A method for intelligent recognition and extraction of features of bridge vehicle-mounted strain influence lines

Abstract

The invention discloses a method for intelligently identifying and extracting features of vehicle-mounted strain influence lines of a bridge, comprising the following steps: decomposing strain test signals by using multi-layer wavelet transform and extracting high-frequency (caused by vehicles) components therein; Intercept and mark the stable segment (when there is no car) and non-stationary segment (when there is a car) in the vehicle-induced strain signal to form a training set and a test set; design a long-term short-term memory classification network, and use the training set data to train the neural network Learning, and use the test set data to test the classification prediction accuracy of the trained neural network; use the sliding window to intercept the vehicle-induced strain signal obtained by online decomposition in real time, and use the trained classification network to classify, predict and mark the real-time intercepted signal , and finally extract the eigenvalues ​​of the identified non-stationary signal (ie, the vehicle strain influence line). Compared with the prior art, the method of the present invention has strict logic, clear physical meaning and rules to follow in implementation.

Description

technical field [0001] The invention belongs to the field of performance monitoring, detection, early warning and evaluation of existing bridge structures, and is a method for intelligently identifying and extracting characteristics of bridge vehicle-mounted strain influence lines, in particular, relates to a bridge structure vehicle-mounted strain influence line based on artificial neural network Intelligent recognition and feature extraction methods. Background technique [0002] Bridges are an important part of China's transportation network, and vehicle loads are the main live loads on bridge structures during their service. The vehicle-mounted strain influence line is one of the main indicators to judge the current state of the bridge structure. With the development of testing technology, it is not difficult to collect a large number of bridge strain responses. However, with the accumulation of data, extracting vehicle-borne strain influence lines (that is, non-statio...

AI Technical Summary

Problems solved by technology

The commonly used methods are as follows: (1) Manual judgment and extraction based on experts: This method is based on the mechanical knowledge and management experience of bridge designers, managers or related scholars, and artificially carries out the on-board influence line of the strain signal. Discrimination and extraction, this method requires practitioners to have certain mechanical knowledge and management experience, and is only suitable for manual operations when the amount of data is extremely small, so it cannot be popularized and applied in the era of monitoring and detection of big data; (2) based on Automatic identification and extraction of empirical thresholds: This method is based on the expert’s usual range of extreme values ​​(such as maximum value, amplitude, etc.) of the vehicle-induced strain influence line. An empirical threshold is given when the extreme value of the vehicle-induced strain signal is greater than When it is equal to the threshold, it is determined that the segment of the signal is the signal of the vehicle strain influence line. However, this method relies too much on the selection of the threshold. If the threshold is large, it is easy to ignore the small vehicle strain influence line. Mastered and applied by engineering practitioners

Images