Reconstruction of road defects and road roughness classification using vehicle responses with artificial neural networks simulation

The road damage assessment methodology in this paper utilizes an artificial neural network that reconstructs road surface profiles from measured vehicle accelerations. The paper numerically demonstrates the capabilities of such a methodology in the presence of noise, changing vehicle mass, changing vehicle speeds and road defects. In order to avoid crowding out understanding of the methodology, a simple linear pitch-plane model is employed. Initially, road profiles from known roughness classes were applied to a physical model to calculate vehicle responses. The calculated responses and road profiles were used to train an artificial neural network. In this way, the network renders corresponding road profiles on the availability of fresh data on model responses. The results show that the road profiles and associated defects can be reconstructed to within a 20% error at a minimum correlation value of 94%.     

Tyre rubber friction on a rough road

One of the most challenging aspects of vehicle dynamics is accurate modelling of the tyre-road interface. Forces between the tyre and road need to be accurately represented in simulation. This is challenging over rough roads since the friction changes along the road due to large surface asperities.The Heinrich/Klüppel friction coefficient estimation model has been implemented on smooth roads in the past. However, this study investigates the applicability of using this model over a rough but hard terrain, such as Belgian paving or cobblestones. The model is based on physical properties that can be determined mathematically or experimentally. The study includes detailed terrain topography and the difference between the top and bottom topography is used to determine the radially averaged PSD. Emphasis is placed on finding and implementing the flash temperature in a practical manner that could also be used in further studies.An experimental setup is built to validate the model. The experimental friction coefficient is compared to the friction coefficient calculated using the Heinrich/Klüppel model. The relative percentage error difference between experimental and friction model results is found to be less than 10% on a smooth concrete road and 20% on a rough road (concrete Belgian paving).     

Autocorrelation model of road roughness

Road surface roughness is the excitation source for the dynamic response of a moving vehicle system. Driving comfort is indicated by either the driver absorbed power level or the vehicle vertical acceleration level. An autocorrelation function model for road roughness is proposed to specify the road surface random characteristics. Subsequently, the power spectral densities (PSDs) for both road roughness and vehicle response, the driver-absorbed power level, are formulated. A road quality index (RQI) in accordance with such energy considerations is defined to catalog the road grade. The laboratory test results show the applicability of the RQI method for road classification using the ISO criteria as a comparison check.     

Country road and field surface profiles acquisition,modelling and synthetic realisation for evaluating fatigue life of agricultural machinery

Fatigue life of agricultural machinery is strongly affected by the surfaces that these machines operate on. In the present paper a sensor-frame was developed to acquire road and field profiles in absolute geo-referenced coordinates. The sensor-frame was validated by measuring discrete trapezoidal bumps with known dimensions resulting in a root mean squared (RMS) error of 6–8 mm. Profiles were acquired from a country road and from a mowed grass field. Using the quarter-car vehicle model, the movement of an agricultural vehicle was simulated for various speeds. The resulted vertical loads were rainflow-counted and the accumulated fatigue pseudo damage was calculated using Palmgren–Miner linear rule. Based on the derived Power Spectrum Density (PSD), the profiles were classified according to ISO 8608 standard. Two methodologies were followed to model and create a number of synthetic realisations for each profile: Direct Spectrum Estimate (DSE) and ISO based modelling. Simulating the produced synthetic profiles with the quarter-car vehicle model, the corresponding pseudo damage was calculated. The accumulated damages from the DSE models were closer to the corresponding ones from the measured profiles. ISO based models could not model the profile irregularities, which proved to contribute the largest part of the accumulated fatigue damage.     

