Search

CN-122022165-A - Automatic evaluation method for non-motor vehicle road environment quality based on artificial intelligence

CN122022165ACN 122022165 ACN122022165 ACN 122022165ACN-122022165-A

Abstract

The invention discloses an artificial intelligence-based non-motor vehicle road environment quality automatic evaluation method which comprises the steps of acquiring parameters such as road domain geometric structure, road surface physical characteristics, traffic flow state, meteorological environment, peripheral facility distribution and the like through multi-source sensing equipment, and completing data noise reduction, outlier identification and feature extraction through a road domain data intelligent analysis processing platform to obtain a road domain feature vector and an environment factor vector. Generating a road suitability evaluation vector through a non-motor vehicle road characteristic adaptation model, excavating factor coupling relations through an environmental factor association analysis modeling model, forming an environmental association influence matrix, fusing characteristics through a non-motor vehicle passing efficiency prediction model, outputting a passing efficiency evaluation result, finally combining a preset index system, and generating a road environment quality final evaluation result through weighted comprehensive calculation and grading, so that comprehensive, accurate and automatic evaluation on the non-motor vehicle road environment quality is realized.

Inventors

  • ZHOU SIYU
  • ZHU MENG
  • XIANG TINGTING

Assignees

  • 安徽农业大学

Dates

Publication Date
20260512
Application Date
20260130

Claims (10)

