (holidays, special events, traffic policies and regulations) .

[0025] Specifically, the data preprocessing module performs preprocessing operations such as cleaning, missing value filling, normalization, and standardization on the collected data to improve the quality and usability of the data.

[0026] Data cleaning: Use the standard deviation method to identify and process outliers. For each data point (historical accident records , real-time traffic flow , weather conditions , road conditions , driver behavior data , random data (holidays, special events, traffic policies and regulations) , ), where is the normalization weight value. Calculate its difference from the average value , and compare it with the standard deviation . If the difference exceeds the threshold (3 times the standard deviation), then consider this data point as an outlier and eliminate or replace it.

[0027] Data standardization: Standardize data from different sources and units so that they can be compared and analyzed on the same scale.

[0028] Improved Z-score standardization

[0029] where x is the multi-parameter data after data cleaning, μ is the mean of the data, and σ is the standard deviation of the data. After standardization, the mean of the data is 0 and the standard deviation is 1.

[0030] Data normalization: Convert the data into dimensionless values between 0 and 1 for easy processing and analysis by subsequent algorithms.

[0031] Minimum-maximum normalization formula

[0032] where x is the multi-parameter data after data cleaning, x min and x max are the minimum and maximum values in the data respectively. After normalization, the value range of the data is between 0 and 1.

[0033] Missing value filling: Use linear interpolation. For the missing value , the two non-missing values before and after it are and , and , then .

[0034] Extract features related to traffic safety from the preprocessed data, which will be used as the input of the deep learning model.

[0035] Specifically, the intelligent analysis module uses a deep learning algorithm (feed-forward neural network FFNN + self-attention mechanism) to train the extracted features and construct a multi-dimensional risk prediction model. The model can automatically identify changes in traffic patterns and predict potential dangerous events, such as vehicle collisions and road congestion.

[0036] Let the network input be M-dimensional: There are a total of K nodes in the hidden layer, and the output is L-dimensional. The length of the input-output sample pair is N. The activation function of the nodes in the hidden layer of the radial basis network is taken as the Gaussian basis function:

[0037]

[0038] The input acquisition data vector is mapped to the hidden layer, and the output of the hidden layer node is:

[0039]

[0040] In the formula: is the normalization constant of the hidden layer node, is the center vector of the Gaussian function of the hidden layer node,

[0041] The linear mapping of dimension conversion is realized from the hidden layer to the output layer of the RBF network, that is, the output of the output layer node k is:

[0042]

[0043] In the formula: is the adjustment weight from the hidden layer to the output layer; is the bias of the output layer node k. As a response to the corresponding input signal, it is output to the workspace as an important regulatory variable according to different backgrounds.

[0044] The above input x is the feature data extracted from the preprocessed data, and the output y is the potential feature danger status data. After the output feature data, an attention mechanism module is added to capture the global dependencies between the input data.

[0045] Specifically, the attention mechanism module calculates the query, key, and value vectors for each element in the sequence, assigns weights to each element through similarity calculation, performs weighted summation on the value vectors according to the weights to obtain the context representation of each element, and inputs the context representation into the subsequent layers of the FFNN model to further enhance the early warning ability of the model.

[0046] Specifically, through the query and the key attention weights , which is a scalar and is obtained by passing the attention scoring function that maps two vectors into a scalar , and then through the softmax operation:

[0047] .

[0048] Additive attention:

[0049]

[0050] Scaled dot-product attention:

[0051]

[0052] By introducing the attention mechanism, linear transformation and non-linear activation are performed on the weights and biases. The neurons in each layer receive the outputs from the neurons in the previous layer as inputs and calculate their own output values. These output values are then passed to the next layer until the output layer is reached. The self-attention mechanism can generate a context-sensitive representation for each input element based on the entire input sequence. This representation not only contains the information of the element itself but also the relationship information with other elements, thus enhancing the representation ability and generalization performance of the model. Then, the model is evaluated through indicators such as cross-validation, accuracy, and recall, and the model is optimized according to the evaluation results to improve the prediction accuracy.

[0053] The intelligent analysis module can also customize personalized early warning thresholds and reminder methods for each driver by analyzing information such as the driver's driving habits, reaction speed, and historical violation records. The specific implementation process is as follows:

[0054] Driver information collection: Collect driver behavior data;

[0055] (2) Driver portrait construction: The portrait includes the driver's driving habits, reaction speed, and risk preference.

[0056] (3)Set personalized warning thresholds for each driver according to the driver profile. For example, set a lower warning threshold for drivers with more aggressive driving habits and a higher warning threshold for drivers with more stable driving habits.

[0057] (4)Select a suitable reminder method according to the driver's preferences and actual situation. For example, for drivers who like visual reminders, display warning information on the in-vehicle display screen.

[0058] By analyzing and mining a large amount of driver information, construct an accurate driver profile, and customize personalized warning thresholds and reminder methods for each driver according to the profile. This personalized warning strategy can significantly improve the pertinence and effectiveness of warning information and reduce the false alarm and missed alarm rates.

[0059] The interactive warning interface is the user interface part of the system of the present invention. Transmit warning information to the driver through various methods such as voice prompts and visual warnings. Specifically, the interactive warning interface includes the following:

[0060] (1)Warning information display: Display warning information through various methods such as visual warnings (such as red warning lights, flashing icons, etc.) and voice prompts (such as voice announcements of warning information). The combination of visual warnings and voice prompts can convey warning information more intuitively and improve the driver's attention.