Suspension settings for optimal ride comfort of off-road vehicles travelling on roads with different roughness and speeds

This paper reports on an investigation to determine the spring and damper settings that will ensure optimal ride comfort of an off-road vehicle, on different road profiles and at different speeds. These settings are required for the design of a four stage semi-active hydro-pneumatic spring damper suspension system (4S₄). Spring and damper settings in the 4S₄ can be set either to the ride mode or the handling mode and therefore a compromise ride-handling suspension is avoided. The extent to which the ride comfort optimal suspension settings vary for roads of different roughness and varying speeds and the levels of ride comfort that can be achieved, are addressed. The issues of the best objective function to be used when optimising and if a single road profile and speed can be used as representative conditions for ride comfort optimisation of semi-active suspensions, are dealt with. Optimisation is performed with the Dynamic-Q algorithm on a Land Rover Defender 110 modelled in MSC.ADAMS software for speeds ranging from 10 to 50 km/h. It is found that optimising for a combined driver plus rear passenger seat weighted root mean square vertical acceleration rather than using driver or passenger values only, returns the best results. Results indicate that optimisation of suspension settings using one road and speed will improve ride comfort on the same road at different speeds. These settings will also improve ride comfort for other roads at the optimisation speed and other speeds, although not as much as when optimisation has been done for the particular road. For improved ride comfort damping generally has to be lower than the standard (compromised) setting, the rear spring as soft as possible and the front spring ranging from as soft as possible to stiffer depending on road and speed conditions. Ride comfort is most sensitive to a change in rear spring stiffness.     

Parameterisation,validation and implementation of an all-terrain SUV FTire tyre model

This paper focuses on the parameterisation, validation and implementation of an FTire model of a Michelin LTX A/T² 235/85R16 tyre. This tyre is designed for both on- and off-road use and is commonly used on all wheel drive SUVs. Quasi-static laboratory and dynamic field tests were conducted to acquire parameterisation and validation test data for the FTire model. Quasi-static parameterisation tests include acquiring vertical tyre stiffness over a flat plate and cleats, tyre footprint sizes and shapes, longitudinal, lateral and torsional tyre stiffness for various tyre normal loads, as well as vibrational tyre responses. Dynamic parameterisation tests include dynamic cleat test data. An Adams model of the tyre testing equipment is implemented to simulate the FTire model and validate it against dynamic validation test results. Finally, the model is implemented on a fully nonlinear multi-body dynamics model of a Land Rover Defender. It is found that the FTire model is able to predict the lateral tyre behaviour well on a smooth road surface. The vertical and longitudinal tyre behaviour on a smooth road surface and on a rough surface are predicted accurately.     

Haul road defect identification using measured truck response

Current management techniques for the maintenance of mine haul roads, such as ad hoc blading, scheduled blading and even maintenance management systems, have shortcomings in complex mining environments. This paper investigates the possibility of using the response of haul trucks to aid the management of haul road maintenance. The question arises as to whether truck response data can be used to recognize road defects at specific locations, in terms of type and size. This is important since different defect types require different road maintenance strategies. A modeling methodology based on dynamic equilibrium of the unsprung mass of a haul truck is proposed and investigated.     

A comparative study of destructive effects resulting from road profile acting on off-road towed vehicles

The more extreme conditions the vehicle is exposed to, the sooner it wears out and deteriorates. In order to determine the forces affecting the lifespan of vehicles we need to know the environmental conditions eliciting these forces.This research aims at elaborating and testing a method which can help to conduct a comparative analysis of forces acting on towed vehicles used in different terrain conditions. Excitation forces acting on a vehicle being towed across terrain cause vibrations which lead to wear and structural deterioration. The rate of deterioration depends on the activating forces resulting from the road profile geometry and the dynamic properties of the vehicle. A knowledge of the relationship between the towed vehicle and the terrain profile will enable the design of an artificial road profile for fatigue testing with which similar stresses arise as during normal use. With the developed comparative method, a connection can be established between stochastic road profiles and road profiles containing artificially built obstacles.     

Spectrum analysis of road roughness for earthmoving machinery

The power spectral density functions of roughness of road surfaces for earthmoving machines such as tractors, scrapers and dumpers were determined by use of an accelerometer mounted on a towed fifth wheel of which the frequency-response function had previously been known. From obtained data, a range of power spectral density curves of the road roughness for these earthmoving machines was derived and expressed mathematically.     