  1. 1. The non-motor vehicle road environment quality automatic evaluation method based on artificial intelligence is characterized by comprising the following steps of S1, collecting geometric structure parameters, road surface physical characteristic parameters, traffic flow state parameters, weather environment parameters and surrounding facility distribution parameters of a non-motor vehicle road domain through multi-source sensing equipment to form an original data set, S2, inputting the original data set into a road domain data intelligent analysis processing platform to conduct data noise reduction, outlier recognition and feature extraction to obtain road domain feature vectors and environment factor vectors, S3, inputting the road domain feature vectors into a non-motor vehicle road domain feature adaptation model, generating road domain suitability evaluation vectors through feature mapping and weight distribution, S4, inputting the environment factor vectors into an environment factor association analysis modeling model, generating an environment association influence matrix through factor correlation calculation and coupling relation mining, S5, inputting the road domain suitability evaluation vectors and the environment association influence matrix into a non-motor vehicle efficiency prediction model, outputting a traffic efficiency evaluation result through multi-level feature fusion and nonlinear mapping, S6, and finally generating a traffic efficiency evaluation result through the road quality index classification and the road quality evaluation system through automatic preset and the road quality index classification.
  2. 2. The automated evaluation method for the environmental quality of the non-motor vehicle road based on the artificial intelligence according to claim 1, wherein the expression of the non-motor vehicle road domain feature adaptation model is: , wherein, The vector is evaluated for road domain suitability, The function is activated for Sigmoid, As the characteristic weight coefficient of the object, In the case of a convolution operation, Is a vector of the physical characteristic parameters of the road surface, For the purpose of the pooling operation, As a vector of the parameters of the geometric structure, Is a bias term.
  3. 3. The automated evaluation method for the environmental quality of the non-motor vehicle road based on the artificial intelligence according to claim 1, wherein the expression of the environmental factor correlation analysis modeling model is: , wherein, For the environmental correlation impact matrix, For the number of categories of environmental factors, For the covariance calculation the sum of the values of the covariance, Respectively the first The vector of the weather-like environmental parameters, For the calculation of the variance of the values, Is a factor coupling coefficient.
  4. 4. The automated non-motor vehicle road environment quality evaluation method based on artificial intelligence according to claim 1, wherein the expression of the non-motor vehicle passing efficiency prediction model is: , wherein, In order to evaluate the result of the traffic performance, For the hyperbolic tangent activation function, In order for the coefficient of fusion to be a function of, For the characteristic splicing operation, the method comprises the steps of, For the weight matrix, MLP () is a multi-layer perceptron operation, As a vector of traffic flow state parameters, Is a bias term.
  5. 5. The automated evaluation method for the environmental quality of the non-motor vehicle road based on artificial intelligence according to claim 1, wherein the feature extraction model expression of the intelligent analysis processing platform for road domain data is as follows: , wherein, In order to extract the feature vector after the extraction, For a linear rectification activation function, For the purpose of extracting the coefficients of the features, For the matrix of the set of raw data, Is used for the analysis and operation of the principal component, For the linear discriminant analysis operation, Is a bias term.
  6. 6. The automated evaluation method for the environmental quality of the non-motor vehicle road based on the artificial intelligence according to claim 1, wherein the comprehensive calculation model expression of the automated evaluation for the environmental quality of the non-motor vehicle road is: , wherein, In order to obtain the final evaluation result, In order to evaluate the number of indicators, Is the weight coefficient of the kth index, As the traffic efficiency score result corresponding to the kth index, For the maximum value of the traffic performance evaluation result, The index correction coefficient.
  7. 7. The automated evaluation method for the road environment quality of the non-motor vehicle based on artificial intelligence is characterized by comprising the following steps of S31, constructing a multistage feature grouping rule based on physical meaning and data distribution characteristics of parameters in a road feature vector, dividing geometric structure parameters and road physical characteristic parameters into basic feature groups, calibration feature groups and auxiliary feature groups, S32, performing dimension unification and format conversion on the features of the groups through an input layer of a non-motor vehicle road feature adaptation model to form a standardized feature matrix, S33, performing convolution kernel sliding operation and feature mapping processing on the standardized feature matrix by a model hiding layer, distributing feature weights by combining with a preset road suitability judgment rule, and S34, performing normalization processing on the mapped feature vectors through an output layer to generate a road suitability evaluation vector with fixed dimension and unified numerical value interval.
  8. 8. The automated evaluation method for the environmental quality of the non-motorized vehicle road based on artificial intelligence according to claim 1 is characterized in that S4 comprises the following steps of S41, decomposing an environmental factor vector into a meteorological environment subvector, a peripheral facility subvector and a traffic interference subvector, determining specific parameter types and data dimensions included by the subvectors, S42, calculating pearson correlation coefficients among internal parameters of the subvectors through an environmental factor correlation analysis modeling model, primarily screening strong correlation parameter pairs, S43, constructing a factor correlation network based on the strong correlation parameter pairs, generating a correlation path matrix through indirect coupling relation among network topology analysis mining parameters, and S44, weighting and fusing the direct correlation coefficients and the indirect coupling intensity to construct an environmental correlation influence matrix with the dimensions consistent with the number of the environmental factors.
  9. 9. The automated evaluation method for the non-motor vehicle road environment quality based on artificial intelligence according to claim 1 is characterized in that the step S5 comprises the following steps of performing dimension expansion processing on a road suitability evaluation vector to enable the road suitability evaluation vector to be consistent with the column dimension of an environment association influence matrix to form an adaptation input matrix, the step S52 of performing element-by-element product operation and feature superposition on the adaptation input matrix and the environment association influence matrix through a feature fusion layer of a non-motor vehicle passing efficiency prediction model, the step S53 of performing nonlinear conversion on the fused features through a multi-layer perceptron structure of the model and extracting high-order features through activation operation of neurons of a hidden layer, and the step S54 of obtaining passing efficiency evaluation results reflecting passing efficiency, safety level and comfort level through linear conversion and range mapping of an output layer.
  10. 10. The automated evaluation method for the road environment quality of the non-motorized vehicle based on artificial intelligence according to any one of claims 1 to 9 is realized by different units and comprises a multi-source data collaborative acquisition unit, a road domain data intelligent analysis processing unit and a road side equipment intelligent analysis processing unit, wherein the multi-source data collaborative acquisition unit is used for acquiring road domain geometry, road surface characteristics, traffic flow, weather and peripheral facility related parameters through sensing equipment, mobile terminals and road side equipment; the road domain data intelligent analysis processing unit receives output data of the multi-source data collaborative acquisition unit, executes noise reduction, outlier identification and feature extraction operations, respectively transmits processed road domain feature vectors and environment factor vectors to the road domain feature adaptation unit and the environment factor association analysis unit, loads a non-motor vehicle road domain feature adaptation model, maps and weight distributes the received road domain feature vectors to generate road domain suitability assessment vectors and transmits the road domain suitability assessment vectors to the traffic efficiency prediction unit, loads an environment factor association analysis modeling model, digs correlation and coupling relations among the environment factors, generates an environment association influence matrix and transmits the environment association influence matrix to the traffic efficiency prediction unit, loads the non-motor vehicle traffic efficiency prediction model, fuses the road domain suitability assessment vectors and the environment association influence matrix, outputs a traffic efficiency assessment result to the quality automation assessment unit, receives the output result of the traffic efficiency prediction unit, and executes weighted comprehensive calculation and grade division in combination with a preset index system to generate and output a road quality final assessment result.

