基于非线性自回归模型的汽车悬架特征提取及状态辨识方法研究

项目名称： 基于非线性自回归模型的汽车悬架特征提取及状态辨识方法研究

项目编号： No.51305194

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 机械、仪表工业

项目作者： 陈茹雯

作者单位： 南京工程学院

项目金额： 24万元

中文摘要： 汽车悬架作为非线性复杂系统，具有不确定性输入，更无法确知其内部相互作用方式，因而经简化处理的汽车振动系统在一定程度上不能完全反映悬架状态，因此采用非线性自回归模型对悬架系统建模，提取状态特征，实现系统状态辨识及发展趋势预测。 首先从时、频域理论方面研究该模型的逼近机理，并辅以数据实验，证明模型对汽车悬架等非线性系统的适用性；然后针对模型的结构特点，改进结构辨识和参数估计方法，提高建模速度、模型辨识和预测精度，形成一套适合于工程实际的非线性自回归模型建模理论体系；最后以汽车悬架振动输出信号为对象，建立非线性自回归模型，通过模型特征量，辨识、比较、诊断系统各种完备或故障状态，区分系统在不同运行工况时的品质信息，并以满足工程应用的精度和实时性要求为目标，优化建模理论，为车辆机械技术状况智能监测提供一种稳定可靠的方法。

中文关键词： 非线性自回归模型；非线性振动；悬架；隔振性能；辨识

英文摘要： There are limitations to indicate the automobile suspension state by traditional modeling method to some extent because of the uncertainty of the suspension's input and internal interaction mode. Hence, the general expression for nonlinear autoregressive model (GNAR model) will be adopted for suspension system feature extraction, state identification and development trend prediction. First, the model approximation characteristics are studied from the time and frequency domain both in theory and data experiments to prove its applicability for the linear and nonlinear systems. Then according to the structural characteristics, the model structure identification and parameter estimation methods are developed to improve modeling speed and identification and prediction accuracy and a set of GNAR model theory is formed which is suitable for engineering applications. Finally, GNAR model is applied to analyze the automobile suspension vibration output signals. The system qualities in different operating conditions are distinguished by identifying and comparing the model feature vectors. The modeling methods are optimized to meet the engineering requirements for precision and real-time target and a reliable method is developed to monitor the technical conditions of automobile mechanical assemblies.

英文关键词： GNAR model；Nonlinear vibration；Suspension；Anti-vibration performance；Identification