Forward kinematics analysis of the 6-3 SPM by using neural networks

A feed-forward neural network has been trained using backpropagation algorithm to solve the forward kinematics problem of the 6-3 Stewart Platform Mechanism (SPM). The forward kinematics problem of the SPM does not have a unique solution since it involves solving a polynomial of order 16. Purely translational, purely rotational and general spatial data sets have been used in training and testing, and then an optimization procedure has been applied to fine-tune the solution. The method yields results fast and accurate enough such that it can be used instead of a gyro and a position sensor for real time control of the mechanism.     

Indoor testing of road vehicle suspensions

The paper presents a method for the indoor testing of road vehicle suspension systems. A car suspension is positioned on a rotating drum located in the Laboratory for the Safety of Transport at the Politecnico di Milano and it is excited as the wheel passes over a cleat fixed on the drum. The wheel accelerations, displacements and the forces/moments acting at the suspension-chassis joints are measured in the frequency range 0–120 Hz. Five special six-axis load cells have been designed and used. Transient wheel motions have been recorded. The influence of the running conditions on the relevant performance indexes related to the vibration behavior of the tire/suspension system has been assessed.     

Rough terrain profiling using digital image correlation

Road profiling is an important aspect of vehicle dynamics simulations especially over rough terrains. The accurate measurement of rough terrains allows for more accurate multi body simulations. Three dimensional road profiles are usually performed by utilising a line scan sensor which measures several points lateral to the road. The sensors range from simple road following wheels to LiDAR sensors. The obtained line scans are longitudinally stitched together using the orientation and position of the sensor to obtain a full three dimensional road profile. The sensor’s position and orientation therefore needs to be accurately determined in order to combine the line scans to create an accurate representation of the terrain. The sensor’s position and orientation is normally measured using an expensive inertial measurement unit or Inertial Navigation System (INS) with high sensitivity, low noise and low drift. This paper proposes a road profiling technique which utilises stereography, based on two inexpensive digital cameras, to obtain three-dimensional measurements of the road. The system negates the use of an expensive INS system to determine orientation and position. The data sets also require subsampling which can be computationally expensive. A simple subsampling routine is presented which takes advantage of the structure of the data sets to significantly speed up the process.     

模糊—神经网络控制算法及其在离心力—振动复合环境试验系统中的应用

针对离心－振动复合环境试验系统所存在的耦合性、非线性和不确定性提出了一种模糊－神经网络控制算法,利用被控对象输入输出信息离线、在线相结合学习系统的动态特性,对时变、非线性系统进行跟踪控制,并研究了该算法在系统中的实现方法。实现表明了控制系统具有良好的跟踪能力。该算法也适用于快速变化这类系统的实时控制。     

Tractor cabin’s passive suspension parameters optimization via experimental and numerical methods

Reduction of transmitted vibrations of tractor cabin which is caused by road roughness is the major objective of this study; consequently operator health can be achieved. This objective is carried out via experimental measurements and finite element modeling. For this purpose the vertical acceleration of the cabin as well as the rear axle of the tractor is measured in different road conditions and forward speeds. However, it should be mentioned that tests were carried out according to the ISO 2631-1985 but no measurements were done on the driver’s seat. Then the finite element model of the cabin’s tractor is developed and the dynamic response of the cabin interior (with the measured axle acceleration as input dynamic force) is obtained. At the third step the suspension parameters are calculated by comparing the accelerations obtained from the model and measurements. Finally the suspension parameters are optimized according to ISO 2631-1985 via iterative method.     

CONSTITUTIVE MODEL OF ARTIFICIAL FROZEN SOIL BASED ON CASCADE-CORRELATION NEURAL NETWORK

为解决BP (back propagation) 神经网络收敛速度慢,网络结构需事先定义等缺点,采用了级连相关神经网络模型来建立人工冻土应力和应变之间的关系. 基于该模型推导了冻土的一致刚度矩阵形式,利用人工冻土三轴试验数据对神经网络模型进行训练,并用其替换有限元计算中的传统本构模型,将计算结果与性质及含水率相同的冻土的试验结果进行了对比,发现该神经网络本构模型很好地反应了材料的非线性,能够改善数值计算结果,与实测结果吻合地很好,比具有相同隐含层神经元个数的BP 模型更接近实测结果.     

