基于积累周期分段编码的汽车车窗防夹学习算法

Anti-pinch Learning Algorithm for Vehicle Windows Based on Accumulated Period and Segment Code

汽车车窗防夹算法设计主要依据电机转动力矩计算出车窗受力情况,其推导计算复杂,软件设计人员需要有跨领域理论知识。以防夹霍尔硬件产生的大量真实数据为依据,利用数据分析和机器学习的技术设计分类模型,把防夹判断转为序列数据的分类。划分霍尔数据序列为正常、防夹、剧烈颠簸三种子分类,并提出了利用跟随周期平均积累差编码和斜率编码的方法分类霍尔数据序列流,这种方法简单、准确、实时地判断出防夹情况,并依此设计了简单的序列流分类防夹算法。

The algorithm designs about vehicle window anti-pinch depend on the window stress calculation according to the motor rotational torque,the software designers are required to learn interdisciplinary theoretical knowledge because of the computing complexity onthe algorithms.Based on a large number of real data produced by anti-pinch hall hardware,the classification model designed with the data analysis and machine learning technology translates the anti-pinch judgment to the sequence data classification.The hall data sequence is divided into three categories:normal,anti-pinch and severe turbulence,and the method of following period average accumulated difference coding and slope coding is proposed to classify the hall data stream.This method is simple,accurate and real-time.Based on above-mentioned,the simple anti-pinch algorithm is designed.