Description

Automatic evaluation method for non-motor vehicle road environment quality based on artificial intelligence Technical Field The invention relates to the technical field of road environment quality evaluation, in particular to an artificial intelligence-based non-motor vehicle road environment quality automatic evaluation method. Background With the acceleration of the urban process, the non-motor vehicle is used as a green low-carbon travel mode, and the road environment quality is directly related to travel safety and experience, so that the non-motor vehicle becomes an important focus of urban traffic control. In the current non-motor vehicle road construction, the multidimensional parameters such as road domain geometry, road surface characteristics, traffic flow state, meteorological environment, peripheral facility distribution and the like are mutually influenced, the traditional manual evaluation mode relies on subjective judgment, and multi-factor collaborative analysis and dynamic evaluation are difficult to realize. Meanwhile, the adaptability requirements of the traffic demands of the non-motor vehicles and the road environment are continuously improved, an efficient and accurate automatic evaluation system is required to be constructed, multisource data are integrated and related through intelligent algorithm mining parameters, scientific basis is provided for road planning optimization, maintenance upgrading and environment improvement, and intelligent development of a non-motor vehicle traffic system is promoted. The prior art has two obvious defects in the non-motor vehicle road environment quality evaluation, namely, the evaluation method lacks the deep fusion capability of multi-source heterogeneous data, fails to effectively correlate the coupling relation between road domain features and environmental factors, obtains an evaluation result through single dimension parameters or simple weighted calculation, and causes the evaluation result to hardly reflect the comprehensive quality of the road environment comprehensively, and the evaluation process lacks a standardized intelligent evaluation model and an efficient processing platform support, relies on manual intervention, so that the full-flow automation of data acquisition, analysis, modeling and evaluation is difficult to realize, the evaluation efficiency is reduced, the consistency and objectivity of the evaluation result are influenced by human factors, and the requirements of large-scale and normalized road environment quality monitoring and evaluation cannot be met. Disclosure of Invention In order to overcome the defects and shortcomings in the prior art, the invention provides an artificial intelligence-based non-motor vehicle road environment quality automatic evaluation method. The technical scheme includes that the method comprises the following steps of S1, collecting geometric structure parameters, road surface physical characteristic parameters, traffic flow state parameters, weather environment parameters and surrounding facility distribution parameters of a non-motor vehicle road domain through multi-source sensing equipment to form an original data set, S2, inputting the original data set into a road domain data intelligent analysis processing platform to conduct data noise reduction, outlier identification and feature extraction to obtain a road domain feature vector and an environment factor vector, S3, inputting the road domain feature vector into a non-motor vehicle road domain feature adaptation model, generating a road domain suitability evaluation vector through feature mapping and weight distribution, S4, inputting the environment factor vector into an environment factor association analysis modeling model, generating an environment association influence matrix through factor-to-road correlation calculation and coupling relation mining, S5, inputting the road domain suitability evaluation vector and the environment association influence matrix into a non-motor vehicle efficiency prediction model, outputting a passing efficiency result through multi-level feature fusion and non-linear mapping, S6, and generating a final road grade evaluation result through automatic combination of the non-motor vehicle passing efficiency and preset road quality evaluation system and a road grade comprehensive evaluation rule. Further, the expression of the non-motor vehicle road domain feature adaptation model is: , wherein, The vector is evaluated for road domain suitability,The function is activated for Sigmoid,As the characteristic weight coefficient of the object,In the case of a convolution operation,Is a vector of the physical characteristic parameters of the road surface,For the purpose of the pooling operation,As a vector of the parameters of the geometric structure,Is a bias term. Further, the expression of the environmental factor correlation analysis modeling model is: , wherein, For the environmental correlation impact