[0061] (2)User interaction: Allow the driver to confirm, ignore or take corresponding measures for the warning information. At the same time, provide a setting function to allow the driver to adjust the reminder method and content of the warning information according to their own preferences.

[0062] Through a reasonable interface layout and diverse information display methods, the readability and understandability of warning information can be significantly improved, and the driver's cognitive burden can be reduced.

[0063] A method for an intelligent highway traffic safety warning system based on artificial intelligence includes the following steps:

[0064] S01. Data collection: Collect multi-parameter data through three channels: sensors, monitoring systems, and weather stations; collect highway traffic data, including key information such as vehicle types, vehicle speeds, driving directions, traffic flows, and meteorological conditions.

[0065] S02. Data preprocessing: Clean, standardize, normalize, and supplement missing values for the multi-parameter data to ensure the quality and consistency of the data.

[0066] S03. Intelligent analysis: Use a feedforward neural network (FFNN) + self-attention mechanism to train the extracted features, construct a multi-dimensional risk prediction model, automatically identify changes in traffic patterns, and predict potential dangerous events;

[0067] Construct a feedforward neural network model for initially extracting the features of traffic data. Set the input layer, hidden layer, and output layer. The input layer receives the preprocessed traffic data, and the output layer outputs the warning results. Train the FFNN model through the backpropagation algorithm to enable it to accurately identify potential risks in traffic data. On the basis of the FFNN model, introduce a self-attention mechanism layer to capture the correlations between different positions in traffic data.

[0068] S04. Warning decision-making: Based on the prediction analysis results, automatically generate warning signals, and provide diagnostic reports and recommended measures according to the warning levels; Deploy the trained model to the highway traffic safety warning system. Real-time collect traffic data and input it into the model for prediction. According to the prediction results of the model, timely release warning information, including warning types, warning levels, recommended measures, etc. Release warning information to drivers and traffic management departments through various channels such as roadside display screens, in-vehicle terminals, and mobile phone APPs to ensure the timeliness and effectiveness of the information.

[0069] S05. Interactive response: Display warning information and multi-parameter data collected in real time.

[0070] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific implementation manners, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Without conflict, the embodiments of the present application and the features in the embodiments can be arbitrarily combined with each other.

Claims

1. An intelligent early warning system for highway traffic safety based on artificial intelligence, characterized in that: It consists of a data acquisition module, a data preprocessing module, an intelligent analysis module, an early warning decision module and an interactive early warning interface. The data acquisition module collects multi-parameter data through three channels: sensors, monitoring systems and weather stations. The data preprocessing module cleans, standardizes, normalizes and supplements missing values ​​for the collected multi-parameter data. After feature extraction, the preprocessed multi-parameter data is trained on the extracted features by the intelligent analysis module using the feedforward neural network FFNN+self-attention mechanism to build a multi-dimensional risk prediction model. The self-attention mechanism is used to capture the correlation between different positions in the multi-parameter data. The query, key and value vectors of each element in the sequence are calculated, and weights are assigned to each element through similarity calculation. The value vectors are weighted and summed according to the weights to obtain the context representation of each element. The context representation is input into the subsequent layers of the feedforward neural network FFNN model, so as to automatically identify changes in traffic patterns and predict potential dangerous events. The interactive response interface displays early warning information and multi-parameter data collected in real time. Specifically, by querying and key The attention weight , which is a scalar obtained by adding the attention score function Maps two vectors to a scalar , and then obtained through softmax operation: ; Additive Attention: ; Scaled Dot Product Attention: ; The multi-parameter data includes historical accident records, real-time traffic flow, weather conditions, road conditions, driver behavior data, and random data; The intelligent analysis module builds a driver profile based on driver behavior data, which includes the driver's driving habits, reaction speed, and risk preference, and then sets personalized warning thresholds for each driver based on the driver profile.

2. According to claim 1, the intelligent early warning system for highway traffic safety based on artificial intelligence is characterized in that: The multi-parameter data cleaning adopts the standard deviation method to identify and process outliers.

3. The highway traffic safety intelligent early warning system based on artificial intelligence according to claim 1 is characterized in that: The multi-parameter data were normalized using improved Z-score normalization to enable comparison and analysis on the same scale.

4. The highway traffic safety intelligent early warning system based on artificial intelligence according to claim 1 is characterized in that: The standardized multi-parameter data were converted into dimensionless values ​​between 0 and 1 using minimum-maximum normalization.

5. The highway traffic safety intelligent early warning system based on artificial intelligence according to claim 1 is characterized in that: The warning decision module releases warning information in a timely manner based on the prediction results of the model, including warning type, warning level and recommended measures, and releases warning information to drivers and traffic management departments through roadside display screens, on-board terminals and mobile phone APPs.

6. A method for a highway traffic safety intelligent early warning system based on artificial intelligence, the method is implemented based on the system as claimed in claim 1, comprising the following steps: S01. Data collection: Collect multi-parameter data through sensors, monitoring systems, and weather stations; S02. Data preprocessing: cleaning, standardization, normalization and missing value supplement of multi-parameter data; S03. Intelligent analysis: Use the feedforward neural network FFNN + self-attention mechanism to train the extracted features, build a multi-dimensional risk prediction model, automatically identify changes in traffic patterns, and predict potential dangerous events; S04. Early warning decision: Based on the prediction and analysis results, early warning signals are automatically generated, and diagnostic reports and recommended measures are provided according to the warning level; S05. Interactive response: Display warning information and multi-parameter data collected in real time.

Images