Road Surface Meteorological Condition Detection:Key Knowledge And Technology

The road surface meteorological condition recognition technology is utilized to help the highway and transportation departments detect and locate adverse road conditions such as ice and snow in time,which is an important decision-making basis for road maintenance and management.Due to the development of complex road surface condition recognition technology,the work of melting ice,removing snow,and closing road sections has gradually entered a new stage of refinement and intelligence.In order to meet the information needs of the next generation intelligent highway network,within the framework of contact detection method,this dissertation studies the road surface condition recognition technology based on resonance principle,spectral absorption principle and relaxation polarization principle.The corresponding experiments have been conducted to verify the performance of the sensing technologies.Based on the above results,the research on complex road surface condition recognition method for multi-source data fusion is carried out.Firstly,this dissertation studies the road surface condition recognition technology based on the statistical characteristics of resonance spectrum,and explores the scientific issue of how to mine and analyze the relationship between the statistical characteristics of resonance spectrum and the thickness of condensation.In the research process,the defects of the traditional resonant sensor with poor stability due to the attenuation of the first-order resonance peak were found.Through the research on the small deflection bending problem,the equivalent bending stiffness model and the finite element simulation results,the method of using multi-order frequency response for road surface condition recognition is proposed.On this basis,the features are selected using correlation coefficient,and the regression model is built to calculate the thickness of the load.The recognition and measurement of ice,water and air are achieved,and the upper limit of the sensor thickness measurement range is greatly increased from less than 2mm to about 10 mm.The performance of the sensor was evaluated by laboratory experiment and filed test.The results show that the measurement accuracy and stability of the sensor have been significantly improved.Then,this dissertation studies the road surface condition recognition technology based on multi-spectral reflection distribution,and explores the scientific issue of how to construct a contact multi-spectral reflection distribution characteristic model for road surface condition recognition.Conventional non-contact optical sensors must have a precise optical structure and large transmission power because of the severe infrared absorption and the long distance between lens and road surface,and thus the prices are extremely high.This dissertation explores the feasibility of contact optical sensors.The power-oriented optical propagation model is established to analyze the performances of the typical leaky and reflective sensor architecture under wear conditions,and the isolated structure of the contact optical sensor is chosen.Combined with the absorption-based multispectral measurement principle,an optical sensor capable of identifying seven kinds of road surface conditions is implemented,and the sensor has strong anti-interference and anti-wear ability through digital average filtering technology and photoelectric signal ratio features.Furthermore,this dissertation studies the road surface condition recognition technology based on complex impedance spectrum characteristics,and explores the scientific issue of how to utilize the characteristics of complex impedance spectrum for road surface condition detection.In view of the fact that the traditional measurement method may have the defects of false alarm and missed alarm when the film is thin,this dissertation takes the parametric simulation as the design basis and effectively improves the measurement response of the complex impedance sensor under the given structural size constraints.The sensor’s ability to recognize small amounts of condensation is improved.At the same time,according to the principle of relaxation polarization,by calculating the equivalent capacitance and conductance curves,and constructing the feature vector with the relative dispersion index of the relevant data,the correct rate of the three categories of 97.5% is achieved in the measurement environment where the temperature,thickness and salinity fluctuate greatly.Finally,based on the above research results and multi-source feature fusion,this dissertation studies the complex road condition recognition method.And the scientific issue of how to achieve complex road surface condition recognition by establishing multi-source feature fusion mechanism is explored on the basis of three road condition recognition technologies and the physical characteristics of road surface condensation.The existing complex road condition sensors rely on knowledge rules to make decision layer data fusion,which weakens the correlation between features to some extent,and the simple feature layer fusion method shows the problem of insufficient stability.In this dissertation,through the multi-level data fusion method,the model has the stability of the decision-making layer fusion model and the correct rate of the feature layer fusion.With the integrated road surface condition sensor being the experiment platform,it has successfully achieved the efficient and reliable recognition of the eight road conditions.The method proposed in this dissertation achieved a correct rate of 97.9% on the laboratory dataset,and demonstrated the road surface condition recognition and thickness measurement capability beyond the non-contact detector in the field test,providing key technical support for constructing the next generation of complex road condition monitoring systems.