PSO-ASVR在三波长路面状态传感器定量标定中的应用

路面状态传感器是路面状态定性识别和定量测量的重要工具,其定量测量性能依赖于定量标定模型的准确性。为了解决路面状态传感器定量标定数据非线性和非均匀分布问题对定量测量的不利影响,提出基于PSO-ASVR(particle swarm optimization-adaptive support vector regression)的路面状态传感器定量标定模型。构建AP(adaptive preprocessing)流程进行标定数据最优化预处理,降低路面状态传感器非均匀分布问题影响下的标定数据处理误差。采用基于结构风险最小化的SVR(support vector regression)算法进行标定数据拟合,并利用PSO(particle swarm optimization)算法实现SVR中参数最优化,降低路面状态传感器标定数据非线性引入的数据拟合误差。不同路面状态条件下标定数据处理实验表明:新方法相比于传统方法在均方根误差RMSE上至少可减小63%,验证了其在提高定量标定模型精度上的有效性,实现了路面状态传感器定量标定误差的降低。

Application of PSO-ASVR in quantitative calibration of three-wavelength pavement state sensor

Pavement state sensor is an important tool for qualitative identification and quantitative measurement of pavement state, and its quantitative measurement performance depends on the accuracy of quantitative calibration model. In order to solve the problem of nonlinearity and nonuniform distribution of quantitative calibration data of pavement state sensor, a quantitative calibration model of pavement state sensor based on particle swarm optimization-adaptive support vector regression (PSO-ASVR) was proposed. Firstly, the adaptive preprocessing (AP) process was constructed to optimize the pre-processing of the calibration data to reduce the calibration data processing error under the influence of the nonuniform distribution of the pavement state sensor. Then, the support vector regression (SVR) algorithm based on structural risk minimization was used to fit the calibration data, and the particle swarm optimization (PSO) algorithm was used to realize the parameter optimization in the SVR to reduce the data fitting error introduced by the nonlinear calibration data of the pavement state sensor. Experiments on calibration data processing under different pavement states show that the root-mean-square error (RMSE) of the new method can be reduced by at least 63% compared with that of the traditional method, which verifies the effectiveness of the new method in improving the accuracy of the quantitative calibration model and realizes the reduction of the quantitative calibration error of the pavement state sensor.