基于胎侧弯曲应变的农用轮胎状态估计方法研究

State Estimation Method of Agricultural Tire Based on Sidewall Bending Strain

摘要:

针对农用轮胎垂向载荷获取困难以及传统模型估算精度低等问题，提出了一种基于胎侧弯曲应变的农用轮胎状态估计方法，根据轮胎胎侧受垂向载荷后的弯曲应变规律，设计了一套集成高精度胎侧弯曲应变传感器、胎温胎压传感器的轮胎状态估计系统。搭建弯曲应变信息采集试验平台并开展多种典型工况测试试验，获取非道路轮胎滚动过程中不同胎压、速度以及负荷下胎侧的应变信号，建立了其滚动过程中轮辋胎侧弯曲应变、胎温、胎压数据集。对应变信号进行降噪、筛选与特征提取获取周期应变曲线与周期特征，构建了多特征加权载荷预测网络与轮速预测网络，对轮胎垂向载荷与速度进行精确且实时估计。结果显示，多特征加权载荷预测网络平均相对误差为1.26%，均方根误差为18.42 kg，相对于传统浅层BP 神经网络平均相对误差降低 27.17%，均方根误差降低 26.32%;速度预测网络平均相对误差为 1.16%，均方根误差为0.10 km/h，相较于BP神经网络平均绝对误差与均方根误差分别降低24.18%与16.67%。通过10折交叉验证试验，证明载荷预测与速度预测网络具有良好的泛化能力。研究表明，提出的基于胎侧弯曲应变的农用轮胎状态估计方法，实现了对农用轮胎垂向载荷与速度等状态信息准确估测。

Abstract:

The typical characteristics of agricultural tires include large load fluctuations, special pattern shapes, harsh working environments, and significant tire body vibration. These features make it difficult to accurately obtain the vertical load of the tire in practical operations. However, vertical load has a significant impact on the performance of agricultural machinery and is a key factor in evaluating and optimizing the efficiency and stability of agricultural machinery operations. A state estimation method for agricultural tires based on sidewall bending strain is proposed to address the difficulties in obtaining vertical loads and the low estimation accuracy of traditional models. A tire state estimation system that integrated high-precision sidewall bending strain sensors, tire temperature and pressure sensors was designed based on the bending strain law of the tire sidewall under vertical load. A bending strain information collection experimental platform was established and various typical working condition testing experiments were conducted through the platform. Strain signals of tire sidewall under different tire pressures, speeds, and loads during the rolling process of non-road tires were obtained. A dataset was established for the bending strain, tire temperature, and tire pressure of the wheel rim sidewall during its rolling process. After denoising, screening, and feature extraction, the periodic strain curve and periodic features were extracted from the strain signal. Furthermore, a multi feature weighted vertical load prediction network （MVL-Net） and speed prediction network based on deep neural network （SDNN） were constructed to accurately and realtime estimate the vertical load and speed of the tire. A dataset of strain signals and tire temperature and pressure was established, and a multi-feature weighted vertical load prediction network （MVLNet） and speed prediction network based on deep neural network （SDNN） were constructed. The prediction results showed that the mean relative error （MRE） of the MVL-Net was 1.26%, and the root mean square error （RMSE） was 18.42 kg, which was 27.17% and 26.32% lower than that of the BP network, respectively. The MRE of the SDNN was 1.16%, and the RMSE was 0.10 km/h, which was 24.18% and 16.67% lower than that of the BP network, respectively. Ten-fold cross validation experiments were conducted, and the results showed that the MVL-Net and SDNN had good generalization ability. Research result showed that the proposed state estimation method of agricultural intelligent tire based on sidewall bending strain can achieve accurate prediction of state information such as vertical load and rotational speed of agricultural tires.