基于级连相关神经网络的人工冻土本构模型1）

为解决BP (back propagation)神经网络收敛速度慢,网络结构需事先定义等缺点,采用了级连相关神经网络模型来建立人工冻土应力和应变之间的关系。基于该模型推导了冻土的一致刚度矩阵形式,利用人工冻土三轴试验数据对神经网络模型进行训练,并用其替换有限元计算中的传统本构模型,将计算结果与性质及含水率相同的冻土的试验结果进行了对比,发现该神经网络本构模型很好地反应了材料的非线性,能够改善数值计算结果,与实测结果吻合地很好,比具有相同隐含层神经元个数的BP模型更接近实测结果。     

基于级连相关神经网络的人工冻土本构模型

为解决BP (back propagation) 神经网络收敛速度慢,网络结构需事先定义等缺点,采用了级连相关神经网络模型来建立人工冻土应力和应变之间的关系. 基于该模型推导了冻土的一致刚度矩阵形式,利用人工冻土三轴试验数据对神经网络模型进行训练,并用其替换有限元计算中的传统本构模型,将计算结果与性质及含水率相同的冻土的试验结果进行了对比,发现该神经网络本构模型很好地反应了材料的非线性,能够改善数值计算结果,与实测结果吻合地很好,比具有相同隐含层神经元个数的BP 模型更接近实测结果.     

方钢管混凝土短柱轴心受压承载力的神经网络模拟

由于方钢管混凝土的侧向约束机构复杂,对方钢管混凝土柱强度承载力的计算至今仍没有一种统一的方法。本文拟采用神经网络方法对轴心受压方钢管混凝土短柱的承载力进行模拟。以混凝土抗压强度、钢管的屈服强度、套箍指标、截面尺寸和宽厚比等五个参数为网络输入,以构件的极限承载力为网络输出,构建多层前馈神经网络来描述它们之间的非线性关系。利用55组试验数据对网络进行训练和测试,并将其预测值与三种承载力计算模型的预测值进行比较。对比结果表明本文建立的神经网络模型对55组试验数据给出了最好的模拟精度,可作为预测方钢管混凝土柱承载能力的一种新方法。     

Crack detection in a rotating shaft using artificial neural networks and PSD characterisation

The diagnosis of cracks in rotating shafts using non-destructive techniques provides a route for avoiding catastrophic failure of these common components. This study measured the dynamic response of a full-scale rotating shaft with three different crack depths. A novel non-destructive system is developed and described. The system uses vertical vibration of the system measured over time and characterises its behaviour using elements of the power spectral density (PSD) gained from a fast Fourier transform of the time-history. The PSDs were used as an input into an artificial neural network (ANN) to detect the presence of cracks using changes in the spectral content of the vibration of the system. A novel method for reducing the amount of data input into the ANN is described. The Peak Position Component Method (PPCM) reduces data transfer by using statistical characterisation of the position of the peaks in the PSD. The peak positions represent a small fraction of the information contained in the total frequency range. The number of the PSD peaks used as input to the neural net is a small fraction of the total frequency range. The ANN was a supervised feed-forward network with Levenberg-Marquardt back-propagation algorithm acting on the PPCM results. The frequency spectrum for the three different crack lengths examined show clear shifts in the peak positions of the PSD and the results clearly demonstrate the feasibility of using the new system to detect cracks in-service.     

Sensitivity analysis of freestream turbulence parameters on stagnation region heat transfer using a neural network

A neural network has been used to predict stagnation region heat transfer in the presence of freestream turbulence. The neural network was trained using data from an experimental study to investigate the influence of freestream turbulence on stagnation region heat transfer. The integral length scale, Reynolds number, all three components of velocity fluctuations and the vorticity field were used to characterize the freestream turbulence. The neural network is able to predict 50% of the test data within ±1%, while the maximum error of any data point is under 3%. A sensitivity analysis of the freestream turbulence parameters on stagnation region heat transfer was performed using the trained neural network. The integral length scale is found to have the least influence on the stagnation line heat transfer, while the normal and spanwise turbulence intensities have the highest influence.